How do I optimize keywords in Google Ads and lower my CPC?

It’s the Search keywords table, with four columns added and a date range long enough to judge by.

1. 01Go to Campaigns › Audiences, keywords, and content › Search keywords.
2. 02Select the columns icon above the table, then Modify columns. Add Conversions, Cost / conv., Conv. value / cost and Avg. CPC, and select Apply. You can save the set under a name and bring it back from the same menu.
3. 03Select the date range at the top right of the screen and set it to the weeks you want to read.
4. 04Select the filter button above the table and filter to one campaign, so each keyword is read beside the others of its campaign.
5. 05To see the weeks one at a time, select the segment icon, then Time, then Week. Day is offered only when the date range is 16 days or less.
6. 06To keep a copy, select the three-dot icon, then the download button. A segment you applied arrives as a column in the file.

What the four columns hold, by Google’s definitions:

• Conversions: the conversions of your primary conversion actions. They are counted to the day of the click, so a click from last week that converts this week adds to last week.
• Cost / conv.: your cost divided by the number in the Conversions column.
• Conv. value / cost: your conversion value divided by your cost, which Google calls an estimate of your return on investment. It only means something where your conversions carry real values.
• Avg. CPC: the cost of your clicks divided by the number of clicks.

Google says complete data for a day isn’t available until 3 p.m. Pacific time the next day, so end the date range at least a day back. Its Days to conversion segment is there, Google says, to understand how long to wait for a given day before you interpret its conversion numbers. The newest weeks go on filling in for that long.

A keyword’s cost per conversion can’t tell you by itself. A keyword with two conversions is one sale away from looking fine, and one with none may have spent too little to have earned one yet. Two questions settle it: how far is the keyword below the rest of its campaign, and how often would luck alone put it there?

1. 01On Search keywords, set the date range to the last 13 whole weeks and filter to one campaign. Add up the Cost and the Conversions of its keywords.
2. 02Take one keyword out of both sums. The other keywords’ cost divided by their conversions is what the rest of the campaign paid for a conversion.
3. 03Divide the keyword’s cost by that price. That is the number of conversions its spend would have bought at their rate.
4. 04If it bought half of that or more, leave it. A keyword that near its campaign isn’t one to act on alone.
5. 05If it bought under half, work out how often luck does that. In a spreadsheet, type =BINOMDIST(its conversions, the campaign’s conversions, its share of the campaign’s cost, TRUE). The answer is the chance that a keyword converting as well as the others would show that few.
6. 06Multiply that chance by the number of keywords you put through the same sum, since each one was another chance for luck to show. If the answer is still under 5%, the keyword is far below its campaign.

So the two rows that look alike in the keyword table are different things. "bed frame" spent what 8.6 conversions cost the rest of the campaign and bought none, and luck doesn’t do that. "storage bed" is two sales where about four were expected: one more would have put it at $81.44 a conversion. Acting on it now is acting on noise.

Cedar & Pine’s purchases carry their order values, so the same comparison holds on return: "storage bed" returned 2.13 times its cost against the other keywords’ 5.39, and "bed frame" returned nothing.

One caution on a range this long. If the campaign’s own rate moved during the 13 weeks, a keyword that spent mostly in its bad weeks looks worse than it is. Segment the table by Week and compare the keyword with the others a week at a time. A keyword under $20 in the whole range has spent too little to compare at all.

• It converted nothing, and luck doesn’t explain it. Read the searches it bought before you pause it. Go to Campaigns › Insights and reports › Search terms, select the columns icon, then Attributes, tick Keyword and select Apply: the column says which of your keywords matched each search. If they are searches for something you don’t sell, the fault is the matching, and a negative keyword stops each. Six of the searches "bed frame" bought were for a garden bed frame, and the negative keywords guide works them. If the searches are right and none converts, pause the keyword.
• To pause one, tick it on Search keywords, select Edit at the top of the table, and pause it. Google says a paused keyword won’t show ads or accumulate new costs until it’s restarted.
• It converts, at more than twice the campaign’s price, and luck doesn’t explain it. Where its bid is yours, lower it: the next section says when it is. Where the bid strategy sets the bid, read its searches for the ones that never convert, add negatives for them, and pause it only if what it buys still isn’t worth the price to you.
• Its match type is too wide. A match type can be changed from the pencil icon beside the keyword, and Google says what that does: you’re removing the original keyword and creating a new one. The original’s statistics stay with it, and show once you adjust the filters to include removed keywords, so the weeks you judged aren’t on the new keyword. The broad match guide covers which match type to move to.
• It’s under half, and within luck. Leave it, and run the same sum next month. The test is the same one with more conversions in it.

Small keywords are the ones this misses. A keyword at a few dollars a week never spends enough to be judged alone, and several of them can share a word that never converts. Add up the keywords of one campaign that hold the word and put the sum through the same steps, as if they were one keyword.

A keyword’s bid is its maximum cost per click, max. CPC: the highest amount you’re willing to pay for a click on your ad. Google says a click won’t cost more than the bid you set, bid adjustments and manual bidding on search partners aside, and that a higher bid generally helps your ad show in a higher position. Who sets it depends on the campaign’s bid strategy:

• Manual CPC: you do. You set a default bid for the ad group and, where you want, a bid for a single keyword, which is used instead of the ad group’s whenever that keyword triggers your ad.
• Maximize clicks: Google sets your max. CPC bids to bring the most clicks within your budget. You can cap them with a maximum CPC bid limit.
• Target CPA, Target ROAS, Maximize conversions and Maximize conversion value: Smart Bidding sets a bid for every query, at the time of each auction. Google’s page on editing keywords says the Max. CPC column then has no pencil button, and its page on columns says a bid marked Auto can’t be edited.
• Enhanced CPC: Google says it is no longer available for Search and Display campaigns since the week of March 31, 2025, and that campaigns still set to it are effectively using Manual CPC.

To change a keyword’s bid under Manual CPC:

1. 01Go to Campaigns › Audiences, keywords, and content › Search keywords and tick the keywords whose bid you want to change.
2. 02Select Edit, then Change max. CPC bids.
3. 03Set a new bid, or raise or lower the bids you have. Leave the amount blank to go back to the ad group’s default bid.
4. 04Select Apply.

For one keyword, the pencil icon beside the amount in the Max. CPC column does the same. A lower bid can show your ad less often and lower on the page, so run the comparison again before you lower it further.

The other way a keyword goes wrong is against its own past. Google Ads compares two date ranges for you:

1. 01Select the date range at the top right of the screen and set it to the last 4 whole weeks.
2. 02Switch on Compare and choose Previous period, then select Apply. The dates apply across your account.
3. 03In the tables Google’s page names, those of campaigns, ad groups and ads, each column now carries the change in percent, and the double arrows in its header open it into four: this range, the one before, the change and the change in percent.
4. 04For keywords, the segment by Week shows the same thing a week at a time: set the range to the last eight whole weeks and add up the last four and the four before.

Then read three things before you decide a keyword has slipped:

• Whether it’s more than a sale or two. A keyword with five conversions a month moves by a fifth when one of them doesn’t come.
• Whether its campaign’s other keywords moved the same way. If they all fell together, the cause is something the whole campaign shares, not the keyword. The not-converting guide covers a campaign whose conversions fell.
• Which part moved. Avg. CPC says whether its clicks got dearer, Conv. rate whether they convert less often, and Value / conv. whether each conversion is worth less. Each sends you somewhere else: the auction, the searches and the page, or what was bought.

The newest weeks also read low until their conversions arrive. Conversions are counted to the day of the click, so a range that ends yesterday is still filling in.

Avg. CPC is the cost of your clicks divided by their number. The price of each click is set in the auction: Google says you pay what’s minimally required to clear the Ad Rank thresholds and beat the Ad Rank of the competitor immediately below you. So a click costs more for one of these reasons, each from Google’s own pages:

• The competition. Google says the competitiveness of an auction can affect your actual cost per click, and that it’s influenced by the competitor immediately below you and by the ones below them. Tick the keyword on Search keywords and select Auction insights to see who shares its auctions and how often they show above you. A keyword needs a minimum of activity to have the report. The Auction insights guide reads it column by column.
• Your ad quality. Google says higher quality ads typically cost less per click, and that with low quality ads your actual CPC may be close to your maximum CPC even when competition is low. Lower quality ads also have higher thresholds to clear. Quality Score itself isn’t an input in the auction. It is the diagnostic that says which part to mend, and the low Quality Score guide covers each part.
• Where your ad shows. The thresholds are typically higher above the search results, so a click there often costs more than one below them, even with no advertiser immediately below you.
• The bid. Under Manual CPC the max. CPC is the ceiling, and a raised bid lets clicks cost more. Under Smart Bidding the strategy bids up or down by how well each search is likely to perform, and Google notes that a click can then cost more than a max. CPC.
• The search. Thresholds vary with the nature of the search terms, the person’s location and their device, so one keyword pays different prices on different searches.

A Google Ads CPC increase is a problem only where it bought nothing. Put Cost / conv. beside Avg. CPC for the same two date ranges. Dearer clicks that convert at the same cost per conversion are worth what they cost. And set the keyword beside the rest of its campaign: if every keyword’s clicks got dearer together, the cause is something the whole campaign shares, not the keyword.

To reduce a CPC in Google Ads, start from what sets the bid, since that decides what you can change:

• Under Manual CPC, lower the keyword’s max. CPC, as above. Google says a click won’t cost more than it.
• Under Maximize clicks, set a maximum CPC bid limit on the bid strategy. Google says it can help control costs where your CPCs are higher than you want, and that a limit set too low can keep the strategy from its goal.
• Under Smart Bidding, the keyword’s bid isn’t yours to lower. What you can change is what the keyword buys and how good the ad is.
• Under any of them, mend the part of Quality Score that Google rates below average, since higher quality ads typically cost less per click, and add negative keywords for the searches that cost the most and convert the least. The wasted search terms guide finds them.

A lower CPC isn’t the goal by itself. A lower bid can show your ad less often and lower on the page, so judge the change by what a conversion costs afterwards.

A keyword written +running +shoes is in broad match modifier, which Google has discontinued. Since July 2021 new keywords can’t be added in that notation, and the ones already in an account keep serving and behave as phrase match keywords, whatever the plus signs suggest.

Google says there are no performance benefits to converting them. If you want each keyword written as what it does:

1. 01On Search keywords, select the keywords you want to change.
2. 02Select Edit, then hover on Change match types.
3. 03Select Change broad match modifier keywords.

Converting creates a new keyword in the match type you chose and removes the old one, and Google says the old keyword’s statistics don’t carry over to the new.

Elsewhere, Google says plus signs added to a keyword or to a negative keyword will usually be ignored, unless one ends a word, as in C++. Quotation marks and brackets aren’t part of a keyword’s text either: typed into Google Ads, quotation marks make a phrase match keyword and brackets an exact match one, and brackets are among the symbols Google lists as invalid in a keyword. A keyword or a negative whose text itself holds them reads as a match type it may not have. Check its Match type column.

Every day Goldbeater reads each enabled keyword’s last 13 whole weeks and sets it against the other keywords of its campaign, brand keywords beside brand keywords and the rest beside the rest, a week at a time. It raises a keyword that spent $20 or more and bought under half the conversions its spend would have at their rate, where so few is under a 5% chance after allowing for every keyword it tested across the account. Where your conversion values are real, its ROAS has to be under half theirs too.

That is the keyword worked above: the finding’s 8.6 expected conversions, its $49 a conversion and its 2.5% chance are the table’s sums, done a week at a time. It adds what the table can’t: each week beside the other keywords’ return, and what the keyword spends a month now, from its last 4 weeks. "storage bed" was tested in the same pass and isn’t raised.

What Goldbeater drafts depends on what the keyword did and on who sets its bid:

• It converted nothing. Goldbeater drafts its pause, whatever the bid strategy, and under Manual CPC offers halving its bid beside it.
• It converts, under Manual CPC. Goldbeater drafts its max. CPC at half today’s bid, a figure you can lower further, and offers the pause beside it.
• It converts, under an automated strategy. There is no bid to lower, so the finding is advice, a step lower in severity: read the searches it buys first, with the pause offered and not recommended.

A finding reads medium, high where the keyword spends $200 a month or more above the others’ price, and low where it still does better than your account’s other campaigns typically do. Five more checks read your keywords:

• Small keywords sharing a word. Goldbeater adds up the keywords of a campaign that share a run of one to three words, leaving out any it raises alone. It raises the word where two or more keywords together spent $20 or more over the 13 weeks and converted nothing while the campaign’s other keywords converted, by the same test, allowing for every word it tested. The finding reads low. It offers pausing them without recommending it, and under Manual CPC advises lowering their bids first.
• A keyword converting worse than a month ago. Its last 4 whole weeks bought under half the conversions its spend would have at its rate in the 4 before, after moving that rate as its campaign’s other keywords’ rate moved, and under half the return too where your values are real. It needs $20 or more in each window and conversions in the earlier one. The finding is advice, medium from $200 in the last 4 weeks and low under it, and says whether its clicks got dearer or convert less often.
• A keyword’s clicks getting dearer. A keyword that spent $20 or more in its last 4 weeks, whose cost per click rose against its own earlier weeks and against its campaign’s other keywords in the same weeks, each by more than its weeks’ own swings explain. It has to have paid $20 or more over the dearer of its earlier price and that price moved as the others’ did. Goldbeater leaves it where its cost per conversion held, and in a brand campaign whose clicks got dearer as a whole. The finding is advice and reads low.
• A keyword’s conversions worth less. Where your conversion values are real, a keyword with five or more conversions in its last 4 weeks and five or more before, whose value per conversion fell against its own weeks and against its campaign’s other keywords, by $20 or more in all. The finding is advice and reads low.
• Text Google reads differently. An enabled broad keyword written with a plus before a word, a negative keyword with one, and quotation marks or brackets in a keyword’s or a negative’s text. For a broad match modifier keyword Goldbeater drafts the phrase keyword and the pause of the old one. For a campaign’s own negative it drafts the same words without the plus signs, and for an ad group’s or a list’s it says where to edit it. A rewrite in the match type the marks stand for is offered, since what they meant is yours to say. The finding reads low, and medium for a keyword that carries your spend.

Each of the four that compare weeks allows for everything it tested, as the first does, so a longer keyword list doesn’t raise more by luck. For the rest of an account, the audit checklist runs through every area in order.

How do I optimize keywords in Google Ads?

Read them on the Search keywords table with Conversions, Cost / conv., Conv. value / cost and Avg. CPC added, over about 13 weeks. Compare each keyword with the other keywords of its campaign: one that bought under half the conversions its cost would have at their price, where luck doesn’t explain so few, is the one to pause, bid down or narrow. Leave the ones that are only noisy.

When should I pause a keyword?

When it has spent several conversions’ worth at the price the rest of its campaign pays and bought none, and its searches are the right ones. Read its search terms first: if it buys searches for something you don’t sell, a negative keyword fixes it without losing the keyword. Google says a paused keyword won’t show ads or accumulate new costs until it’s restarted.

Why is my CPC so high in Google Ads?

A click’s price is set in the auction: Google says you pay what’s minimally required to clear the Ad Rank thresholds and beat the competitor immediately below you. It rises with more competition, lower ad quality, a place above the search results, a higher bid, or a bid strategy bidding more for a search it expects to do well.

How do I lower my CPC in Google Ads?

Under Manual CPC, lower the keyword’s max. CPC bid. Under Maximize clicks, set a maximum CPC bid limit. Under Smart Bidding the keyword’s bid isn’t yours to lower, so improve the part of Quality Score that Google rates below average and add negative keywords for costly searches that don’t convert. Judge the result by what a conversion costs, not by the click.

Can I change a keyword’s bid under Smart Bidding?

No. Smart Bidding sets a bid for every query at the time of the auction. Google’s page on editing keywords says that under automated bidding the Max. CPC column has no pencil button, and a bid marked Auto can’t be edited.

Does changing a keyword’s match type reset its statistics?

It moves them out of sight. Google says that when you edit a keyword or change its match type, you’re removing the original keyword and creating a new one. The original’s statistics stay with it, and show once you adjust the filters to include removed keywords.

Do broad match modifier keywords still work?

They still serve, as phrase match. Google discontinued broad match modifier: since July 2021 new keywords can’t be added with the plus-sign notation, and the old ones behave as phrase match keywords. Google says there are no performance benefits to converting them.