SEO & Content

Google AI Mode and B2B Lead Generation: Where the Click Went, and What to Build Now

By Oliver Grant· Chief Digital Officer·October 4, 2026·15 min read
The headline 'Where the click went' in white on a dark navy background, below faint search-result cards with one highlighted card and lines tracing a path from it

Most of the commentary on Google AI Mode is written for publishers, because publishers sell attention and AI Mode is plainly taking some of it. A B2B company that sells a service or a software contract is in a different position. It never wanted the click for its own sake; it wanted the click because the click was the only reliable way to get onto a buyer's shortlist and, eventually, into a conversation with sales. Google AI Mode changes how that shortlist gets made, and that is the problem worth solving, much more than the traffic chart in Search Console.

This piece sets out how we think about AI Mode for lead generation, what the reliable data actually says, and what we would build if we were starting from your site today, with three worked scenarios at the end.

What Google AI Mode is now, in numbers that can be checked

AI Mode is the fully conversational tab in Google Search. It rolled out to everyone in the US on 20 May 2025, and by October 2025 Google said it was available in more than 200 countries and territories. At I/O in May 2026, Google reported that AI Mode had passed one billion monthly users, with queries more than doubling every quarter since launch. Sundar Pichai repeated the billion figure on the Q2 2026 earnings call and added that AI Mode is "driving an incremental increase in Search queries overall."

Two behavioural details from Google's own US data matter more for B2B than the headline number. The average AI Mode query is about three times the length of a traditional search, and planning-type queries are growing faster than the rest. A buyer typing a three-line question about migrating a claims system, with their constraints spelled out, is doing in one query what used to take a week of reading vendor blogs.

The click data is consistent in direction even where the studies differ in method. Pew Research Center tracked the real browsing of 900 US adults and found that people clicked a traditional result in 8% of visits when an AI summary appeared, against 15% when it did not. Semrush's study of 69 million AI Mode sessions found that 92–94% ended without a visit to any external site. That dataset was US desktop traffic in the first weeks after launch, so treat it as a direction rather than a constant, but nothing published since points the other way.

How often searchers click through when Google shows an AI summary
Clicked a result, no AI summary
15 %
Clicked a result, AI summary shown
8 %
Clicked a link inside the summary
1 %
Share of Google search visits that led to a click. Source: Pew Research Center, browsing data from 900 US adults, March 2025.

The click didn't disappear. It moved later.

Here is the reading we think most B2B teams are missing. The informational click, the one that brought someone to your "what is revenue-based financing" explainer, is the click AI Mode absorbs most completely, because the answer is the whole point of that query. The commercial click survives, but it arrives later in the journey, it often arrives as a branded search or a direct visit rather than an organic click on a generic term, and it comes from someone who already has a view.

The buyer research supports that picture. In G2's April 2026 survey of 1,076 software buyers, 51% said they now start research in an AI chatbot more often than in Google, up from 29% a year earlier, and 69% said a chatbot recommendation had led them to a different vendor than the one they planned to buy from. 6sense's 2025 buyer study of more than 4,000 buyers found that 94% of buying groups had ranked their preferred vendors before first contact, and that they bought from that early favourite 77% of the time.

Put those findings next to each other and the job changes shape. Previously your site earned the first visit by ranking for an informational query and then had weeks to educate the buyer. Now the education mostly happens inside the answer, the shortlist is drafted there, and your first visit comes from someone checking whether you are as good as the answer suggested.

AI Mode moves the B2B click later, and the click that survives carries more intent
THE TEN-BLUE-LINKS JOURNEYInformational query“what is…”Click to explainerfirst visitReads on your siteyou teach themNurture sign-upweeks of emailLeadlow intent, earlyTHE AI MODE JOURNEYConversational queryfollow-ups, contextAI answernames a shortlistBranded searchor a direct visitDecision pageproof, scope, toolsLeadhigh intent, latecyan = where your site first gets a visit
The same buyer and the same need, before and after AI Mode. Cyan marks where your site first gets a visit: in the AI Mode journey it comes later, and the visitor arrives with a shortlist already formed.

That is why "AI Mode is killing our traffic" and "our leads are holding up" are both true for many of the B2B sites we look at. Sessions measured the old journey and leads measure the new one, and the two have come apart.

Three jobs your site now has to do

We find it useful to stop thinking about "SEO for AI Mode" as a single discipline and split it into three jobs, each with a different owner and a different measure of success.

Job one: be named in the answer

The answer is the new first page, and the brands it cites get a measurable halo. Seer Interactive's analysis of 3,119 queries across 42 organisations found that organic click-through on queries with an AI Overview fell by 61%, but brands cited inside the Overview earned 35% higher organic CTR and 91% higher paid CTR than brands that were not cited. The penalty for being absent is larger than the penalty for being summarised.

Being named comes down to three things a model can work with. First, your pages need to state facts plainly enough to be extracted: who you serve, what you do, which systems you integrate with, what a typical engagement looks like, and what it costs or how it is scoped. Second, the claims need corroboration outside your own domain, because answer engines weigh third-party mentions, review platforms and trade press heavily when building a shortlist. Third, the technical basics have to be sound, which we covered in detail in what controls your citations in AI search, and how to measure them.

The part most teams skip is the second one. A beautifully structured site with no independent footprint reads, to a model, as a company with only its own word to go on.

Job two: convert the visitor who already has the summary

When the visitor arrives late with a shortlist, a page that opens by explaining the category wastes their time. They have read the explainer, in AI Mode, thirty seconds ago. What they need is the material the answer could not give them with confidence: evidence, specifics and a way to test fit.

That means building what we call decision pages. These include a scoping page that explains what drives cost and gives honest ranges or a calculator, a comparison page that says plainly where you are and are not the right choice, a proof page with named outcomes and the method behind them, and an integration or implementation page that answers the "will this work with what we have" question in detail. Each of these is also exactly the kind of page an answer engine likes to cite, because it contains specific, checkable statements, so the work on job two feeds job one.

The form itself deserves attention too. A late-stage buyer tolerates a short, purposeful form far better than a gated ebook, and a well-built AI lead scoring layer can then tell sales within seconds whether the enquiry is a shortlisted buyer or a student writing an assignment.

Job three: measure the journey you actually have

Google's own help documentation confirms that clicks from AI Mode count as clicks in Search Console, with standard impression rules, but they are blended into your normal Web totals with no separate filter. You cannot isolate AI Mode performance from Search Console alone, so the measurement has to come from the edges of the journey.

The signals we track are branded search impressions and clicks over time, direct and branded-organic conversions, referral sessions from assistants such as ChatGPT and Perplexity, a fixed panel of buyer-style prompts checked weekly to see whether you are named and how you are described, and a "how did you hear about us" field with an AI assistant option on every enquiry form. The last one is low-tech and it is often the most honest number in the stack. Those signals then need to be joined to pipeline rather than reported side by side, which is the job our InsightAX reporting layer does for clients, so a fall in blog sessions can be read against what happened to qualified leads and revenue in the same period.

Be careful with the conversion multiples circulating on LinkedIn. Ahrefs reported that AI search drove 0.5% of its traffic but 12.1% of its signups, and Semrush has claimed AI visitors are worth 4.4 times an organic visitor. Both figures come from ChatGPT-style referrals rather than AI Mode, the first is one company's own site and the second does not publish its method. They are encouraging, and they fit our journey model, but you should test the claim on your own data before you put it in a board deck.

Ads are already eligible to appear in AI Mode, and Google announced the expansion at Google Marketing Live in May 2025. Until recently the route in was broad match with automated bidding, AI Max for Search or Performance Max. In September 2026 Google's Ads Liaison confirmed a small test in which exact and phrase match keywords can serve text ads in AI Mode for queries with explicit, direct intent. It is a test rather than a launch, and Google Ads still does not separate AI Mode delivery from standard Search in keyword reporting.

Our view for B2B accounts is to treat AI Mode reach as something you opt into deliberately, on campaigns where the conversion signal is strong enough to steer automated bidding, which in practice means feeding qualified-lead or opportunity data back from the CRM rather than raw form fills. A campaign optimising to form submissions will happily buy conversational traffic that never becomes pipeline. We go into the setup on our PPC page.

Three scenarios, and what we would do in each

These are composites of situations we see repeatedly, not individual client case studies, and the numbers in them are illustrative.

Scenario one: the managed IT provider whose blog traffic fell by 40%

A managed services firm with a strong content library sees organic sessions to its blog down about 40% year on year, and the marketing lead is under pressure to explain it. Enquiries are roughly flat, and branded search impressions are up.

What is happening is the journey shift described above. The explainer posts that used to bring in "what is SOC 2" readers are now summarised in the answer, while buyers who see the firm named in that answer search for it by name later. The wrong response is to publish more explainers to win the traffic back. The right one is to change what the blog is measured on, from sessions to citation presence on a prompt panel and assisted conversions, and then to take the ten highest-value explainers and rebuild each as a citable reference with clear definitions, a dated methodology and a link to a matching decision page. Effort then shifts to two or three decision pages per core service, starting with the scoping page, because that is the question buyers most often take to an assistant and leave without a confident answer.

Scenario two: the compliance software vendor that is never on the shortlist

A fintech compliance vendor runs its own buyer-style prompts through AI Mode and two other assistants and finds that it is never named. Three competitors appear every time, and one of them is smaller.

The diagnosis is usually about evidence rather than on-page SEO. The competitors have reviews on the platforms buyers use, analyst or trade-press mentions, partner listings with the core banking platforms, and a public comparison page, while our vendor has a well-designed site that describes itself in aspirational language without stating plainly which regulations it covers or which systems it connects to. The plan is to rewrite the core service and product pages so that every capability is stated as a specific, checkable claim, to publish an honest comparison page and an integrations directory, to run a structured review programme with existing customers, and to pursue two or three pieces of genuine trade-press coverage. Then the prompt panel is re-run monthly, because that is the only instrument that shows whether the shortlist is moving.

Scenario three: the logistics operator whose buyers ask Google to call around

A regional freight and warehousing operator sells to mid-sized e-commerce brands. Google has been rolling out the ability for AI Mode to call businesses for pricing and availability on a searcher's behalf, and Google says businesses control participation through their Business Profile settings.

The opportunity is to be the business that answers well when that call or query comes in. That means a Business Profile with current service categories and hours, a public page that states minimum volumes, lead times and the main pricing drivers, and a phone and enquiry flow that can respond in minutes, which for most operators means an AI lead qualification agent handling the first exchange and routing serious enquiries to a person. A buyer who gets a clear answer from one operator and a voicemail from three others has made their choice before anyone in sales knows the enquiry existed.

A 90-day plan if you are starting now

In the first month, set up measurement before touching content: define the prompt panel of 20 to 40 buyer questions, add the "how did you hear about us" field, segment branded and non-branded search in reporting, and connect lead quality to source in the CRM. In the second month, rebuild the core service pages as fact-dense, extractable pages and ship the first two decision pages, starting with scoping and comparison. In the third month, start building the external evidence base through reviews, partner listings and trade press, refresh the highest-value explainers into citable references, and review paid campaigns so that any AI Mode reach is optimising to qualified leads rather than form fills.

None of this abandons classic SEO. Crawlability, speed and structured data remain the price of entry, which is why we still treat them as the foundation of our SEO work. What changes is the goal. In AI Mode, the job of a B2B site is to make the answer recommend you and then to give the buyer who checks a reason to get in touch.

Frequently asked questions

Related services

How we work in this space

Continue reading

More articles

Sources & references

← All postsGet a proposal →