Your Google Ads account, audited every day.

Search terms that took real spend over the last 30 days and converted nothing, in enabled Search, Standard Shopping and Performance Max campaigns alike. Each is rated by how likely its clicks were to bring nothing at the rate the campaign's other searches convert, so a few clicks with none is an early read, since it could still convert: it's listed under Early, and what it cost isn't counted as wasted spend until Goldbeater can tell. Goldbeater drafts an exact negative keyword for each, or pausing the campaign's own keyword where the search is one. While a few clicks could still explain it, or where the search converted in that campaign over the last year, or names your brand in a brand campaign, the change is offered rather than recommended.

A word that runs through many of a campaign's searches, or a country or region the campaign doesn't target, where none of the searches containing it converted, in the last 30 days or over the year. Words like free or jobs, and the words customers in your line of business rarely buy with, count on less evidence only where your own searches with them convert less than your others. Goldbeater reads your line of business from your keywords and ads, and you can change it in Settings. One search is often too small to judge, but the word they share is not, and one phrase-match negative stops all of them. Each is rated by how likely its searches' clicks were to bring nothing at the rate the campaign's other searches convert; while a few clicks could still explain it, it's an early read, with no figure yet, and the negative is offered rather than recommended. Its figure is spend some search-term findings also count, so a campaign's total counts it once.

• A word running through many searches, none of which converted
• Searches with a word like free or jobs, looking for something other than a purchase
• Searches naming a country or region the campaign doesn't target
• Searches with a word customers in your line of business rarely buy with

A search your campaign reaches only through a looser keyword's matching, or a word that runs through several of its searches, that converts, but at far worse cost than the campaign's other searches over the last 30 days: under half the conversions its spend would have bought at their rate, beyond what chance allows for everything tested, and a return under half theirs where your conversion value is real. A brand search is judged beside the campaign's other brand searches. A negative keyword would give those conversions up, so Goldbeater drafts none: it shows what the search costs a conversion against the campaign's own price, names the keyword that bought it, and says what brings the price down, the keyword's bid under Manual CPC, its match type, or the campaign's target.

• A search reached through a looser keyword, converting at far worse cost than the campaign's other searches
• A word running through several searches that convert at far worse cost than the campaign's others

A search one of your keywords keeps matching that people click far less often than the keyword's other searches: under half the rate, beyond what chance allows for every search tested, by Fisher's exact test on the keyword's own 30 days. A search nobody clicks costs nothing, so this isn't money lost; it's your account's own evidence that the people making that search aren't looking for what the ad offers, usually another product's name a broad keyword reached. Goldbeater never raises one that converted, names your brand or is one of the campaign's keywords, and leaves a search that has cost real money to the wasted-search check. It offers an exact-match negative keyword and never recommends it, since only you know whether the search is for what you sell; where another keyword's ad is clicked on it as usual, it names the ad group to add the negative to instead.

An exact-match keyword you paused whose search the campaign still buys over the last 30 days through its other keywords, as a close variant of another exact keyword or a broader match of a phrase or broad one: you switched the search off and the account still pays for it. Goldbeater leaves out a keyword whose words are still live in the campaign in any match type, since that was a choice, and a search that has cost real money with no conversion, which the wasted-search check raises with its negative. It offers an exact-match negative keyword where the search converted nothing in 30 days or a year, and never recommends it, since a pause can be for a season or a bid and Goldbeater can't see when or why it was made; where the search converts, it says the choice: enable the keyword again, or block the search and give those conversions up.

A Search or Standard Shopping campaign that bought searches with no negative keyword anywhere: none of its own, none on its ad groups, and no negative keyword list. It pays for every search its keywords or products are matched to, and nothing turns away the ones that don't fit. Goldbeater names the searches it raised there as wasted, each with the negative that stops it.

A negative keyword list left off a campaign it fits, where most of the campaigns it fits carry it, more than chance would give, or where even one carries it and this campaign bought searches the list would have kept out: that campaign keeps paying for the searches the list keeps the others off. Goldbeater applies the list only where none of its negatives would block the campaign's own keywords or a search that converted there over the last year, never where the list's name says it isn't for that campaign (a Non-FINRA list on a FINRA campaign), and never outside the kind of campaign its carriers are all named for (a NonBrand list on a brand campaign). It names the searches the list would have kept out, and recommends it only where their clicks bringing no conversion is unlikely by chance; until then it's an early read, and the list is offered.

A negative keyword list whose campaigns are all paused or ended, so it keeps nothing out. Goldbeater shows which live Search campaigns it would fit, where none of its negatives blocks their keywords or a search that converted there over the last year, and what it would have kept out of each; whether one of them should carry it is yours to say. A list on a campaign built and not launched yet is waiting for it, and isn't counted.

A negative keyword that most of the campaigns named like this one keep as their own, more than chance would give, left off it: a copy made without them, or a campaign that never got them, keeps paying for the searches the others turn away. Goldbeater adds them only where none would block the campaign's own keywords or a search that converted there over the last year, names the searches each would have kept out, and recommends them only where those searches' clicks bringing no conversion is unlikely by chance; until then it's an early read, and they're offered.

An enabled keyword whose own words a negative keeps out: the campaign's own negative, a negative keyword list attached to it, the account's list, or a negative on its own ad group. Either a keyword you think is working is kept off its searches, or the negative is keeping the campaign off traffic it was built for. For the campaign's own negative Goldbeater offers both fixes, removing it or pausing the keyword; a list reaches every campaign that carries it, so there, and on an ad group, it offers pausing the keyword. It recommends the pause only where another campaign takes those searches and the keyword no longer buys anything, since a negative doesn't block close variants.

The same negatives kept separately on several campaigns, or on every ad group of one. Each copy is edited on its own, and a new campaign or ad group starts without them. Goldbeater groups them by the campaigns that carry them, so each finding is one negative keyword list to make, or one campaign to hold them.

• The same negatives kept separately on several campaigns, where one list would do
• The same negatives on every ad group of a campaign, where the campaign would do

A search that has converted at least twice in the last year, for no more than its campaign pays for a conversion, and reaches the campaign only through a broader keyword's matching, so a change to that keyword can lose it. Google puts an exact keyword identical to a search ahead of the looser ones, so Goldbeater drafts the search as an exact keyword in the ad group that converted it, recommended where it converted twice in the last 90 days, so one whose season has passed is offered, not pressed. The figure is what other ad groups spent on it over the year, which the keyword moves.

A search Performance Max has converted at least twice in 30 days, where one of your Search campaigns serves the same places and has no keyword for it and didn't buy it. As an exact keyword there it's answered by the ad and page you wrote, ahead of Performance Max. Goldbeater drafts it in the ad group whose keyword sits inside the search, and offers it rather than recommending it where Performance Max pays less for a conversion than the Search campaign does.

Two of your enabled campaigns can show on the same searches. Google picks one each time by its own rules, not by which one you meant, so a search can land in the campaign with the worse page or target, and its results are split across both. Goldbeater raises a search one of them converted, and recommends keeping it in one only where that one converts it better by more than chance, and never by negating your brand in a brand campaign. The figure is the spend that would move, not money saved.

A search one of your Search campaigns bids on as a keyword, bought by Performance Max instead. It gets the ad and page Performance Max assembled rather than the ones you wrote for it, and its results are counted in Performance Max rather than the campaign built for it. Google puts an exact keyword identical to a search ahead of Performance Max, so that's the change Goldbeater drafts. The figure is the spend that would move, not money saved.

A search that converts in one ad group of a campaign and is bought by others too, each answering it with its own ads and page and its history split between them. An exact keyword in the ad group whose keyword it is, or that converts it best, puts it there first, so that's the change Goldbeater drafts. The figure is the other ad groups' spend on it, which would move, not money saved.

A search that converts in one ad group of a Standard Shopping campaign and is bought by others too, so its clicks and sales are split between ad groups with different products and bids. A Shopping campaign has no keywords to hold a search in one ad group, so the finding says which ad groups to add it to as a negative keyword. The figure is the other ad groups' spend on it, which would move, not money saved.

Searches naming your brand bought by a campaign built for other searches, or searches that don't name it bought by your brand campaign through broad match or AI Max. Each campaign's numbers then mix the two, and bidding set for one kind of click pays for the other. Your brand is the names Goldbeater reads from your account's name, its ads and what your brand campaigns buy, or the terms you set in Settings. Where you keep a negative keyword list of the brand that keeps them all out, Goldbeater applies it. The figure is spend that would move, not money saved.

• Searches for your brand bought by a generic campaign
• Generic searches bought by your brand campaign

A broad-match keyword whose searches Google matched only as broad convert at under half the rate of those it matched as phrase or exact, too far under to be chance. Goldbeater drafts the keyword as phrase match with the broad one paused, which keeps the searches that carry its meaning. The figure is the broad-only spend that converted nothing; its search-term findings count some of the same money, so a campaign's total counts it once.

An exact or phrase keyword whose close variants, the misspellings, plurals and rewordings Google matches it to, bought under half the conversions their spend would have at the rate of the searches matching the keyword as written, too far under to be chance. Google Ads has no setting for close variants, so the spend on the ones that never convert goes on until they're added as negatives.

A phrase or exact keyword where none of the searches Google reports it showed on has the keyword's words: every one is a close variant, matched on its meaning. That's how phrase and exact match are meant to work, so nothing is wrong yet. It's an early read, with no figure, that the keyword is reaching other searches than the ones you wrote it for, usually before any of them has cost enough to judge. Goldbeater lists the searches with how Google matched each and what they cost, from five searches on, and never where one of them converted. For a phrase keyword it offers exact match where Google's own labels say that should stop some of them, and never recommends it; otherwise the lever is negative keywords, which block the search you add and its misspellings, not its plural or a synonym.

A keyword that over the last 13 weeks brought under half the conversions its spend would have bought at the rate your campaign's other keywords converted at in the same weeks, too far under to be chance, and under half their return where your conversion value is real. Goldbeater drafts pausing one that converted nothing, and halving the bid of one that did under Manual CPC; under automated bidding, which sets its bid itself, it names the searches to check first and offers the pause. The estimate is what it spends above the campaign's own price a month, at what it spent in the last four weeks.

• A keyword spending for months with no conversion while the rest of its campaign converts
• A keyword paying far more for its conversions, or returning far less, than the rest of its campaign
• A keyword paying far more for its conversions, or returning far less, than the rest of its campaign under automated bidding

Two or more of a campaign's keywords sharing a word or phrase, none of them enough to judge alone, that together over the last 13 weeks spent real money and converted nothing where the campaign's other keywords converted, too far under to be chance. Goldbeater offers pausing them until you've read the searches they buy, and under Manual CPC names their bids to lower first.

A keyword that in the last four weeks brought under half the conversions its spend would have at its own rate the four weeks before, allowing for how the rest of its campaign moved, too far under to be chance. The rest of the campaign held up, so the cause is the keyword's own: the finding says whether its clicks got dearer or stopped converting, and nothing is changed until you know which.

A keyword whose cost per click in the last four weeks rose against its own weeks before and against the rest of its campaign's keywords in the same weeks, each by more than its weeks' own ups and downs explain, where what it paid over is real money and its conversions didn't come cheaper. The finding says where to look: under Manual CPC its bid and the rivals bidding against it, under automated bidding the auction and the parts of Quality Score Google rates below average.

Where your conversion value is real, a keyword with five or more conversions in each period whose value per conversion in the last four weeks fell against its own weeks before and against the rest of its campaign's keywords in the same weeks, each by more than its weeks' own ups and downs explain. Its return fell because each conversion is worth less, not because fewer came: the finding says to read which conversion actions and purchases are behind it, and nothing is changed until you know why.

A keyword with five or more conversions in 30 days, converting at least as well as its campaign and not under its campaign's target, in a campaign converting at least as well as your other campaigns typically do, losing a fifth or more of its searches to ad rank. Ad rank is the bid and the quality: under Manual CPC Goldbeater drafts a higher bid; otherwise the finding names the part of Quality Score Google rates below average, the campaign's target, or its budget: always on Maximize conversions or Maximize conversion value with no target, which bid to spend it, and on another strategy where the campaign loses searches to budget.

• A Manual CPC keyword converting well, losing searches to ad rank
• A keyword converting well, losing searches to ad rank on a part of Quality Score Google rates below average
• A keyword beating its campaign's target, losing searches to ad rank
• A keyword converting well in a campaign short of budget, losing searches to ad rank
• A keyword converting well, losing searches to ad rank, with nothing in its campaign to change for it

A Manual CPC keyword whose maximum CPC is under Google's estimate of the bid that reaches the first page of results, in a campaign converting at least as well as your typical one that already pays at least that much for a click on average, where the keyword itself isn't losing money. Google says an ad under that estimate is less likely to show on the first page. Goldbeater drafts the bid at the estimate.

A Manual CPC keyword whose maximum CPC reaches the first page but is under Google's estimate of the bid that shows an ad among the top ads, which generally sit above the results, in a campaign with five or more conversions in 30 days, converting at least as well as your typical one and already paying that much for a click on average, where the keyword itself isn't losing money. Goldbeater drafts the bid at the estimate, recommended where the keyword has converted.

Quality Score falling on keywords that still carry spend. A lower score usually means paying more for the same position, and Google's three parts of it say which one moved: expected click-through rate, ad relevance or the landing page.

Keywords Google scores 4 or lower out of 10 for Quality Score that spent in the last 30 days, gathered by ad group, each with the three parts Google rates it on: expected clickthrough rate, ad relevance and landing page experience. Google says higher quality ads can lead to lower prices and better ad positions. The finding names each part rated below average, and whether the fix is the ads or the landing page. A keyword naming a competitor scores low by nature, since the searcher is looking for the competitor, so it's judged instead by what its clicks cost per conversion against your other campaigns.

• Keywords spending at a low Quality Score
• Competitor keywords whose clicks cost far more a conversion than your other campaigns pay
• Competitor keywords scoring low whose clicks pay, come near, or are too new to judge

A keyword Google rates 1 to 3 for Quality Score, with spend, in an ad group whose other keywords mostly rate 5 or more, and which holds no more low-scoring keywords than well-rated ones. The ad group's ads and page suit the rest and not this one, and Google says higher quality ads can lead to lower prices. It wants an ad group of its own, or a pause.

An ad group where most of the spend is on keywords Google rates below average for expected clickthrough rate or ad relevance, or a landing page where most of the spend reaching it comes from keywords rated below average for landing page experience. Google says higher ad quality generally brings better ad positions at a lower cost; the finding says which part, and whether the fix is the ads or the page.

• Ads Google expects to be clicked less often than others for your keywords
• Ads Google rates as a poor match for the keywords they serve
• A landing page Google rates below average for the keywords sending people to it

An ad group whose ads Google rates relevant to some of its keywords and below average for others, each side spending in the last 30 days, with most of the spend on the keywords the ads suit and a tenth or more on the rest. Google says higher ad quality generally brings better ad positions at a lower cost, and its advice for such an ad group is to split it: the keywords the ads don't suit get an ad group and an ad of their own, and the rest keep the ads that suit them.

A Search ad group holding far more keywords than your other ad groups do, past the usual line for an outlier, where more than half of them took no click in the last 30 days and that many idle is beyond chance at the rate the keywords in your ad groups of usual size go without one. The keywords that buy nothing are dead weight around the ones that carry the ad group, and Google asks for non-serving keywords to be cleaned up, since the excess makes an account harder to manage and more error prone. Goldbeater lists them with Google's own status on each, offers pausing them and recommends nothing: a keyword added last week, or sold in another season, is yours to know.

A keyword Google disapproved under one of its advertising policies shows ads on no search, and one it limited shows only where and when the policy allows. Goldbeater names the policy Google gave, and whether the keyword was converting before it stopped.

• A keyword Google's policy review disapproved, so it shows on no search
• A keyword Google's policy review limited to where and when the policy allows

Only one copy can show on a search, and Google picks which, so the ad and page that serve may not be the ones you meant. The history splits between the copies, and every edit has to be made in each. Broad keywords whose words differ only in order, spacing or a plural are copies too, since broad match buys the same searches for both. Across two campaigns serving the same places and languages, where one copy takes every click, too many to be chance, the other is only a second set of settings to keep in step.

• The same keyword, in the same match type, live in two ad groups of one campaign
• Broad keywords whose words differ only in order, spacing or a plural, buying the same searches
• The same keyword in two campaigns serving the same places, where one copy takes every click

Keywords and negatives in force whose words say one thing and whose setting another: a keyword written +running +shoes serves as phrase match since Google retired broad match modifier in 2021, a negative written with plus signs blocks what the words without them block, and quotation marks or brackets in a keyword's words set no match type. Whoever reads the account next judges it by what it says. Goldbeater drafts the rewrite for a keyword or a campaign's own negative.

• A keyword still written in broad match modifier, which Google serves as phrase match
• A negative keyword written with plus signs, which Google ignores
• A keyword or negative with quotation marks or brackets in its words, set to another match type

Ads Google disapproved, or that a policy keeps out of every place the campaign targets, and headlines or descriptions it disapproved inside a responsive ad. A disapproved ad serves nothing, and a disapproved headline leaves its ad one short until it's replaced. An ad approved with limits serves less than you think: it says where or to whom, and whether a review, a certification or nothing is the answer.

• An ad Google disapproved, so it serves nothing
• An ad a policy keeps out of the places its campaign targets
• A headline or description Google disapproved inside an ad that still serves
• An ad Google approved with limits: kept out of some countries it targets, from your own audience lists, or serving without a rejected part

An enabled campaign with no ad group, or whose ad groups are all paused, and a Performance Max campaign with no enabled asset group, show no ads: Google marks each Not eligible. It reads as live in your campaign list while it does nothing, so a campaign you meant to run isn't running. Goldbeater offers pausing it where it isn't meant to run.

An enabled ad group whose ads are all paused or disapproved shows nothing, in Search, Shopping or Display, and a Search ad group with no keywords matches no search at all; a Search campaign none of whose ad groups has a keyword is marked Not eligible and serves nothing. An ad group running a single ad gives Google nothing to choose between, where it asks for at least two responsive search ads in each, and under Smart Bidding Google asks for keywords grouped by theme rather than one to an ad group, in however many match types. Without Smart Bidding a single keyword is a way to control what the ad group buys, unless it's broad match, where Google says Smart Bidding is critical: one bid then buys every search Google relates to it.

• An ad group with no ad Google shows, so nothing in it shows
• An ad group with no keywords, so it matches no search
• An ad group running a single ad, with nothing to compare it against
• An ad group holding a single keyword under Smart Bidding, where Google asks for themed ad groups
• An ad group holding a single broad match keyword without Smart Bidding, one bid for every search Google relates to it

A Search ad group showing more than three enabled ads, or a Display ad group more than four, Google's own numbers, whose conversions over the last 8 weeks couldn't tell any of its ads from luck even converting half as often as the others, and could compare fewer ads than it runs. Every ad takes a share of the same impressions, so none reaches a verdict and the ad tests never call one. Goldbeater says how many its conversions can compare, which to pause first, and never fewer than Google asks for.

A responsive search ad whose headline or description names a sale, deadline or event that is over, such as "Sep 4-8 only" three weeks later. Searchers read an offer that has ended, and each click costs the same. Where another ad in the ad group can carry it, pausing the ad is drafted.

An ad whose headline, description or path says a word you've listed in Settings as one your ads must never say: a competitor's name, a claim you can't make, a word you've decided against. Goldbeater reads every enabled ad for each word, whole and in any capitals, and the ads of a campaign waiting to launch, when a word is cheapest to take out. With no words listed it raises nothing.

A line you wrote that holds a common misspelling, such as recieve for receive: an ad's headline or description, a sitelink, callout, structured snippet, promotion or price asset, or a Performance Max asset group's text. Google shows the line as you wrote it. Goldbeater reads each line against a published list of common English misspellings that each have one right spelling, so the same line reads the same way every time, and only where the ad group's or asset group's own text is written in English. It never raises a word you use on purpose: one in a keyword, a product title, a brand term, a business name, a campaign or ad group name or your site's address, anything Google fills in, or a word in capitals. The finding names the line, the word and its right spelling.

• An ad's headline or description that misspells a word
• A sitelink, callout, snippet, promotion or price asset that misspells a word
• A Performance Max headline or description that misspells a word

An ad that brought under half the conversions per impression of the other ads in its ad group over the weeks they ran together, and no more per click, by more than chance explains across every ad in the account. The impressions it takes go to the weaker message. Goldbeater drafts pausing it, and never an ad group's last ad.

A headline or description with which its ad drew under half the clicks per impression it drew without it, over eight weeks, by more than the account's lines differ from their ads on their own. Every impression it takes is one the ad's better lines could have had.

An ad whose click-through rate over the last four weeks fell by a quarter or more from the four before, judged week by week beside the other ads in its ad group and the rest of its campaign, so a fall their searches share isn't counted. The same impressions bring fewer customers to your site.

A Search campaign converting enough that a four-week, half-and-half experiment could tell a difference of a fifth in conversions from luck, in an account that hasn't run an experiment or ad variation in a year. Changes to its bidding, keywords or pages are then judged by before and after, which seasons and auctions blur. Nothing is lost by not testing, so it's offered, and low.

Ad groups where Google rates no ad better than Poor or Average for ad strength: headlines that say the same thing, too few of the ad group's keywords in them, or too many pinned, so Google has fewer good combinations to try. A Demand Gen ad is judged by the pictures in each shape, the headlines and the description Google's own table asks of a Good one. It reads ads before they launch too, when a rewrite costs the least. Google doesn't use the rating in the auction; it's feedback on the ad's variety.

• Responsive search and display ads Google rates Poor or Average
• Demand Gen ads short of the Good ad strength Google asks for

An ad group whose ads say everything in a few words a line: no headline, or no description, long enough to say an offer in full, judged against how long the account's other ad groups' longest lines run. Google asks for headlines of a variety of lengths, so the short ones stay; what's missing is one or two that use the room. Lengths are counted as Google counts them.

• An ad group none of whose headlines uses the room a headline has
• An ad group none of whose descriptions uses the room a description has

Two responsive ads in one ad group sending people to the same page with mostly the same headlines or descriptions, or one ad holding a line twice. Each ad assembles its own combinations, so two alike ads leave Google's rotation nothing to learn between them, and a line twice leaves the ad one fewer to test. Ad groups of one campaign with much the same ads, more alike than the account's own campaigns are, sending people to different pages: the split buys nothing in relevance. Not two ad groups for competitors, whose ads can't carry the competitor's name that would tell them apart. An ad group whose headlines are its own but whose every description also runs, under other headlines, for other pages, where most of the account's ad groups have a description of their own: whichever Google shows was written for no page in particular. And text pasted in the wrong encoding, which searchers see as written.

• Two responsive ads in one ad group sending the same page nearly the same headlines or descriptions
• A responsive ad holding the same headline or description twice
• Ads showing garbled characters, such as — where a dash was meant
• Ad groups of one campaign running much the same ads to different pages
• An ad group with no description of its own, each also running for other pages

A headline or description pinned to a position with nothing pinned beside it shows there on every impression, so Google has nothing to test in that place. Where one has to show there, Google asks for two or three pinned to the position.

Expanded text ads still enabled, which Google stopped anyone editing on June 30, 2022. They serve, but their words stay as written, and an ad group with no responsive search ad beside them has no text anyone can change. Where ad groups sending people to different pages all show the same first headline, second headline or description in every expanded text ad they run, none can be edited to say what its own ad group is about, so those are the ones to replace first.

• An expanded text ad still running in an ad group
• Expanded text ads in ad groups for different pages all showing the same headline or description

A Search campaign whose Dynamic Search Ads all run through a single ad group, with its clicks landing on several pages of your site in the last 30 days: one ad and one set of targets for every page it reaches. Google upgrades Dynamic Search Ads campaigns to AI Max from February 2027 and moves their dynamic ad groups to standard ad groups with their settings carried over, so Goldbeater says to upgrade by hand first and split what the ad group reaches by theme, and never to add dynamic ad groups the upgrade would replace.

An ad group, or a whole campaign, set to rotate its ads indefinitely under a bid strategy that follows the setting. Every ad gets its turn whatever its results, which suits a test you run by hand; Google calls it not recommended for most advertisers, and its optimized rotation shows the ads expected to do better.

Text customization or final URL expansion on in a Performance Max campaign, or in a Search campaign with AI Max, where both are on unless someone turns them off, and Performance Max's image and video enhancement, on the same way. Google then writes headlines and descriptions from your pages and ads, may send a click to another page of your site than the ad names, and crops, animates and shortens the images and videos you gave it. The one campaign of its kind left with one on, where the rest have it off, is flagged as an oversight; otherwise it is yours to judge.

• Text customization on, so Google writes your ad text
• Final URL expansion on, so Google chooses your landing page
• Image enhancement or landing page images on, so Google edits your images and adds its own
• Video enhancement or sourced videos on, so Google makes new versions of your videos and finds more

A page your ads, keywords, asset groups, sitelinks, promotions or price assets send clicks to, or one Google chose for a search itself, that answered with an error when Goldbeater opened it, and again when it asked a second time: page not found, a server error, a domain that doesn't resolve, or a certificate browsers refuse. So does a page that loads but whose title or heading says it wasn't found, or that sends a visitor to the home page instead, and an app asset whose store has no such app. Every click to it is billed, and the visitor meets the error instead of the page.

• An ad, keyword or asset group sending clicks to a page that doesn't load
• A sitelink, promotion or price asset sending clicks to a page that doesn't load
• A page that loads but says it isn't there, or sends visitors to the home page
• An app asset sending taps to an app the store doesn't have

A final URL of an ad, keyword, asset group, sitelink, promotion or price asset that redirects to another page when Goldbeater opens it: to another site, which Google's destination requirements disapprove, or within your site to a page that moved for good. Every click makes a round trip to the old address first, and the ad names a page the site no longer serves.

Ads, keywords, asset groups, sitelinks, promotions or price assets whose final URL or mobile final URL starts http://. Chrome tries the https address first, and from Chrome 154 in October 2026 it asks before opening a public site that has none, so where your site has no https, a new visitor's paid click meets a warning before your page.

Ads Google's own review found pointing at a page it couldn't load, couldn't reach, or that isn't where the ad says it goes. While the ad still serves, each click is billed whether or not the page loads.

• An ad Google disapproved or limited because it couldn't load the landing page
• An ad Google disapproved or limited because its landing page isn't where the ad says it goes

A page your ads send clicks to whose HTML carries no Google tag, on a site whose other pages carry one. Google asks for its tag on every page a click can land on, since that's where it records the click's details; a conversion after landing on a page without it can go unrecorded.

Of the ten pages your ads send the most spend to, those where a quarter of real visitors on phones waited longer than Google's lines for good: 2.5 seconds for the main content to show, or 200 milliseconds for a tap to answer, from Google's PageSpeed Insights. A visitor who waits often leaves before the page shows what the ad offered, and the click is billed either way.

A page whose clicks bring under half the conversions the same clicks bring on the other pages of the campaigns sending them there, and whose conversions cost more than theirs, too far under to be chance. Each campaign is its own measure, so a page is never judged against a campaign with another goal. The usual causes are a conversion tag missing on the page, a page that doesn't say what the ad did, or searches it wasn't built for; the estimate is the conversions its spend would have bought at what the other pages paid for one.

An ad group whose keywords Google mostly rates below average for landing page experience, on a page that other ad groups also send people to and whose keywords Google mostly rates average or above there: the page answers their searches and not this ad group's. Goldbeater takes an ad group's page to be the one most of its spend reached, and raises it only where so many of its keywords rated below average is more than chance explains across every ad group tested. Mending the page for this ad group would cost the ones it suits, so the fix is a page of its own, and where the same ad group in another campaign already lands on a page Google rates well, the finding names it.

Campaigns running without sitelinks, or with fewer than the six Google recommends; with only one, Google shows none. They cost nothing to add, make the ad larger on the page, and usually lift its click-through rate. It reads campaigns before they launch too.

Callouts, at least four, and structured snippets, under more than one header with four values each, are free lines of text on your ad, and a call asset, images, a logo, prices, an offer, your app or a message button that another of your campaigns already shows fits the rest too. A campaign without them shows a smaller ad than it could, and usually earns fewer clicks for it. It reads campaigns before they launch too.

• A campaign showing no callouts, or fewer than Google recommends
• A Search campaign showing no structured snippet
• Structured snippets under one header, where Google asks for more than one
• A structured snippet header with fewer than the 4 values Google asks for
• A Search campaign without the call asset your other campaigns show
• A Search campaign without the prices your other campaigns show
• A Search campaign without the promotion your other campaigns show
• A Search campaign without the lead form your other campaigns show
• A Search campaign without the images your other campaigns show
• A Search campaign without the business logo your other campaigns show
• A Search campaign without the app your other campaigns link to
• A campaign without the message button your other campaigns show

Pages your Search ads send clicks to that link a phone number for visitors to call, in an account with no call asset at all. A call asset puts that number on the ad, so a visitor on a phone can call without opening the page, and Google can count the calls as conversions.

An account none of whose campaigns shows a call asset, a lead form or images, where Google recommends one on its Search campaigns from signals only it has. Goldbeater says what the campaigns recorded that the asset would serve (calls counted, leads taken) and what they spent without it, and leaves the choice yours: Google's image recommendation hands it the choice of pictures, so Goldbeater asks for your own instead.

• No campaign shows a call asset, where Google recommends one
• No campaign shows a lead form, where Google recommends one
• No Search campaign shows images, where Google recommends letting it pick some

A campaign targeting a radius around a place, the targeting of a business people come to, in an account with no Business Profile linked. Its ads can't show your address, a map to you or how far away you are, which location assets add.

Sitelinks, callouts, snippets, images, a business name or a logo that Google disapproved, or that a policy keeps out of every place their campaigns target, wherever they're linked: the account, a campaign or an ad group. Google won't show it there until it's fixed, so the ads carry less than you gave them. One approved with limits shows in some of the places its campaigns target and not others, and the finding says which. It reads campaigns before they launch too.

• A sitelink, callout or other asset Google disapproved
• A sitelink, callout or other asset a policy keeps out of the places its campaigns target
• A sitelink, callout or other asset Google approved with limits, in a country its campaigns target

A sitelink, callout, structured snippet, promotion or business name that says a word you've listed in Settings as one your ads must never say. It shows under your ads wherever it's linked, the account, a campaign or an ad group, and one Google wrote for you counts as yours. Goldbeater reads each asset's every line for each word, whole and in any capitals, before a campaign launches too. With no words listed it raises nothing.

Groups Google rates Poor or Average for ad strength: too few, or too similar, assets to assemble a good ad for every placement. And groups with no video or portrait image of their own, where Google shows a video it made from your images, or has no image of yours for the placements that take a portrait one.

• An asset group Google rates Poor or Average for ad strength
• An asset group with no video of its own, so Google makes one from its images
• An asset group with no portrait image of its own

An image, video, headline or other asset Google disapproved, or a video YouTube no longer plays, still in a Performance Max asset group. It never shows and still holds one of the group's places; where it leaves the group under Google's minimum of its kind, the group can't assemble its ads.

A headline, long headline, description or business name in a Performance Max asset group that says a word you've listed in Settings as one your ads must never say. Performance Max builds its ads from every line the group holds, so one line reaches all of them, and a line Google wrote counts as the group's own. With no words listed it raises nothing.

An asset group with no audience signal, or one whose signal holds only interests, custom segments or demographics and none of your customer lists or site visitors. Google names your own data as the signal to give Performance Max first; Goldbeater names the lists in your account to add.

The videos, sites and apps a Performance Max campaign showed on that a published list, or their own name, marks as made for children, as games or as job sites, where together they showed more than a stray placement would. Google reports only impressions for them, no cost or conversions, and says the report is for brand safety rather than performance, so the finding says where your ads appeared and how often, never that money was wasted. Performance Max takes placement exclusions only for the whole account.

• Performance Max ads showing on videos, sites and apps for children
• Performance Max ads showing on games sites, in game apps and on gaming videos
• Performance Max ads showing on job and employment sites

Google has suspended the Merchant Center account your products come from, which takes them off Shopping ads and free listings until it's reinstated, or has warned that it may, with a deadline to fix the issue. The finding says what its products spent in your ads, and where in Merchant Center the issue is and a review is asked for.

An enabled Shopping or Performance Max campaign that has no product it can show, so it runs no Shopping ads at all: Merchant Center has none for it, or its feed label, inventory filter or listing groups leave every one out. The finding says which, from Google's own reason.

Products that can show in Shopping ads and that none of your campaigns advertises, as Google counts them, in a feed you advertise from. The usual cause is an asset group or ad group that lists its products one item ID at a time, so a product added to your feed since falls into an excluded Everything else. The finding says what the products are and which group leaves them out.

Listing groups in a Performance Max asset group, or product groups in a Standard Shopping ad group, that name an item ID, brand or product type no product in your feed can show today or has shown in the past year. An empty group costs nothing itself; it means the groups stopped following the feed, and the products that replaced those are usually left out. The finding lists the groups to check and remove.

Products that were showing in your ads and now can't, because Merchant Center found a problem such as a product page Google can't reach, or that show for fewer searches because their data is missing an age group, a color or an identifier. Until the product data is fixed, your ads don't offer them where the problem applies. The finding says what the products kept out of your ads spent, apart from those still showing.

• Product pages Google couldn't load, keeping the products out of your ads
• Products held back for their data: a detail missing, wrong or against a policy

A feed where the products that can't show in ads today, disapproved or marked out of stock, took more of the feed's clicks over the last 30 days than the products still showing, too many to be the usual sell-through: a feed update that failed, a store that marked everything out of stock, or a policy sweep. The finding lists the products, starting with the stock data, and never says to remove them.

A product that spent several conversions' worth at its campaign's price and sold nothing, or sold under half what the campaign's other products sell for the same spend, too far under to be chance, and under half their return where your conversion value is real. The finding says where to exclude it or lower its bid; the estimate is what its sales cost above the campaign's own price.

A product group of several products in a Standard Shopping campaign that spent several conversions' worth and sold nothing, or sold under half what the campaign's other groups sell for the same spend, too far under to be chance. The finding says to lower its bid under Manual CPC, or to split it to find the products that don't sell; the estimate is what its sales cost above the campaign's own price.

A product group of several products in a Standard Shopping campaign that brings over twice the conversions the campaign's other groups buy for the same spend, too far over to be chance, and over twice their return where your conversion value is real. Under Manual CPC, where it still misses searches it could show for, the finding names a bid a fifth higher; otherwise it says when a campaign of its own would let it have more.

A product group in a Standard Shopping campaign under Manual CPC whose maximum CPC is above the most a click on it has shown it can return, its conversion value over its clicks taken at the best chance allows. The auction usually charges less than the bid, so nothing is lost yet; a competitor or a dearer hour can charge up to the bid with no decision of yours. The finding names a ceiling at what a click returns.

A product group in a Standard Shopping campaign under Manual CPC whose products' prices are so far apart that its one bid overpays for a click on the cheap ones while the dear ones could earn more: at the campaign's conversion rate, a click on some returns less than the bid and on others more. Where Google sets bids itself it bids per product, so a large group costs nothing there. The finding says how to split it.

Two Shopping or Performance Max campaigns that both paid for the same products in the last 30 days. Which one serves each search is decided auction by auction, so each has part of the products' clicks and sales, and a bid or target in one settles only part of what they cost. The finding lists the products and what each campaign paid; Standard Shopping campaigns set to different priorities are tiered on purpose and left out.

Where your products carry more than one brand, titles that don't name the product's brand in their first 70 characters or so, which is what shoppers usually see, and titles that name a different brand from the one the product carries, so one of the two is wrong. Where none of a brand's titles name it though your other titles name theirs, the brand itself is likelier wrong. The finding lists the products and what they spent in your ads.

• Product titles that don't name the brand where shoppers read them
• Products whose brand none of their titles name, likely set wrong
• Product titles naming a different brand from the product's own

Products that can show in ads but carry no custom label, product type or brand that your listing groups divide by, so each falls into an Everything else group: excluded there in Performance Max, so that asset group doesn't advertise it, or given the catch-all bid in Standard Shopping. The finding lists the products and the detail to add in your product data.

A target cost per conversion below every week's in the last 13, or a target return above every week's, while the campaign leaves budget unspent or loses searches on ad rank. Google bids low to chase a target out of reach: the campaign buys less and still misses it. A target changed in the last four weeks waits, since the weeks before it were bid to another.

A campaign with its own target cost per conversion or return that beat it in every week of the last 13 with five or more conversions, each by more than luck explains, while losing under a fifth of its searches to budget and under a fifth to ad rank. Google aims to keep a campaign's cost per conversion, or its return, at its target, so one this far ahead isn't steered by it and leaves Google room to bid more. A tighter target trades some conversions for a lower cost or a higher return, which is yours to weigh, so it's offered and never recommended.

A campaign bidding to a target CPA or target ROAS on fewer than 15 conversions in 30 days, or one Google reports limited by data. Google recommends judging a bid strategy over at least 30 conversions, 50 for Target ROAS, so a target set on fewer is set, and judged, on too little to know whether it's right.

A campaign on Maximize conversions or Maximize conversion value with no target, converting at least 30 times in 30 days, or 50 for conversion value: the counts Google recommends judging a bid strategy over, and more than the 15 it asks for before a Target ROAS on Search and Shopping. Without a target, Google aims to spend the whole budget, so the budget decides what each conversion costs, and a bigger budget buys dearer ones. One whose bid strategy or target changed in the last four weeks waits, since part of its month was bought under the earlier one.

A campaign's bid strategy or its target changed on two or more days in the last two weeks. Google says a change inside a conversion cycle makes the bidding harder to assess, and that a target changed often mid-cycle may take longer to reach. A changed strategy also goes back into learning, and results can swing while it learns. So one changed this often never settles long enough to judge.

A bid strategy Google reports held back by its maximum CPC bid limit over the last 30 days: it would have bid higher. Google advises against a limit on Target CPA and Target ROAS, and its best practices for value-based bidding say to avoid CPC limits, which covers Maximize conversion value; on Maximize clicks and Target impression share, a higher limit buys more clicks or impression share at a higher price, which is yours to weigh.

A Search, Shopping or Display campaign bidding to a target CPA or ROAS whose ad groups returned differently over the last 13 weeks, too far apart to be chance, its paused ad groups counted in the campaign's return. Google meets the target on average, so an ad group returning less keeps buying while the others carry it. Goldbeater drafts a target for each ad group whose own conversions differ from the rest beyond chance, from its own return weighed toward the campaign's by its conversions, at most 30% a step: higher where it returns less than the step asks, lower where it returns more and still loses searches, or a Display ad group's impressions, to ad rank, on a campaign its target holds back.

• An ad group returning far less than the rest of its campaign, under the campaign's one target
• An ad group returning more than the rest of its campaign and losing searches or impressions to ad rank, under the campaign's one target

An ad group in a Search, Shopping or Display campaign without a target of its own that over the last 13 weeks brought under half the conversions its spend would have at the rate of the campaign's other ad groups in the same weeks, too far under to be chance, and under half their return where your conversion value is real. Keywords raised on their own are left out of its figures, and one still among your account's better buys isn't raised. The finding says what to do by how the campaign bids: under Manual CPC, lower its bids; under automated bidding, read what it buys, then give it a campaign with a target of its own or pause it.

• An ad group spending for months with no conversion while the rest of its campaign converts
• An ad group paying far more for its conversions, or returning far less, than the rest of its campaign, with no target of its own

A campaign converting better than your typical one, by more than chance explains, losing a fifth or more of its searches to ad rank, and more to rank than to budget, where you have a lever: a target CPA or ROAS it isn't missing, set on enough conversions, or manual bids. A bigger budget won't win those back; a looser target or higher bids can, and where Google has simulated its target, the finding gives the estimate. Campaigns bidding to spend their budget, and ones held by a bid limit, are left to the checks that judge those. A campaign Google says is still learning after a change waits, and one whose budget, target or bid strategy changed is judged on the days since.

• A campaign on a target ROAS converting well, losing searches to ad rank
• A campaign on a target CPA converting well, losing searches to ad rank
• A campaign on manual bids converting well, losing searches to ad rank

A campaign whose name says Brand, week by week against its own recent weeks: shown at the very top of fewer of its searches than in any of the eight weeks before, or paying more a click than in any of them by more than their spread explains, without showing higher for it. Where the searches it lost went to its budget, the finding says so; otherwise, on a search for your own brand, both usually mean another advertiser bidding on it, and the finding names any change to its budget or bidding from the change history.

• A brand campaign showing first on fewer of its searches
• A brand campaign losing its first place on its own searches to its budget
• A brand campaign's clicks costing more than they did

Auto-apply is on for recommendations that change a bid strategy or its target, a budget, keywords and targeting, or the words in ads. Google applies each when it finds it relevant, and lists what it applied on the History tab of the Auto-apply page and in Change history, where you can review it. The finding counts every type switched on and lists what Google applied to campaigns, budgets and keywords in the last four weeks.

A campaign that has spent at least three conversions' worth in 30 days, at its own target or at what your typical campaign pays for one, and made no sale or lead, counted or not. The figure is its whole spend. Its search-term findings count part of the same money, so a campaign's total counts it once. A campaign that has spent on fewer than 14 days is an early read, with no figure yet: a conversion can follow its click by days, and automated bidding is still learning. The finding shows its last 13 weeks, and what its clicks recorded instead, with an invalid-traffic tool's flags apart from visits.

A campaign at or above its own target CPA or ROAS while losing a fifth or more of its searches to its budget, judged on the days since its budget or bidding last changed. Where it beats the target by more than chance explains on a budget and target that have run two weeks, Goldbeater drafts one a fifth higher, which you can take up to double; at about its target, or where an automated rule changes the budget, it offers the raise without recommending it. The finding lists the budget's changes, and where Google has simulated the budget drafted, its estimate.

• A campaign beating its target, held back by its budget
• A campaign at about its target, held back by its budget

A campaign that never came near its daily budget in 30 days, beside one that meets its own target and loses searches to its budget. Part of the first budget can move to the second without changing any day the first spent, or the account's total.

A budget shared by several campaigns, where one takes most of the spend and converts significantly worse than another on it that loses searches to budget. The campaign that converts better runs short of a budget it shares.

A campaign that spends each day's budget on most days and loses a fifth or more of its searches to it, judged on the days since its budget or bidding last changed, while it misses its target by more than chance explains or has none. More budget buys more of the same clicks; whether they're worth it at the return the campaign gets now is your call. With no target and a fair price, Goldbeater offers a raise a fifth higher, never recommended, and where Google has simulated that budget, its estimate. A campaign on Maximize conversions or Maximize conversion value with no target spends its budget by design, so there Goldbeater reads Google's budget simulator and not the share: it raises the campaign where the simulator prices what a raise adds within three times what your typical campaign pays, and says what Google's own forecast prices its recommended budget at where Google recommends one.

• A campaign limited by its budget while it misses its own target
• A campaign limited by its budget, paying far more a conversion than your others
• A campaign with no target, limited by its budget
• A campaign limited by its budget that converts nothing
• A Maximize campaign with no target, priced for more budget by Google's simulator

A campaign that loses a fifth or more of its impressions to its budget, more than to rank, on a daily budget that covers fewer than ten clicks at what a click costs it: its own average, or with no click yet, what its ad groups bid where its ads showed. Google spreads a small budget's ads through the day, so the campaign shows on only some of its searches and a day can end with no click; Google's guideline is a budget of at least ten times what a click costs. The other budget checks wait for 14 days of spend, which a budget this small can be slow to reach, so Goldbeater says this from the first day Google reports, as an early read until the campaign has shown for a week. More budget and cheaper clicks are both yours to choose, so nothing is drafted: the finding shows each day's spend, each ad group's bid, and the budget Google recommends where it has one.

A campaign Google recommends a bigger budget for, saying the return improves. Its own forecast says what the raise adds a week, so Goldbeater prices each added conversion, or the return on the added spend, against the campaign's own target or what your typical campaign pays, and says what its budget checks find beside it. Within that price, raise in steps; past it, Goldbeater doesn't advise the raise. Where its budget check already prices that campaign for more budget, that finding carries Google's recommendation, so one budget is one finding.

• A bigger budget Google recommends, whose added conversions cost no more than you pay
• A bigger budget Google recommends, whose added conversions cost more than you pay

A campaign Google forecasts a rise in searches for, from a date it names, and recommends a target and a bigger budget to capture it. Only Google can forecast the rise, so Goldbeater says it as Google's, beside what the campaign paid a conversion or returned over its last 30 days: a target looser than that lets each conversion cost more as searches rise, so it advises a target near what the campaign achieves, and the budget raised for the rise's days only.

The month's spend so far, projected to its end at the last seven days' pace, heading past your monthly spend target or well short of it by more than the account's day-to-day swings explain. Until you set a target in the account's Settings, the month is judged against what the account spent the month before, and raised only when it's heading further from that than its months usually differ. Past a target is money you said you wouldn't spend; short of it is volume the month won't buy.

• A month's spend heading past your target, or past last month's
• A month's spend falling short of your target, or of last month's

From the last seven days of Google's change history: one of the ten campaigns or keywords that spend the most paused or removed, or the daily budget of one of the three biggest campaigns changed, and still so. A change to what carries the account moves its results, whoever made it. The finding says how it was made, what a paused campaign was bringing, and what else changed in the same minutes; a budget change's value is what it moves a month.

• One of your top campaigns paused or removed
• A top campaign's daily budget changed
• One of your top keywords paused or removed

A campaign that showed ads on every one of the two weeks before and none yesterday: its ads disapproved, its end date passed, everything in it paused, or, where every campaign stopped, a payment declined. Each day it stays dark is a day of the sales it was bringing.

• One campaign that suddenly stopped showing ads
• Every campaign in the account stopped at once

For an account paying by monthly invoicing, the account budget's spending limit used up within a week at the last seven days' spend, or its end date within a week with no budget approved to follow it. Google stops every campaign's ads when either happens, so each day after is a day of the sales the whole account was bringing.

• An account budget's spending limit about to run out
• An account budget ending with no budget after it

A day, or a run of days to yesterday, spending at least twice the campaign's highest day of the four weeks before, on more clicks than chance explains: a budget or bid raised by mistake, a target loosened, broader match. The money goes at the new rate every day until someone sets it back.

The whole account, or a campaign carrying a twentieth of its spend, showing under a quarter of its usual impressions or clicks for the day of the week and spending less, or over four times as many, where chance doesn't explain it. A fall is sales missed each day it lasts; a jump is traffic from somewhere new.

• Impressions or clicks falling sharply, short of stopping
• Impressions or clicks jumping far past usual

A conversion action your campaigns count that recorded on every one of the two weeks before and nothing for days since, by the day each conversion happened, where its campaigns' clicks make so long a silence unlikely and it's longer than any the action showed in the quarter before. The days Google is still reporting are left out, and uploads are judged by their uploads instead. Bidding learns from what it counts, so until the tag is mended it learns from nothing.

• One conversion action silent for days after recording every day
• Every conversion tag and Analytics import silent at once

The whole account's last four weeks or more each converting fewer of its clicks than every week before them, and bringing fewer conversions in total, by the day each conversion happened and in the actions it counts now; or the same for conversion value while conversions held. Judged against the account's own weeks rather than a set percentage, so a season's usual swing isn't raised. A tag, a page or an import broken partway is the usual cause.

• Fewer conversions from each click, week after week
• Less conversion value from each click, week after week

Clicks billed from outside the places you chose to target, with the share of spend they take, country by country. Whether they are worth anything depends on whether you sell there, and on whether people there bought at your usual price; the figure is the spend outside your targets.

A campaign that targets places, whose location option is Presence or interest, Google's default, so its ads also show to people outside those places who search for them or show interest in them. For a business that serves people where they are, that's spend on people who aren't there, and the finding says how much went outside in the last 30 days. Goldbeater drafts Presence: people in, or regularly in, your locations.

No location targeted, so the ads can show in every country and territory, and the budget goes wherever the searches are. For a business that sells in some countries, spend in the others reaches people it doesn't sell to; the finding lists where the spend and the conversions came from.

A campaign whose name says a country or a region (UK, EMEA, Germany) while it targets somewhere else, somewhere beyond it, or no location at all: often a copy of another campaign that was never pointed at its own market. Goldbeater reads your own names, with no rule for you to keep: a code counts where most of your campaigns that carry it target what it stands for, beyond what chance would give, and a country's full name counts by itself. The finding shows what your campaigns named most like it target. A language in a name (FR, French) is read the same way on Display, Video, Demand Gen, App and Performance Max campaigns; on Search campaigns Google no longer uses the language setting and matches an ad by the language it's written in, so those aren't judged. Which is wrong, the name or the targeting, is yours to say.

• A campaign named for a country or region that targets somewhere else, or no location at all
• A Display, Video, Demand Gen or Performance Max campaign named for a language its languages leave out

A country your campaign targets that brings under half the conversions its other countries buy for the same spend, too far under to be chance, and under half their return where your conversion value is real. Only the countries it targets now are compared, and where you count a cart or a form on the way to a sale or a qualified lead, only the last step counts. The estimate is what its conversions cost above the campaign's own price; whether a customer there is worth more is yours to judge.

A country your campaign targets that brings over twice the conversions its other countries buy for the same spend, too far over to be chance, and over twice their return where your conversion value is real. Under manual bidding or Maximize clicks, where the campaign targets it by name, the fix raises its bids there by a fifth; where Smart Bidding already bids by location, the finding says when a campaign of its own would let it have more.

A state, province or other region of a country your campaign targets that brings under half the conversions the rest of that country buys for the same spend, too far under to be chance, and under half their return where your conversion value is real. Only the places it targets now are compared, and where you count a cart or a form on the way to a sale or a qualified lead, only the last step counts. Where it bought nothing, the fix excludes it from the campaign's locations; where the campaign targets it by name and bids manually or to Maximize clicks, the fix halves its bids there. The estimate is what its conversions cost above the campaign's price in the rest of the country.

A state, province or other region of a country your campaign targets that brings over twice the conversions the rest of that country buys for the same spend, too far over to be chance, and over twice their return where your conversion value is real. Where the campaign targets it by name and bids manually or to Maximize clicks, the fix raises its bids there by a fifth; where Smart Bidding already bids by location, the finding says when a campaign of its own would let it have more.

A city in a region your campaign targets that brings under half the conversions the rest of that region buys for the same spend, too far under to be chance, and under half their return where your conversion value is real. Only the places it targets now are compared, among its costliest cities, and where you count a cart or a form on the way to a sale or a qualified lead, only the last step counts. Where it bought nothing, the fix excludes it from the campaign's locations; where the campaign targets it by name and bids manually or to Maximize clicks, the fix halves its bids there. The estimate is what its conversions cost above the campaign's price in the rest of the region.

A city in a region your campaign targets that brings over twice the conversions the rest of that region buys for the same spend, too far over to be chance, and over twice their return where your conversion value is real. Where the campaign targets it by name and bids manually or to Maximize clicks, the fix raises its bids there by a fifth; where Smart Bidding already bids by location, the finding says when a campaign of its own would let it have more.

A three-hour block of the day in which a campaign spent real money over the last 30 days and converted nothing, where its own conversion rate says some should have, too far under to be chance. An ad schedule that leaves those hours out sends the budget to the hours that convert, and Smart Bidding works within one.

A three-hour block of the day that converts, or a day of the week, bringing under half the conversions the rest of the campaign's week buys for the same spend, too far under to be chance, in a campaign bidding manually or to Maximize clicks. Its bids are the campaign's until an ad schedule bid adjustment says otherwise; the finding says where to lower them. The estimate is what its conversions cost above the rest of the week's price.

A three-hour block of the day or a day of the week bringing over twice the conversions the rest of the campaign's week buys for the same spend, too far over to be chance, in a campaign bidding manually or to Maximize clicks. Its bids are the campaign's until an ad schedule bid adjustment says otherwise; the finding says where to raise them by a fifth.

A device, such as mobile phones, that brings under half the conversions your campaign's other devices buy for the same spend, too far under to be chance, and under half their return where your conversion value is real. Where the bid strategy applies a device bid adjustment, the fix halves the device's bids; where it uses one only at −100%, as Target ROAS and Maximize conversions do, the fix switches off a device that converted nothing, and one that converted is its landing page to look at first, then a campaign of its own. The estimate is what its conversions cost above the campaign's own price.

• A device converting far below the rest of its campaign
• A device converting far below its campaign under Smart Bidding

A device, such as mobile phones, that brings over twice the conversions your campaign's other devices buy for the same spend, too far over to be chance, and over twice their return where your conversion value is real. Under manual bidding, Maximize clicks or Target CPA its bids are the campaign's until a device bid adjustment says otherwise, and the fix raises them by a fifth; where Smart Bidding already bids by device, the finding says when a campaign of its own would let it have more.

A Search campaign also showing on Google search partners' sites or the Display Network, where that spend brings under half the conversions it buys on Google Search, too far under to be chance, and under half the return where your conversion value is real. Google turns search partners on for every new Search campaign. The estimate is what those conversions cost above the campaign's Google Search price.

• Search partners converting far below Google Search
• Display expansion converting far below Google Search

An age range, gender, household income band or parental status that brings under half the conversions the rest of your campaign's audience buys for the same spend, too far under to be chance, and under half their return where your conversion value is real. Where it bought nothing, the fix excludes it from the campaign; people Google couldn't place are never judged. The estimate is what its conversions cost above the campaign's price for everyone else.

• An age range converting far below the rest of its campaign
• A gender converting far below the rest of its campaign
• A household income band converting far below the rest of its campaign
• Parents, or people who aren't, converting far below the rest of a Display, Video or Demand Gen campaign

An age range, gender, household income band or parental status that brings over twice the conversions the rest of your campaign's audience buys for the same spend, too far over to be chance, and over twice their return where your conversion value is real, in a campaign bidding manually or to Maximize clicks. Its bids are the campaign's until a demographic bid adjustment says otherwise; the finding says where to raise them by a fifth. People Google couldn't place are never judged.

• An age range converting far better than the rest of its campaign
• A gender converting far better than the rest of its campaign
• A household income band converting far better than the rest of its campaign
• Parents, or people who aren't, converting far better than the rest of a Display campaign

An audience segment your campaign observes or targets, such as an in-market segment or a remarketing list, that brings under half the conversions the rest of the campaign buys for the same spend, too far under to be chance, and under half their return where your conversion value is real. Where it bought nothing, the finding says to exclude it; where the bid strategy applies an audience bid adjustment, to lower its bids. The estimate is what its conversions cost above the rest of the campaign's price.

An audience segment your campaign observes or targets that brings over twice the conversions the rest of the campaign buys for the same spend, too far over to be chance, in a campaign bidding manually or to Maximize clicks. Its bids are the campaign's until an audience bid adjustment says otherwise; the finding says to raise them by a fifth.

A Search or Shopping campaign set to Targeting, so its ads show only to the people in its audiences, when nothing in its name says it's meant for them; or a campaign named for an audience, such as remarketing, set to Observation, so its ads show to everyone. The first loses every searcher outside those audiences; the second spends a remarketing budget on people who never visited.

• A Search campaign shown only to its audiences by mistake
• A campaign named for an audience that shows its ads to everyone

Lists collecting people that no campaign that serves targets, observes or excludes, or that sit only in paused campaigns and audiences nothing signals, each place named. The work was done and none of it reaches the auction, unless Smart Bidding is picking up your customer lists by itself; the finding says where to check, and when a list belongs to something you no longer advertise.

A customer list your campaigns exclude, target or use as a signal that Google says hasn't been updated in 90 days or more, or sizes at under the 100 people it needs to serve. An exclusion list gone stale lets every customer since its last upload see ads meant for new ones. Where the account already keeps a fresher customer list, the finding names it to put in its place.

A site, app, YouTube channel or video your Display, Video or Demand Gen campaign paid for that brought under half the conversions the rest of the campaign buys for the same spend, too far under to be chance, and often nothing at all. Apps are judged one by one and together, a placement you picked by hand is taken out of its ad group rather than excluded, and in Demand Gen, YouTube, Gmail and Discover are channels to choose rather than placements. The estimate is what it cost above the campaign's own price.

• A site, YouTube channel or video converting far below the rest of its campaign
• A mobile app converting far below the rest of its campaign
• A campaign's mobile apps together converting far below the rest of it
• A placement you picked converting far below the rest of its campaign

A site or app drawing over twice the clicks your campaign's other sites and apps would from the same impressions, too far over to be chance, while its clicks convert far less than theirs or not at all. That's the mark of clicks nobody meant, most often in games and apps, and you pay for each one Google's filters don't catch.

A site, app, YouTube channel or video your Display, Video or Demand Gen campaign paid to show on, whose name holds a word you've listed in Settings as one you never want your ads beside. Each finding lists a campaign's placements for one word, with what each cost, so you can read them before excluding them. With no words listed it raises nothing.

The YouTube channels, sites and apps your Display, Video or Demand Gen campaign paid for that a published list, or their own name, marks as made for children, as games or as job sites: the kinds advertisers most often keep their ads off. Goldbeater gathers each kind into one finding a campaign, lists the placements, and says how they converted beside the rest of the campaign: far below it, with what that cost, or no differently, where keeping your ads off them is a question of where you want to be seen.

• Ads showing on YouTube channels, sites and apps for children
• Ads showing on games sites, in game apps and on gaming channels
• Ads showing on job and employment sites

A Display or Video campaign, live or about to launch, that keeps its ads off no site, app or video, itself, in an ad group, through a list or on the account; and a Display campaign that excludes no category of apps, so its ads can show in any of them. Nothing is lost until it spends, and once it does Goldbeater judges each placement and its apps together.

• A Display or Video campaign with no placement excluded anywhere
• A Display campaign that can show its ads in any mobile app

A list of sites, apps or videos kept out, carried by most of your campaigns of the same kind and named alike, more than chance would put on them, and missing from one, which still shows where the others don't.

None of the five kinds of sensitive content Google lets an account exclude (tragedy and conflict, sensitive social issues, profanity, sexually suggestive, sensational and shocking) is excluded, while enabled campaigns, Performance Max among them, spend on the Display Network. The ads can appear beside any of them; the finding says what they spent there, and its share of the account's spend.

A Display or Video campaign with no frequency cap. For Display that's the setting Google recommends, letting it decide how often your ads show, so the finding says what's left to Google and when a cap is worth setting; for Video, nothing limits how often one viewer sees the same ad.

A Display, Video or Demand Gen ad group with optimized targeting on, which shows its ads beyond the audiences you chose to people Google expects to convert, while its campaign bids by clicks or impressions, or records no conversions. Google recommends it with Maximize conversions or Target CPA; without conversion data it widens your reach with nothing to steer it.

An account spending with no conversion action counting toward its goals: none set up, or every one secondary. Every campaign's conversions read nothing, no campaign's results can be told apart from another's, and a bid strategy that bids to conversions has nothing to bid to.

A conversion action your campaigns count that recorded conversions before and has recorded none for longer than its own rate explains, often a tag removed from the site: Smart Bidding keeps bidding toward it, and where no campaign bids on conversions, their Conversions column reads short. Or imported conversions that stopped arriving or are failing, which Google reports within days.

• A conversion your campaigns count that stopped recording
• Conversion imports that stopped arriving
• Conversion imports that are failing

A campaign whose last few weeks' clicks brought under a quarter of the conversions its clicks brought the month before, in the actions it still counts: a tag stopped firing, an import stopped arriving, or something changed on the page. Smart Bidding learns from what the campaign counts, so it adjusts to the fall as if it were real.

A campaign whose clicks reach the checkout and then buy under a quarter as often as your other campaigns' do, too far under to be chance. Either its sales are made and not recorded, so its reports and its bidding can't see them, or its checkout is losing buyers you already paid to bring that far. Google Ads shows the same figures for both, and one test purchase from one of its ads tells them apart. A sale is counted once however many conversion actions record it, a campaign in its first two weeks is an early read, and the estimate is the purchases its checkouts would have ended in at your other campaigns' rate.

The whole account's last three weeks or more each converting more of its clicks than every week before them, and more in total, by the day each conversion happened. A sale or a new page explains it; a tag firing twice, or an action counting what another already does, inflates what bidding buys toward. Where Goldbeater already says the action that rose is counted twice, it's left to that finding.

Two conversion actions counting one event: two tags on the same purchase, a begun checkout counted beside the purchase it leads to, or a lead counted as a form and again as the qualified lead it becomes. Each inflates the numbers Smart Bidding optimizes toward, so it buys more of whatever produced the extra count.

• Two conversion actions counting the same event
• An add-to-cart or begun checkout counted beside the purchase
• A lead counted twice, as a form and again as a qualified lead

One conversion action counted at least 80% as many times as the campaign was clicked, where a sale or lead tag on the confirmation page records a share of the clicks. It usually fires on a page every visitor reaches, and bidding and every report read each visit as a result.

A campaign whose clicks make purchases, sign-ups or leads that its own conversions leave out, because the action is secondary or isn't among the campaign's goals, while what it does count is next to nothing. A conversion bid strategy optimizes toward the little it counts, and its reports judge the spend by it. Leaving an action out can be deliberate, so whether to count it is your call.

An add-to-cart, a begun checkout or a lead that bidding counts, carrying as much value as its share of the sale it leads to or more. If a fifth of checkouts end in a sale, a checkout is worth at most a fifth of one, and it has earned nothing yet. Value bidding chases it as if it were revenue.

A sale or lead that a campaign bidding on conversion value counts, recorded at a value of 1 or less each, Google's default for a new action and what a tag passing no order value leaves. Value bidding then bids on how many conversions there are rather than what they're worth, and a target ROAS on it means nothing.

Conversion actions feeding one campaign's bidding that credit conversions by different attribution models, or that count the same kind of conversion over different conversion windows, so the strategy weighs numbers counted by different rules.

A conversion action bidding learns from that gives each conversion to the last ad click, so the searches and ads that started the customer earn nothing and bidding cuts them back. Data-driven attribution shares the credit among the clicks that led to it.

A conversion action bidding learns from that counts conversions for fewer than the 7 days after a click Google recommends. A customer who buys later is never counted, and bidding learns only from the quickest.

Auto-tagging switched off while conversions your bidding counts rely on it, a custom parameter your tracking uses that nothing gives a value, so it arrives empty, or one click's ID pasted into a final URL, so every click carries it. The spend is real, and the record of where it came from is lost or wrong.

• Auto-tagging off while conversions rely on the click ID
• A tracking parameter no level gives a value, so it arrives empty
• One click's ID pasted into a final URL, so every click carries it

Call reporting switched off while your campaigns show a call asset. Google shows your own number and tracks none of the calls, so no call from an ad is reported or counted as a conversion.

A brand Search campaign spending while Google Ads holds no organic result for any search, which it does only with a Search Console property linked. Without it the paid and organic report is empty: you can't see what your ads add on searches your site already ranks for, or the searches you rank for that no keyword covers.