If you run paid search for a US business, you have probably had the same conversation I’ve been having with clients since the start of this year. Spend is steady. Quality Score hasn’t moved. The keywords are the same ones that worked last year. But clicks are down, and nobody can point to a single change in the account that explains it.
Most of the time, the account isn’t the problem. The search results page is.
What changed at the top of the SERP
Google’s AI Overviews now sit above the traditional results on a meaningful share of queries. They appear on roughly 15% of searches and pull their answers from the pages that used to earn those clicks, which has cut average organic click-through rates by about 18%.
The paid side takes a harder hit than most people expect. One analysis found that when an AI Overview is present, Google Ads click-through rate falls from 19.70% to 6.34%. That is not a rounding error. On a campaign built around a handful of high-volume head terms, an Overview showing up on those terms can quietly cost you two-thirds of your traffic while every in-platform metric looks normal.
The mechanic is simple. The user gets a synthesized answer before they ever reach your ad. If the answer is good enough, the search ends there.
This is not the end of paid search
I want to be careful here, because there is a lot of panic content on this topic and most of it is overstated.
Google has not removed ads from the equation. It has moved them. Ad eligibility has been expanded inside AI Overviews and AI Mode through Performance Max, Shopping, broad match Search campaigns, and AI Max for Search. The inventory still exists. What changed is how you qualify for it.
The old way to win was tight keyword control: exact match, narrow themes, manual negatives, a human steering every lever. That approach is still useful for protecting budget, but it is increasingly the wrong tool for getting into AI-generated results. Campaign types like Performance Max, Shopping, and broad match Search matter more now because they give Google the signals it needs to match your ads to longer, messier, conversational search journeys.
Which means the lever has moved from keyword selection to data quality. The accounts doing well in 2026 are the ones feeding the system better inputs.
What “better inputs” actually means
I’ll be specific, because this phrase gets used loosely.
1. Fix your conversion tracking before you touch anything else.
Automated bidding is only as smart as the conversion signal you give it. If you are still counting every form fill as one conversion when half of them are junk leads, the model is optimizing toward junk. The shift in 2026 is toward conversion quality optimization and server-side tracking instead of chasing raw lead volume. Offline conversion imports, value-based bidding, and lead scoring fed back into the platform are where the real gains sit.
2. Treat first-party data as an asset, not a compliance headache.
Customer match lists, email segments, past purchasers, high-value lookalikes. First-party data has become one of the sharpest competitive advantages in paid search, in many cases mattering more than how much budget you have. A smaller advertiser with clean, well-segmented customer data can out-target a bigger one who is running on platform defaults. 6smarketers
3. Expect smart bidding to fail sometimes, and know what failure looks like.
Automation is not a set-and-forget system. Conversion volume drops, seasonality shifts, a tracking break upstream, and a Target ROAS strategy can spiral without any obvious alarm. Build a weekly check on conversion volume and bid strategy status so you catch it in week one instead of week six.
4. Give the algorithm creative to work with.
Text alone is thin input now. Search advertising should no longer be treated as a text-only channel, and creative assets need to support the campaign rather than sit beside it. Images, video assets, and strong product feeds all feed the same matching system.
5. Stop measuring Google in isolation.
Privacy-first attribution and multichannel expansion are shaping how performance is measured in 2026, and last-click reporting will increasingly tell you a story that isn’t true. Even a basic blended CAC view across channels beats a platform-reported ROAS number you can’t reconcile with revenue.
The channels worth testing right now
Two are worth naming for US advertisers specifically.
Microsoft Advertising benefits from Copilot’s share of AI-assisted search, and CPCs there are still below Google in most B2B categories.
Reddit is the one more people should be looking at. Reddit has grown into a genuine performance channel, with automated targeting and creative selection through Reddit Max, dynamic product ads, and commerce integrations, and performance advertising now makes up the majority of its ad revenue. For considered purchases where buyers research heavily before converting, it’s reaching people at a stage Google misses.
Where this leaves you
The honest summary is that paid search got harder to do casually and more rewarding to do properly. The advertisers losing ground are mostly the ones running the same account structure they built in 2022 and wondering why the numbers drifted. The ones gaining are not necessarily spending more. They have cleaner conversion data, segmented first-party audiences, and creative that gives the platform something to optimize.
None of that is exotic work. It’s mostly unglamorous account hygiene done consistently, which is usually what separates a good account from an average one.
If your CTR has dropped this year and you aren’t sure whether it’s the SERP, the tracking, or the structure, that’s a diagnosable question. Start by pulling your top 20 terms by spend and checking which of them now trigger an AI Overview. That one check tells you a lot.
MashGrowth builds and manages performance campaigns for US businesses, with a focus on conversion tracking, first-party data, and paid search that holds up in AI-driven results. Book a call