First, confirm the rise is real
An account-level average hides mix. If a cheap segment shrank and an expensive one grew, the average rises even though nothing became more expensive. Before blaming the auction, break cost per click down by campaign, keyword, device, location and hour, and compare each segment with itself over the same weekdays in the earlier period. If most segments are flat and the mix moved, the cause is mix. If the same keyword on the same device now costs more, the price really rose.
Put cost per acquisition next to cost per click in every view. A price increase that comes with a better conversion rate may leave CPACost of each conversion. unchanged, which makes the rise a reporting curiosity rather than a fault. If the real question is that volume fell rather than that price rose, the impressions drop diagnosis is the page to follow.
Diagnosis table
Diagnosis
| Symptom | Check | Likely cause | Fix |
|---|---|---|---|
| Same keywords cost more; outranking rate of others rising | Auction insights over time; top-of-page rate | More or stronger competitors in the auction | Defend only profitable terms; improve quality rather than bidding blindly |
| Price jumps after a settings change | Change history for bid strategy, target or impression share goal | Strategy changed, target loosened, or a target impression share goal set | Restore or tighten the target; reserve impression share goals for terms that justify them |
| Rising price with falling component ratings | Quality Score and its three components over time | Weaker relevance, expected CTR or landing page experience | Repair the weak component on high-spend keywords |
| New, dearer queries appearing in search terms | Search terms report sorted by cost; match type of the triggering keyword | Broad match drifting into pricier, adjacent queries | Add negatives, tighten keywords, or separate the pricier intent |
| Account average rises while segments are flat | Device, location and hour reports | Traffic mix moved toward costlier segments | Set adjustments or split campaigns if the costly segments do not pay |
| Rise lines up with a known peak | Prior-year cost per click by week | Seasonal demand and competition | Plan budgets and targets for the peak rather than fighting it |
Fix, verify and prevent each cause
Competition intensified
The help centre explains that the price of a click is set by the auction: what you pay depends on the Ad Rank needed to hold your position against the advertisers below you. When a new competitor enters, or an existing one raises bids or improves quality, the price of holding the same place rises. Auction insights shows overlap, position-above rate and outranking share by domain, and the tutorial on reading auction insights works through those measures in order and turns them into a decision to defend, concede or improve quality.
Fix: rank keywords by value and decide which positions are worth paying for. Defend those; let the others slide to a position that costs less. Improving ad and landing page quality lowers the price of a given position more durably than outbidding.
Verify: on the defended terms, conversions hold and acquisition cost stays within target.
Prevent: read auction insights monthly so a new entrant is noticed before it shows up in the price. The Quality Score guide owns the work on quality components.
A bid strategy or target change
Switching to a different strategy, loosening a target CPA or ROAS, or adopting a target impression share goal can all lift bids by design. Target impression share in particular asks the system to show at a given position regardless of what each click returns, which is useful for brand defence and expensive everywhere else.
Fix: find the change in change history and check whether it was meant to buy more volume at a higher price. If not, move the target back gradually or return to the earlier strategy.
Verify: the bid strategy report shows average bids settling, and acquisition cost returns to range.
Prevent: tie every target to a documented acquisition cost or return, using the logic in the Smart Bidding guide.
Quality slipping
When Quality ScoreGoogle's 1–10 keyword diagnostic. components drop, holding the same position requires a higher bid, because the auction weighs quality alongside the bid. A landing page redesign, a slower site, or ads rewritten without the query's language are common triggers. This page does not explain the components; it only asks which one changed.
Fix: compare component ratings before and after the rise. Repair whichever moved: restore relevance, improve the landing page, or rewrite ads people are more likely to click.
Verify: ratings recover and the cost per click on those keywords eases without a bid change.
Prevent: check quality ratings after every landing page or ad release.
Broad match drifting into dearer queries
Broad match lets a keyword reach related searches, and some related searches sit in more competitive, more expensive corners of the market. The average climbs because the keyword now buys different clicks.
Fix: sort the search terms report by cost and read the matches that are new. Add negatives for the irrelevant ones, and move relevant but pricier intent into its own ad group with its own expectations. The match behaviour itself is covered in the match types guide.
Verify: the share of spend on newly matched, high-price queries falls in the following weeks.
Prevent: review search terms on a fixed rhythm, and more often after switching keywords to broad.
Traffic mix shifted
A campaign whose traffic moves toward a dearer device, a more competitive city or evening hours will show a higher average with no change in any unit price. This is the mix effect from the first section, now named as the cause.
Fix: decide whether the costlier segments convert well enough to justify their price. If they do, the rise is fine. If not, apply adjustments where the bid strategy allows them, or separate the segment so it can carry its own budget and target, as the account structure guide describes.
Verify: segment-level acquisition cost converges on the account target.
Prevent: report key segments separately, not only the account average.
Seasonal peaks
Demand peaks bring more searchers and more advertisers, and both push prices up. The budgets and pacing guide covers how to plan spend around them.
Fix: compare with the same weeks in a previous year. If the rise matches the seasonal pattern, plan for it: adjust budgets and accept a higher price where conversion rates rise too.
Verify: prices fall back after the peak without any intervention.
Prevent: keep a seasonal calendar and set expectations for each peak in advance.
Prevention: watching CPC alongside CPA
Cost per click on its own is a poor alarm, because it rises for good reasons as often as bad ones. A prevention routine pairs it with outcome metrics so a rise only triggers work when it matters. Report cost per click, conversion rate and acquisition cost together, for each major segment rather than only for the account. Set an alert on acquisition cost, and treat a click-price change as context for that alert. Check auction insights monthly for the campaigns that carry most of the spend. Read the search terms report by cost after any move toward broad match. Keep the seasonal calendar in the reporting template so a peak is labelled before anyone asks about it. The CPC calculator gives the maximum click price a target acquisition cost can carry at a given conversion rate, which turns "CPC is high" into a yes-or-no question.
A software firm whose average CPC climbed for a month
A project-management software firm sees average CPC rise from 3.10 to 4.05 over four weeks. Segmenting shows desktop CPC on its core exact-match keywords flat at around 3.20. The rise comes from two places: broad-match keywords now match "enterprise project portfolio software" queries at about 7.50 a click, and mobile's share of clicks fell from 45% to 30%, removing cheaper clicks from the mix. CPA has gone from 96 to 118. The firm adds negatives for enterprise portfolio terms, moves the relevant ones into a separate ad group with its own target, and leaves the device mix alone because mobile converts poorly. CPA returns to 101 within three weeks.
Common mistakes
Reading the account average instead of like-for-like segments. Capping bids before knowing why the price rose. Treating a seasonal peak as a fault. Setting a target impression share goal on generic keywords and being surprised by the bill.
