Guide

What are the Google Ads targeting options, and which ones pay off?

Google Ads targeting by device, demographic, audience and city: where each one's results are reported, and when a bid adjustment or an exclusion is the fix.

In short
  • Google Ads sorts its targeting options into audience targeting, who sees an ad, and content targeting, where it shows. Devices, locations and the ad schedule are set on the campaign beside them.
  • Each has its own report: devices under Insights and reports › When and where ads showed, and age, gender, household income and audience segments on the Audiences page.
  • A device, age range, audience or city is worth acting on when it bought under half the conversions its spend would have at the rate of the rest of its campaign, and so few would be unlikely by chance.
  • A bid adjustment changes the bids for one of them by a percentage, and only manual bidding and Maximize clicks apply every kind. Target CPA applies a device adjustment to its target, and the rest of Smart Bidding uses one only at −100%, which switches the device off.
  • Goldbeater reads every campaign's devices, demographics, audiences and cities every day, and raises one that converts far below or far above the rest of its campaign by more than chance, with the change drafted where a setting fixes it.
Goldbeater checks this every 24 hours
  • A device converting far below the rest of its campaign
  • A device converting far better than the rest of its campaign
  • An age range, gender, income band or parental status converting far below the rest of its campaign
  • and 6 more

Each week Goldbeater's AI agent follows up on the findings. A free first look shows how many problems these and 154 other checks find on one of your accounts, and what they cost. No card.

The manual way comes first, in Google Ads itself. Then the same problem as Goldbeater finds it, every day, without anyone asking.

Google Ads targeting options: who sees your ads, and where

Google sorts the ways to target an ad into two kinds. Audience targeting is who you want your ads to reach: people by who they are, their interests and habits, what they’re actively researching, and people who have already interacted with your ads, website or app. Content targeting is where your ads show: the keywords, topics and placements you choose.

A campaign’s own settings sit beside them: its locations, its ad schedule, and a bid adjustment for each device. Every one of these splits a campaign into parts, and Google Ads reports what each part cost and converted. This guide reads those parts: whether a device, an age range, an audience or a city returns what the rest of its campaign does, and what you can change when it doesn’t.

TargetingIts partsWhat you can change
DevicesComputers, mobile phones, tablets, TV screensA bid adjustment, down to −100%
DemographicsAge, gender, household income, parental statusAn exclusion, or a bid adjustment
Audience segmentsYour data, in-market, affinity and moreTargeting or Observation, a bid adjustment, an exclusion
LocationsCountries, states, citiesAn exclusion, or a bid adjustment

Hours and days have a guide of their own, the ad schedule guide, and countries and states are in the location targeting guide. Placements are in the Display campaign guide, which reads where a campaign’s ads appeared.

Where Google Ads reports each kind of targeting

Devices and places are on one page, and demographics and audience segments on another:

  1. Go to Campaigns › Insights and reports › When and where ads showed and select the Devices tab, which has a row for computers, mobile phones, tablets and TV screens.
  2. For places, select the Matched locations tab on the same page.
  3. For demographics, go to Campaigns › Audiences, keywords, and content › Audiences. In the Demographics module choose Campaign view, select a tab (Age, Gender, Household income or Parental status), then Show table.
  4. For audiences, find the Audience segments module on that page, choose Campaign view, and select Show table.
  5. Set the date range to the last 30 days. Each table shows cost and conversions, and the columns icon adds more.

Any other table splits by device too: select the segment icon above it, then Device. A segment can split a table by time as well, but Google Ads segments by Day only when the date range is 16 days or less, so read a month as its total.

Two things about what the rows hold. The Devices tab can show a row called Other: Google says some devices, feature phones among them, fall under none of the four. And an audience’s row counts only the people in that segment who also met the campaign’s targeting, so the rows don’t add up to the campaign. On Search, Google says a person in several of a campaign’s segments is counted under the one with the highest bid adjustment.

How to tell a weak device from a noisy one

A row with a high cost per conversion, or with no conversions at all, isn’t a finding yet. Small parts of a campaign swing from month to month, and across a few dozen rows some will look bad by luck. Work each one against the rest of its campaign:

  1. Take one campaign’s figures by device, from the Devices tab or the Device segment. For one device, add up the other devices’ cost and their conversions, and divide the first by the second. That is what they paid for a conversion.
  2. Divide the device’s own cost by that price. The answer is what its spend would have bought at their rate.
  3. If it bought at least half of that, leave it. If it bought under half, ask whether so few could be luck.
  4. In Google Sheets, BINOMDIST answers that. Give it the device’s conversions, the campaign’s conversions, the device’s share of the campaign’s cost, and TRUE. It returns the chance that a device converting like the rest would have that few, or fewer.
  5. Multiply the chance by the number of rows you compared, in every campaign, since each was another chance for luck to show. Under 5%, the gap is worth acting on.
By device, 30 days
Sample data
DeviceCostConv.Cost / conv.ExpectedBy luck
All devices$2,01621$96.00——
Computers$1,17021$55.71——
Mobile phones$7060—11.30.01%
Tablets$1400—1.622%
Sofas — Phrase sold every sofa on a computer, and two rows read 0 conversions. Mobile phones spent $706, which would have bought 11.3 at the $62.38 the other devices paid: none is a 0.01% chance, and 0.13% after the 11 rows compared across the account. Tablets spent $140, which would have bought 1.6: none happens about one month in five. Computers have nothing to be set against, since no other device converted.

That is the whole difference between the two rows. Both bought nothing, and only one spent enough for nothing to mean something. A part that converts far better takes the same sum turned over: over twice what its spend would have bought, where so many is as unlikely.

An age range, a gender, an audience segment and a city are read the same way, each beside the rest of its campaign, and a city beside the rest of its state in that campaign.

The count misses two things. A conversion can be worth more on one device than another, so where your account records a real conversion value, compare the return as well, and act when both are under half. And a device that buys nothing is sometimes a page that fails on it. Open your landing page and your checkout on that device before you change a bid; the landing page guide has what to look for.

Google Ads bid adjustments, and which bid strategies apply them

A bid adjustment is a percentage increase or decrease in your bids for one part of a campaign. Google’s example is a maximum CPC bid of $1 with a 20% increase on mobile devices, which bids $1.20 for a search on a phone. What an adjustment does depends on the campaign’s bid strategy:

Bid adjustmentRangeManual, Maximize clicksTarget CPAOther strategies
Device−100% to +900%AppliedMoves the target−100% only
Location−90% to +900%AppliedNot usedNot used
Ad schedule−90% to +900%AppliedNot usedNot used
Audience−90% to +900%AppliedNot usedNot used
Demographic−90% to +900%AppliedNot usedNot used

The other strategies are Target ROAS, Maximize conversions, Maximize conversion value and Target impression share. Of Smart Bidding, Google’s page says: “If you make a manual bid adjustment to your automated Smart Bidding strategy, it won’t be supported.” Under Target CPA a device adjustment changes the target, not the bid: with a $10 target, +40% on mobile makes the target $14 there.

Two things hold under every strategy. A device at −100% gets no ads. And where an adjustment isn’t used, Google says Smart Bidding “will still consider targeting and exclusions”, so an age range, an audience or a place you exclude stays excluded.

The campaign type matters too. A Search campaign takes every kind in the table. Performance Max, App, Smart and Demand Gen campaigns take none. To set one:

  1. For a location, an ad schedule, an audience or a demographic group, go to Campaigns › Audiences, keywords, and content and select its page. For a device, go to Campaigns › Insights and reports › When and where ads showed.
  2. Select the row, then the pencil icon in its Bid adj. column.
  3. Choose Increase or Decrease, enter a percentage, and select Save. To remove an adjustment, delete the value.

Adjustments of different kinds multiply, and together they can’t raise a bid by more than 900% or lower it by more than 90%.

Google Ads device targeting is a bid adjustment on each device

Google’s page on device targeting sets it one way: a bid adjustment for each device, on the Devices tab. Computers are desktops and laptops with screens larger than 7 inches. Mobile is hand-held devices that include a phone. Tablets are a row of their own. TV screens are devices that stream TV content, and Google offers them as a target in Display and Video campaigns alone.

A device adjustment can sit on a campaign or on an ad group. Where both set one for the same device, the ad group’s is used, unless the campaign’s is −100%. That figure is how a device is switched off: Google’s instruction is to decrease the bid by 100% for that device, and a device adjustment is the one kind that goes that low. What you can do about a weak device follows the bid strategy:

  • Manual bidding or Maximize clicks: lower the bids on a device converting far below the rest, and raise them on one far above.
  • Target CPA: the same adjustment, which raises or lowers the target on that device.
  • Target ROAS, Maximize conversions, Maximize conversion value or Target impression share: a device that buys nothing can be switched off at −100%. One that converts at a worse price has no bid to lower, because Smart Bidding already sets bids by device. Look at the page it lands on first, and if nothing is wrong there, consider a campaign of its own for that device, with its own budget and target.
  • Performance Max: no bid adjustment. It has device targeting controls of its own, which Google announced as fully available in August 2025.

Display, Video and Demand Gen campaigns have advanced options besides: operating systems, device models, carriers and wireless networks.

Google Ads demographic targeting: exclusions, bids and the Unknown group

Demographic targeting reaches people likely to be in an age range, a gender, a parental status or a household income band. A Search campaign has age, gender and household income. Display, Video and Demand Gen campaigns add parental status. Household income runs from the top 10% to the lower 50%, and Google offers it on Search in 20 countries, the United States among them.

By default every demographic group is targeted. Each tab also has an Unknown row: the people whose age, gender, parental status or household income Google hasn’t identified. Leave it in. Google warns that excluding Unknown “might prevent a substantial number of people from seeing your ads”.

You can change two things for a group. The first is an exclusion, which works under any bid strategy:

  1. Go to Campaigns › Audiences, keywords, and content › Audiences.
  2. Under Exclusions, select Edit exclusions, choose Campaign or Ad group, and select the one to change.
  3. On the Browse tab, tick the demographic segments to exclude, and select Save.

An exclusion on the campaign covers every ad group in it. Google’s example is a spa for women that excludes Male once, at the campaign.

The second is a bid adjustment on an age range, a gender or a household income band, from −90% to +900%. It changes bids in a campaign that bids manually or to Maximize clicks, and no Smart Bidding strategy uses it. So under Smart Bidding, exclude a group that buys nothing, or one whose customers are worth less to you. For a group that converts at a worse price, see whether your ads and landing page speak to it. Performance Max has age controls of its own, and gender controls in beta, by Google’s announcement of August 2025. A Demand Gen campaign takes no bid adjustment at all, and the Demand Gen guide reads one by age.

Audience targeting in Google Ads: Observation, Targeting and in-market audiences

An audience segment is a group of people, built by Google or by you. Google lists your data segments (people who visited your website or app, and your customer lists), in-market segments, affinity segments, detailed demographics, life events and custom segments. A Search campaign can use your data, in-market, affinity and detailed demographics.

In-market audiences are the ones nearest a sale. Google describes them as people who are “researching products and are actively considering buying a service or product like yours”. Affinity segments are wider: people grouped by what they’re passionate about, their habits and interests.

Every audience is added under one of two settings, and the setting decides more than the audience does:

  • Targeting narrows the campaign or ad group to the people in its audiences. Google recommends it for Display campaigns.
  • Observation leaves who sees your ads unchanged. It reports how each audience performs, and lets you set a bid adjustment on it. Google recommends it for all Search campaigns, and for Shopping.

So a Search campaign meant for every searcher belongs on Observation, and a remarketing campaign meant for past visitors belongs on Targeting. The other way round, the first shows only to the people on its lists, and the second spends a remarketing budget on everyone. To see or change the setting, select Edit audience segments in the Audience segments module, choose the campaign or ad group, and select Targeting or Observation.

Under Smart Bidding, Google uses first-party segments added under Observation as signals, so an observed list informs the bidding with no adjustment from you. Under manual bidding or Maximize clicks an audience takes a bid adjustment from −90% to +900%. Google’s example raises bids by 25% for people who viewed your website in the last 30 days.

An audience can be excluded too, on Search, Display, Demand Gen, Standard Shopping, Video and Performance Max campaigns. Google recommends observing how a segment performs before you exclude it. Its page says you can remove “specific types of segments (for example, affinity or your data)”, and carries a note that audience exclusions “are only applicable for first-party data”. Your own lists can be excluded on either reading. For one of Google’s segments, see whether the Exclusions module offers it.

Lists no campaign uses are the subject of the unused audience segments guide, and the lists you upload of the Customer Match guide. The RLSA guide works a list of past visitors in a Search campaign by hand.

A city is read against the rest of its state

The location targeting guide reads a campaign by country and state. One level down is the city, in the same report: Google says its location reports show performance “at the most specific level available for targeting”, which in the United States reaches the postal code.

Set a city beside the rest of its state in the same campaign, with the sum above. Cities multiply the rows to compare, so start with the ones that spent the most. A city with a few dollars of spend can’t be told from luck.

What you can change is what a state takes. An exclusion works under any bid strategy. A location bid adjustment, from −90% to +900%, works under manual bidding or Maximize clicks, and where a city and its country both carry one, Google uses the more specific: in its example of +50% for France and +100% for Paris, people in Paris get Paris’s alone. Smart Bidding already bids by where a person is, in Google’s words “down to the city”, and uses no location adjustment.

What it looks like when Goldbeater finds it

Every day Goldbeater reads each enabled campaign’s last 30 days by device, by age range, gender and household income, by audience and by city, and sets each part against the rest of its campaign with the sum above. It raises one that bought under half the conversions its spend would have at the rest’s rate, or over twice, on at least $20, where a gap that size is unlikely by chance once every pair tested in your account is allowed for (5% overall). Where your conversion value is real, its return has to be under half, or over twice, as well. Where you count a cart or a form on the way to a sale, only the last step counts.

  • A device far below the rest of its campaign. Under manual bidding, Maximize clicks or Target CPA, Goldbeater drafts the bid adjustment that halves the device’s bids, never lower than −90%, and offers −100% beside it for a device that bought nothing. Under Target ROAS, Maximize conversions, Maximize conversion value or Target impression share, it drafts −100% for a device that bought nothing; for one that converted, the finding is advice: its landing page first, then a campaign of its own. In Performance Max it is advice too. The finding carries an estimate of what the device cost above the other devices’ price.
  • A device far above it. Under manual bidding, Maximize clicks or Target CPA, Goldbeater drafts a fifth more on the device’s bids, never past +900%. Under the other strategies and in Performance Max there is no bid to raise, and the finding says when a campaign of its own would let the device have more. A device whose other devices are mostly one raised as far below isn’t raised: it’s the same gap.
  • An age range, gender, household income band or parental status far below the rest of the campaign’s audience. Unknown is never judged, and parental status is read on Display, Video and Demand Gen campaigns. For an age range, gender or income band that bought nothing, Goldbeater drafts its exclusion on a Search, Display, Video or Demand Gen campaign; for parental status it advises one. For a group that converted, it advises a −50% bid adjustment in each ad group where the campaign bids manually or to Maximize clicks, and under Smart Bidding, a look at the ads and landing page before excluding anyone.
  • One far above it, in campaigns bidding manually or to Maximize clicks alone, since those apply a demographic bid adjustment. The finding advises +20% on the group in each ad group, and drafts nothing.
  • An audience far below or far above the rest of its campaign or ad group, observed or targeted. Below, the finding says to exclude one that bought nothing, and to lower one that converted by −50% where the bid strategy applies an audience adjustment. Above is judged in campaigns bidding manually or to Maximize clicks alone, and advises +20%. Both are advice, with no change drafted.
  • An audience setting at odds with a campaign’s name. For a Search or Shopping campaign or ad group set to Targeting, with audiences in it, whose name doesn’t say it’s meant for an audience, Goldbeater drafts the switch to Observation. For one named for an audience (remarketing, RLSA, visitors, customer lists) and set to Observation, the finding says to switch it to Targeting if that’s what it’s for.
  • A city far below or far above the rest of its region, in the same campaign, among the 1,000 campaign and city pairs that spent the most in the places your enabled campaigns target. For a city that bought nothing, Goldbeater drafts its exclusion where that only narrows what the campaign targets. For one the campaign targets by name under manual bidding or Maximize clicks, it drafts half the bids, or a fifth more for a city far above. Anywhere else the finding is advice.
Finding
Sample data
HighHigh severityCost per conversion

Sofas — Phrase spent $706 on mobile phones and converted nothing, against $62 a conversion on its other devices

Sofas — Phrase

$706Estimated, a month

In the last 30 days mobile phones brought no conversion value for $706, where the campaign's other devices returned a ROAS of 6.77.

This campaign's own ROAS on its other devices, last 30 days

What Goldbeater saw

The campaign spent $706 on mobile phones, 35% of its spend, for 0 conversions where 11.3 were expected at its other devices' rate: under a 0.2% chance of so few by luck, allowing for the 11 campaign and device pairs tested across the account.

Cost
$706
Share of the campaign's spend
35%
Expected at its other devices' rate
11.3

The last 30 days

FigureMobile phonesIts other devices
Spend$706$1,310
Conversions, beyond chance021
Cost per conversion—$62.38
Conversion value$0$8,870
ROAS0.006.77

Mobile phones

Its other devices

Why it matters

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.

What to do

Stop showing Sofas — Phrase's ads on mobile phones: a −100% bid adjustment, the one device adjustment its bid strategy uses. It's the change below.

What would change
  • Stop showing ads on the deviceMobile phones in Sofas — PhraseNo adjustment−100%

It reaches your account when you approve it in the change set and press Apply. An applied change can be undone.

This is “A device converting far below the rest of its campaign” on a sample account. Goldbeater runs it on yours every 24 hours, and your AI analyst answers what you ask about any finding.

Get a free first look
Sample data. Cedar & Pine Interiors is an invented account, and its figures are made up.

That is the campaign worked above. The 11.3 conversions expected are the table’s, and the finding’s chance is the table’s 0.13% after the 11 rows compared, rounded up. Tablets bought nothing too and aren’t raised: $140 is too little to tell from luck. Sofas — Phrase bids to a Target ROAS, which uses a device adjustment at −100% alone, so that is the change drafted. A device that sells nothing can be a page that fails on it, so load the landing page and the checkout on a phone before you approve it.

Hours and days are in the ad schedule guide, countries and states in the location targeting guide, and the audit checklist has every targeting line with where to find it in Google Ads.

Questions

What are the targeting options in Google Ads?
Google sorts them into audience targeting and content targeting. Audience targeting is who sees your ads: demographics such as age and gender, and audience segments such as in-market segments and your own lists. Content targeting is where they show: keywords, topics and placements. A campaign’s locations, ad schedule and device bid adjustments sit beside them.
Do bid adjustments work with Smart Bidding?
Mostly not. Google says a manual bid adjustment on a Smart Bidding strategy isn’t supported. Devices are the exception: Target CPA applies a device adjustment to its target, and Target ROAS, Maximize conversions and Maximize conversion value use one only at −100%, which stops the ads on that device. Targeting and exclusions still apply under every strategy.
How do I exclude a device in Google Ads?
Set its bid adjustment to −100%. Go to Campaigns, Insights and reports, When and where ads showed, open the Devices tab, select the pencil icon in the device’s Bid adj. column, and decrease the bid by 100%. A device adjustment is the one kind that goes to −100%; the others stop at −90%.
Should I exclude the Unknown age or gender group?
Usually not. Unknown is everyone whose age, gender, parental status or household income Google hasn’t identified, and Google warns that excluding it might prevent a substantial number of people from seeing your ads. Exclude it only if you mean to restrict the campaign to a narrow audience.
Does adding an audience to a Search campaign narrow who sees my ads?
Only under the Targeting setting, which shows the ads to the people in the audience alone. Under Observation, which Google recommends for all Search campaigns, your reach is unchanged: the audience is reported on, and can take a bid adjustment.
What are in-market audiences in Google Ads?
Segments of people Google judges to be in the market: researching products and actively considering buying a service or product like yours. Search, Display, Video, Standard Shopping and Demand Gen campaigns can use them. On a Search campaign, add them under Observation to see how each converts beside the rest of the campaign.
How much data do I need before judging a device or an age range?
Enough that its spend would have bought several conversions at the rest of the campaign’s rate. A part that bought nothing can’t be told from luck until its spend would have bought about three, and more when you compare many rows at once. Below that, a zero is noise.

How Goldbeater checks this

These checks run every day, and whenever you ask.

Each finding shows its evidence, an estimate of what it costs a month where it costs money, and, where a setting fixes it, a change drafted for you to approve.

A check is code that runs the same way every time, so a new finding means your account changed. Each week Goldbeater's AI agent follows up on what the checks find, and your AI analyst answers any question about your account.
ChecksA finding is worth
  • A device converting far below the rest of its campaignFixing it lowers the Cost per conversion
  • A device converting far better than the rest of its campaignFixing it raises the Conversions
  • An age range, gender, income band or parental status converting far below the rest of its campaignFixing it lowers the Cost per conversion
  • An age range, gender, income band or parental status converting far better than the rest of its campaignFixing it raises the Conversions
  • An audience converting far below the rest of its campaignFixing it lowers the Cost per conversion
  • An audience converting far better than the rest of its campaignFixing it raises the Conversions
  • An audience setting at odds with what the campaign is forFixing it lowers the Risk
  • A city converting far below the rest of its regionFixing it lowers the Cost per conversion
  • A city converting far better than the rest of its regionFixing it raises the Conversions

7-day trial · nothing charged for 7 days

Try it for a week. See the money it saves. It pays for itself.

Start the 7-day trial

Or connect one account for a free first look, no card needed: how many problems it finds, and what they cost.