AI Overviews Came for Your Commercial Keywords

For two years the industry told itself a comforting story about AI Overviews: they take the informational traffic, the top-of-funnel explainers, the definitions nobody was going to buy from anyway. The money keywords were safe. That story is now out of date. AI Overviews on commercial keywords grew 71% in six months, according to Semrush research published on 2 July 2026, and the growth is concentrated in exactly the categories where a click is worth the most.

Commercial keywords are the ones your revenue actually runs on. Not “what is X,” but “best X for Y,” “X vs Z,” “X pricing,” “top X companies.” Someone in research mode, comparing options, close enough to a decision that a competitor’s name in the answer costs you the deal.

Those are the SERPs Google is now putting an AI Overview on top of. The pages that serve them (service pages, category pages, comparison pages, pricing pages) were mostly never written to be quoted by a machine. That is the gap this piece is about.

Do AI Overviews show up on commercial searches now?

Yes, and that is the change. Semrush analysed more than 600,000 keywords from its US desktop database between November 2025 and April 2026 and found the share of commercial-intent SERPs carrying an AI Overview grew 71% over those six months. Every one of the ten industries in the study saw growth.

The mirror-image finding is the one worth sitting with. Over the same window, the share of transactional SERPs showing an AI Overview fell 5%. Google is not pushing the Overview all the way down to the point of purchase. It is planting it in the research stage, the part of the journey where a buyer is still deciding who to trust.

Semrush’s reading of why is straightforward: the longer the research phase in a category, the more room there is for an AI answer to compress it. Internet & telecom, Finance, and Travel, all decisions that take weeks and involve real money, showed the widest gap between commercial and transactional Overview coverage. In Food & drink, where people research and choose lunch inside an hour, the pattern flipped: Overviews on 16% of transactional SERPs and 14% of commercial ones.

Exposure is not evenly distributed. If you sell something people deliberate over, you are in the blast radius.

Which commercial queries trigger AI Overviews most?

The expensive ones. When Semrush split its keyword set into four CPC tiers, AI Overviews appeared most often in the most expensive tier in nearly every industry. Across the priciest Jobs & education keywords, more than 60% carried an Overview, a higher share than any other industry.

The CPC comparison makes the same point from the other side. Jobs & education keywords with an AI Overview averaged $5.02 CPC against $1.51 for keywords without one. Finance showed $4.84 against $2.14. In high-CPC categories, the presence of an Overview and the price of a click move together.

Two things follow. First, the queries where you already pay the most for a click are the queries most likely to have an AI answer sitting above your ad. Second, Google is not treating this as a conflict. Semrush found that Google Ads and AI Overviews now appear together on the same SERP roughly twice as often as a year earlier, building on its earlier finding that SERPs carrying both grew more than 394% through 2025.

Google has been explicit about the direction. On Alphabet’s Q2 2026 earnings call, reported by Search Engine Land on 23 July 2026, Chief Business Officer Philipp Schindler said AI Max is opening “billions” of previously unmonetized searches to ads, and that Gemini has improved Shopping ad relevance for complex queries by roughly 20%. The company is testing an AI Mode ad format called Highlighted Answers that places labelled sponsored links inside AI-generated list responses.

Read that alongside the Semrush data and the strategy is legible. The commercial research stage is being converted into inventory: organic citation slots and paid slots on the same page.

How much does an AI Overview actually cost you on a money keyword?

Less than the headlines suggest, and it depends almost entirely on whether you are cited. That is the most useful correction in the current data, and it comes from the most recent large study available.

Seer Interactive tracked 53 brands across paid and organic search from January 2025 through February 2026, covering 5.47 million queries, 2.43 billion organic impressions and 296.9 million paid impressions, and published the results on 24 April 2026. Where a brand was not cited in the Overview, organic CTR fell 67% across 2025 on informational queries, which Seer translates to roughly 13,000 fewer organic clicks per million impressions. On transactional queries, not-cited organic CTR nearly halved, from 4.17% to 2.15%.

Where the brand was cited, CTR held up comparatively well for most of the year.

Then the trend broke. After eighteen months of decline, Seer found organic CTR on AIO-present queries rebounded 85% across January and February 2026, closing much of the gap with non-AIO queries. Their own model had predicted continued decline. It did not happen. Seer is careful to call this directional rather than a benchmark, and so are we. But planning on the assumption that AI Overview traffic only ever goes down is no longer supported by the numbers.

Two caveats keep this honest. Pew Research Center, analysing the browsing behaviour of 900 US adults in March 2025, found users clicked a link 8% of the time when an AI summary was present against 15% when it was not, and clicked a source cited inside the summary just 1% of the time. Fewer clicks is real.

And Seer’s own revision on commercial queries is blunt: they had previously assumed commercial terms were relatively safe from AI Overviews, and they no longer do. Their 2026 note adds that commercial no-AIO organic CTR declined meaningfully across 2025 without significant Overview exposure, meaning something beyond AI Overviews is compressing commercial click-through, and it has not been fully explained. Diagnose before you conclude. We covered how to separate these effects in a walkthrough of stable rankings with falling traffic.

How do you rank in AI Overviews for commercial keywords?

You stop trying to rank and start trying to be quoted. Seer put the shift in one line in their 2026 update: the goal is not to answer the question, it is to be the source Google points to when it answers the question. Appearing on an AI Overview SERP without a citation is the worst position on the page. You absorb the CTR decline and get none of the attribution.

For commercial queries specifically, that means four things.

Make a claim the Overview can lift. Commercial queries are comparative. “Best X for small teams” needs a sentence that actually resolves the comparison, with the qualifier attached: who it is best for, under what constraint, at what price band. Marketing copy that says you are the leading provider gives a model nothing to extract. A sentence that says a 12-person team on a $400 monthly budget should choose the mid tier gives it something to quote.

Put the answer where it can be found. The first 40 to 60 words under a heading do disproportionate work. Lead with the resolved answer, then support it. Burying the conclusion under three paragraphs of context is a structural choice that costs you citations.

Show the numbers. Prices, timelines, capacities, specifications, service areas. Commercial buyers filter on constraints, and models extract constraints far more reliably than they extract adjectives. A pricing table on the page is more citable than a “contact us for pricing” button, and it also filters your leads.

Be consistent off-site. Overviews synthesise across sources. If your own page, your directory listings, and third-party roundups disagree about what you offer or what it costs, a model has a reason to hedge or skip you. This is the same discipline behind earning visibility when the click never happens.

How do you optimize service, category and comparison pages for AI Overviews?

Treat each page type as answering a different commercial question, then build the structure that answers it directly.

Three habits break commercial pages for AI extraction, and all three are common. The first is putting the substance in an image. A pricing chart rendered as a graphic is invisible to text extraction. The second is one-sided comparison pages: a page that finds your product superior on all fourteen criteria reads as marketing to a model as clearly as it does to a buyer. The third is thin category pages, a headline and a filter bar and a product grid, with no prose a model can quote.

The trade-off is real: making a page more extractable can mean fewer clicks, because the answer is available in the Overview. Being cited and not clicked still beats being neither, and the citation carries brand presence into the moment of comparison. That is a different measurement problem, and the subject of where buyers now start their research.

How do you track AI Overviews on your commercial keywords?

Segment your keyword set by intent first, then check Overview presence within the commercial segment only. Blended reporting hides this entirely: an informational decline and a commercial decline average into a number that tells you nothing about which pages to fix.

A workable monthly routine:

1. Split your tracked keywords by intent. Commercial and transactional separately, not lumped as “non-brand.”
2. Flag Overview presence on the commercial set. In Semrush this is the Organic Rankings report, Positions tab, filtered to Intent: Commercial and SERP Features: AI Overview.
3. Record citation status, not just presence. Whether an Overview appears matters far less than whether you are inside it. Track cited versus not cited as separate cohorts. That is the split that drives the CTR difference in Seer’s data.
4. Watch impressions and clicks separately. Seer makes the point that a click drop with stable impressions is a different diagnosis than a click drop with falling impressions. Conflating them sends you after the wrong fix.
5. Re-run monthly, judge on trend. Overview composition varies between runs and Google is actively iterating. One snapshot is not a measurement.

One reporting gap is worth flagging to your paid team. Search Engine Land noted in its Q2 2026 coverage that Google Ads still does not show which paid queries triggered an AI Overview, so the paid side of this is a blind spot you have to fill from organic SERP data.

The instinct when a new SERP feature lands on your money keywords is to defend the ranking. That instinct is now aimed at the wrong target. The position under an Overview is worth a fraction of what it was; the slot inside it is worth more than a position ever was. Commercial pages were built for a page where ranking was the whole game, and most of them still read that way.

Rewriting them is unglamorous work — a pricing table here, a genuine “choose the other one if” paragraph there — and it is the part of an AI-search programme with the shortest line to revenue.