Google Ads CPCs are up 10 to 25 percent across nearly every industry this year, and the usual reaction is to blame competition. Competition is part of it. The larger part is that advertisers handed match-type control to automation and stopped reading what they were buying.
Why CPCs rose, structurally
- AI Overviews pushed organic results down the page, so more of the same demand now arrives through paid placements. Less free traffic means more paid competition for identical intent.
- Smart Bidding produced a collective escalation. When every advertiser's algorithm independently bids toward its maximum tolerable cost, the auction clears higher for everyone. Nobody made a bad decision; the aggregate result is still inflation.
- Lower barriers brought more advertisers in. AI campaign tooling let smaller businesses enter auctions they previously could not manage, thickening the bidding pool.
None of those are reversible by tactics. What is controllable is how much of your budget lands on queries that were never going to convert.
The number worth acting on
Analysis across 200-plus accounts found 15 to 40 percent of spend going to irrelevant broad match queries that never convert. On a $20,000 monthly budget that is $3,000 to $8,000 a month buying traffic with no commercial relationship to what you sell.
Automation did not stop needing supervision. It stopped announcing when it needed supervision.
AI Max extends this further by combining broader matching, AI-generated text variations and final URL expansion. Google reports roughly a 7 percent average lift in conversions at similar efficiency, which is a genuine result. It is also an average, and the accounts dragging it down are the ones where nobody constrained the expansion.
What actually goes wrong
- Query drift toward the adjacent and useless. Broad match with AI expansion reaches semantically related terms, and semantic relatedness is not commercial intent. "How to do X yourself" is adjacent to "buy X service" and worth nothing to you.
- Final URL expansion routing clicks to the wrong page. A visitor with buying intent landing on a blog post converts at a fraction of the rate they would on the page built to sell.
- Top-of-funnel traffic diluting your conversion rate, which then misleads every optimisation decision you make afterwards.
- PMax cannibalising brand search and reporting it as performance. Cited by roughly a third of practitioners as a live concern, and covered in detail in PMax best practices. The fix is brand exclusions, and it is still frequently skipped.
The control layer to put back
- Read the search terms report weekly, not monthly. Automation changes what it buys faster than a monthly cadence can catch.
- Build negative lists as a standing asset, shared across campaigns. Most accounts treat negatives as occasional housekeeping rather than infrastructure.
- Turn off final URL expansion unless you have specifically verified that the pages it chooses convert. It is opt-out, and the default rarely suits a considered purchase.
- Keep exact match campaigns for proven money terms, budgeted separately, so your reliable demand is not competing with an experiment.
- Feed the algorithm better data rather than fighting it. Clean conversion values and accurate signals change what automation optimises toward, which is more effective than manual bid resistance. That argument runs through how AI changed this job.
The honest framing
This is not an argument against automation, which wins on speed and scale and is not going away. It is an argument that the work moved rather than disappeared. The job is now setting constraints, supplying clean data, and auditing what the machine bought, and the accounts skipping that step are funding the CPC inflation everyone else complains about.
If your costs climbed this year and nothing else obviously changed, pull the search terms report before raising budget. The answer is usually sitting in it. For the full sequence, the audit checklist runs measurement first, then structure, then keywords, in that order for a reason.
Frequently asked questions
Why are Google Ads CPCs rising in 2026?
CPCs are up 10 to 25 percent across most industries for structural reasons: AI Overviews push organic listings down so more demand arrives through paid placements, widespread Smart Bidding adoption creates collective bid escalation as every algorithm bids toward its maximum tolerable cost, and AI tooling has lowered the barrier for smaller advertisers entering the same auctions.
How much budget does broad match waste?
Analysis across more than 200 accounts found 15 to 40 percent of spend going to irrelevant broad match queries that never convert. The waste concentrates in accounts where nobody reviews the search terms report regularly, because AI-driven query expansion changes what it buys faster than a monthly review catches.
Should I use AI Max?
It can work, with Google reporting around a 7 percent average lift in conversions at similar efficiency, but it needs constraints. Turn off final URL expansion unless you have verified the pages it selects convert, keep exact match campaigns for proven terms, and review search terms weekly. Unconstrained, it reaches semantically related queries that carry no commercial intent.
Does Performance Max steal traffic from brand campaigns?
It can, and roughly a third of practitioners cite cannibalisation as a concern. Without brand exclusions, PMax serves against your own brand searches, converts demand you already had, and reports it as campaign performance, which inflates apparent ROAS while raising what you pay for traffic that was already yours.