Search and AI Visibility: Twelve Months Since AI Moved Into the SERP

Six months ago, we wrote about Google's AI Overviews landing in the search engine results page (SERP) and what that meant for brand visibility. The tone back then was cautious. Curious. A bit wait-and-see.

How quickly things have changed.

The speed of change has been faster than most experts predicted, and the gap between businesses that are standing still and those that proactively adapt their content and websites to rank better in a world of AI is widening by the week. If you've noticed your organic traffic softening, your branded impressions flattening, or your content showing up in AI answers without the click, you're not imagining it.

The SERP you remember doesn't exist

Google's AI Overviews have gone from novelty feature to default behavior on nearly half of all queries. Pew Research, analyzing 68,000 queries, saw the click rate drop from 15% without AI Overviews to 8% with them. Other industry studies have put the hit on the top-ranking result as high as 58-61% when an Overview appears.

Zero-click is the new default. Searches that trigger an AI Overview now average an 83% no-click rate. Even traditional queries hover around 60%.

So what's actually changed since October?

The behaviors driving the new SERP

• Answers replace listings - users get a synthesized response at the top of the page and, in most cases, stop reading. The ten blue links are still there. They're just not where attention lands.
• Citations carry more weight than rankings - brands cited inside the AI Overview earn more clicks than brands in the blue-link results below it. Position one of page one now sits under a layer of synthesized authority.
• Brand queries behave differently than topic queries - topic queries get the full Overview treatment. Branded terms still behave more like the old SERP, which is why direct demand is holding up better than top-of-funnel traffic for most B2B sites.
• Prevalence has doubled - roughly 48% of queries triggered an Overview in March 2026, up 58% year-on-year. Six months ago it was a feature. Now it's the default interface.

The short version: you're not losing clicks because your SEO broke. You're losing clicks because the page changed.

Buyers moved faster than the agency playbook

The traffic story is only half of it. The bigger shift is that B2B buyers have quietly moved the top of their research funnel into chatbots.

G2's March 2026 research found 51% of B2B software buyers now start their purchase research in an AI chatbot rather than Google. MarTech's coverage of the same shift puts AI tool usage across the full research process at 73%. The buyer's first impression of your brand, in many cases, is now a paragraph generated by ChatGPT, Perplexity, or Gemini, summarizing what the internet appears to think about you.

Three follow-on effects worth tracking:

What the AI-mediated funnel is doing to pipelines

• Shortlists form before anyone visits your site - B2B sales research shows 95% of winning vendors are on the buyer's Day One shortlist, and the top-ranked vendor wins 77% of the time. If the assistant doesn't name you on day one, you're usually not in the room later.
• Dark funnel activity is becoming the largest phase of the journey - buyers research, rank, and narrow before any tracked engagement fires. Your attribution software can't see what ChatGPT told them.
• Buying cycles are compressing - the average B2B cycle shortened from 11.3 to 10.1 months year-over-year. AI-mediated research isn't just changing where buyers look. It's changing how quickly they decide. This shortening of buying cycles could suggest an acceleration of discovery-to-action in B2B pipelines; however, the alternative theory is that we now have 1.2 months less visibility and awareness of what people are researching. This means when you do get wind of a new lead, they're likely already 4-5 weeks deeper into their own due diligence than you might realize, and you have a lot of catching up to do to inform their own assessments, as they'll have made their way into other sales pipelines by that point also.

The 1.2-month compression warrants a closer look because, operationally, it's worse news than it sounds.

Most of that compressed time happens before your marketing automation platform has a cookie on anyone. A prospect who arrives on your site with a demo request used to be somewhere in the middle of their evaluation. Now, based on the same research, they're closer to the end. They've asked ChatGPT to summarize what features and functionality they require, had Perplexity compare three vendors, pulled reviews off G2, skimmed LinkedIn posts, and quietly formed a shortlist. You're either on it, or you're not. By the time a form is completed the due diligence is already done.

That shift changes three things about how programs run.

• Lead scoring under-predicts intent, because models tuned to 2023-era journeys treat a first content download as early-stage when, in an AI-mediated journey, it's frequently late-stage.
• Response times have to tighten up even more. A 48-hour reply on a new lead felt fine two years ago. Now those two days are spent letting the prospect deepen their diligence with the vendors already on their shortlist, and let's be honest, your competitors already have automated, personalized SDR outreach in place within minutes or hours of the inquiry.
• Thirdly, first-touch attribution stops being useful altogether. The first real touch was an assistant answer you can't see and a LinkedIn citation you may or may not have written. Attribution has to move from "which channel sourced this" to "where was our product or brand visible during the lead's research". When you can't categorically identify first-touch attribution, you should at the very least track if your coverage is improving to increase your non-cookied influence.

The practical response isn't to panic. It's to assume every new inbound lead is further along than the scoring says, route them faster, brief sales with what the buyer most likely already knows, and audit where your brand shows up during the hidden phase of research.

The uncomfortable part: only 22% of marketers are actively tracking AI visibility, according to Similarweb's gen-AI stats report. The buyer has moved. Most programs haven't.

Search visibility and AI visibility are one problem now

Terminology is starting to get in the way. Answer engine optimization (AEO), generative engine optimization (GEO), and SEO are being sold as separate practices. For B2B teams with finite time, it helps to collapse the different modalities of how people research online and think about a single outcome: Search/AI visibility, the combined presence of your brand across ranked results, synthesized answers, and assistant-generated shortlists.

The techniques underneath are more continuous than the acronyms suggest.

What actually moves the needle across both surfaces

• Schema markup and structured data - clean schema paired with FAQ blocks has been tied to a 44% lift in AI citations in industry testing. Google's Gemini-powered AI Mode uses schema to verify claims and assess source credibility during answer synthesis, and the same markup feeds traditional rich results.
• Semantic completeness per page - large language models (LLMs) reward content that answers a question fully on its own page, without making the reader click through three levels of navigation. Modular, question-led subheadings read better to both humans and retrieval systems.
• Entity clarity and consistency - your organization, your products, your people, and their relationships need to be described the same way across your site, LinkedIn, review platforms, and any third-party sources. Inconsistent entities are a signal to ignore.
• LinkedIn as a citation surface - a SEMrush analysis of 89,000 LinkedIn URLs cited in AI responses found LinkedIn sitting above Wikipedia and Reddit as the #1 B2B citation source. Native LinkedIn articles account for half to two-thirds of those citations.

If we had to pick one place to start, it's introducing more schema. A clean Organization, Product, FAQ, and Article markup will make your content much easier for AI to find.

The honest read for the next six months

Traditional search isn't dead. It's still the largest channel for most B2B sites. But the ceiling on traffic from it has quietly been lowered, and the fastest-growing distribution channel in most of the accounts we're working with is AI referrals. They still represent an estimated 12-18% of total referrals, but the trendline has been steep since late 2024.

The brands that will benefit most are the ones that stop debating whether to invest in AI visibility and start treating it as the same discipline that governs their SEO program: structured content, clear entities, consistent citations, and a willingness to be specific about what they know.

“A shift in where buyers look doesn't have to be a loss. It just has to be planned for.”

Paul Wright, Marzipan

Frequently Asked Questions

Where Marzipan fits in

If you're auditing where AI visibility sits inside your current SEO program, or mapping a Search/AI visibility roadmap for your B2B site, book a session with us and we'll run it against your existing content, schema, and LinkedIn footprint.

For a read on how AI arrived in search in the first place, the original piece that preceded this one is still a useful primer on the mechanics. More progress checks like this one land in our insights library.