Zero-Click Search and AI Answers: What It Means for Traffic

Zero-click search means a search ends without a visit to any website, because the answer is shown on the page. AI answers extend this: they can fully answer many informational questions. Simple informational traffic is most at risk. Pages that help with decisions, tools and original data tend to keep earning clicks, and visibility inside the answer becomes a goal in its own right.

On this page

• What does zero-click mean?
• Do AI answers reduce clicks?
• From MentionX data: how many links AI answers offer
• Which queries are most affected?
• What does it mean for different businesses?
• How do you measure value without clicks?
• How do you set up the measurement?
• How should your content mix change?
• Which prompts give you a chance at all?
• Is zero-click always bad?
• How do you explain zero-click to leadership?
• Which page elements still earn clicks from AI answers?
• A quick example
• Worked example: Ledgerline rethinks its blog
• What should you do this quarter?
• What mistakes should you avoid?
• Common questions
• Checklist
• Related
• Sources

Updated 31 October 2026. Anything marked "Our view" is MentionX's opinion, not documented engine behaviour. The worked example uses a fictional company.

What does zero-click mean?

A zero-click search ends on the results page. The user reads a featured snippet, a knowledge panel, a calculator, a map pack, or now an AI answer, and stops. This existed long before generative AI: featured snippets were designed to show an answer first (Google). AI Overviews, AI Mode and assistants such as ChatGPT search extend it to more complex questions.

Do AI answers reduce clicks?

It depends on the query and who's measuring.

• Google says AI Overviews show links to a wider range of sources, and that it sees the links included in AI Overviews get more clicks than if the page had appeared as a traditional web listing for that query (Google).
• Many publishers report falling click-through rates on informational queries.
• Studies by third parties vary widely by method, industry and date.

The honest conclusion: some queries lose clicks, some don't, and you need your own data.

From MentionX data: how many links AI answers offer

Method. We ran read-only aggregate queries over completed MentionX runs from 22 August to 29 October 2026 (UTC): 2,154 runs across 18 projects, mostly B2B software, on six engines called through their APIs with web search or grounding on where available. We counted the citations each engine returned. We don't have click data from these answers, so this shows how many chances to click exist, not how many clicks happen.

The share of runs with no citations varied widely by engine: 4.1% on Perplexity (73 runs), 12.6% on Google AI Mode (214), 30.2% on OpenAI (864), 47.0% on Copilot (100), 50.7% on Gemini (803) and 99.0% on DeepSeek (100).

What it means. Where an AI answer has no links, the only possible click is a brand search afterwards. Where it has seven, a reader who does click picks one of several. Either way, being cited is not the same as getting traffic.

Limits. This is a convenience sample of MentionX projects, not a market survey. API results can differ from what a user sees in a consumer app, and Perplexity's sample is small.

Our view: treat a citation as a chance at a click and a mention as a chance at a later visit. Count both, but don't promise traffic from either.

Which queries are most affected?

What does it mean for different businesses?

How do you measure value without clicks?

• AI visibility rates: mention, citation and recommendation rates for your prompt set
• Branded search: rising brand queries suggest AI answers are driving awareness
• Self-reported attribution: add "AI assistant" to "How did you hear about us?"
• Impressions vs clicks: Search Console shows impressions and CTR for your pages (Google)
• AI referral traffic: a custom channel group in GA4 (Google)

How do you set up the measurement?

You can do most of this with free tools in an afternoon.

Search Console. In the Performance report, compare impressions, clicks and CTR for your informational pages over the same months year on year (Google). Filter by page, then by query. Pages whose impressions held while CTR fell are your zero-click candidates.

GA4. Create a custom channel group with a channel for AI assistants, matching referrers such as chatgpt.com, perplexity.ai, copilot.microsoft.com and gemini.google.com (Google). Place it above the default referral channel so those sessions aren't lumped in with other sites. Expect small numbers; the trend matters more than the total.

Your forms. Add "AI assistant (ChatGPT, Perplexity, etc.)" as an option in "How did you hear about us?". It catches people who read an answer, then typed your name into a browser.

AI visibility. Run a fixed prompt set on a schedule and record mention, recommendation and citation rates per engine. Without this, you only see the clicks you lost, not the attention you kept.

Our view: none of these numbers is precise on its own. Read them together and look for the same direction across several.

How should your content mix change?

1. Accept that simple answers will be taken. Keep accurate definitions; they build entity understanding and are often cited.
2. Invest in content that needs a visit: calculators, templates, detailed comparisons, case studies, original research.
3. Make every page useful as a source: clear facts, named data, an obvious reason to click.
4. Build direct channels: email, community, product-led growth.
5. Value being named in answers, even without the click.

Which prompts give you a chance at all?

Zero-click hurts most when you're absent from the answer. In MentionX data, how often the tracked brand was named depended heavily on the kind of prompt:

Problem-style questions, which are exactly the informational queries most likely to end without a click, named the tracked brand in only 5.7% of answers. If your blog lives on those queries, you're losing the click and often not getting the mention either.

Is zero-click always bad?

No. If an AI answer recommends your product by name, the user may never click the citation but search your brand later or go straight to your site. That value doesn't show up as a click on the original query, which is why attribution beyond last-click matters.

How do you explain zero-click to leadership?

Present three numbers together, so a traffic drop isn't read as failure:

If impressions and recommendations hold while clicks fall, the market is changing how it consumes answers, not rejecting you. Report the trend, the likely cause and the content you're shifting toward.

Which page elements still earn clicks from AI answers?

• Something to do: a calculator, template, configurator or free trial
• Something to verify: detailed pricing, specifications, full comparison tables
• Something unique: original data, case studies, expert opinion
• Something current: prices, availability, events

Make these obvious near the top of the page, so a visitor arriving from a citation finds them straight away.

A quick example

Illustrative. A SaaS blog's "what is customer churn" article loses half its clicks over a year, while Search Console impressions remain steady. Meanwhile, the company's recommendation rate for "best churn analysis tool" prompts rises, branded searches grow and more leads choose "AI assistant" as their source. The team stops judging the blog only by sessions.

Worked example: Ledgerline rethinks its blog

Ledgerline is a fictional accounts-payable tool for mid-sized companies. Its blog has 60 informational articles, such as "what is three-way matching" and "how to calculate days payable outstanding". Over a year, Search Console shows impressions on those pages holding roughly steady while clicks fall by about 40%.

The team checks a sample of the queries by hand. Most now show an AI Overview. Ledgerline's definition page is cited in some of them, but the answer fully explains the term, so few people click.

Rather than delete the pages, the team sorts them into three groups:

1. Keep as is: short definitions that are often cited. They still teach the engines what Ledgerline knows.
2. Add something worth visiting: the DPO article gets a free calculator, and the three-way matching article gets a downloadable checklist and a worked table with real-looking numbers.
3. Merge: eight thin articles that overlap are combined into two thorough guides.

In parallel, the team adds "AI assistant" to the demo form's "How did you hear about us?" field and starts tracking recommendation rate on 30 best-of and comparison prompts. Their monthly report now shows clicks, impressions, AI visibility and self-reported source together. Illustrative: the numbers here are made up to show the method.

What should you do this quarter?

1. Pull Search Console data for your top 100 informational pages and flag those with stable impressions but falling CTR.
2. Check whether those queries now show AI answers, and whether you're cited.
3. Add a visit-worthy element (tool, table, data) to the ten most affected pages.
4. Start tracking AI recommendation rate and branded demand alongside clicks.

What mistakes should you avoid?

• Measuring success only by sessions
• Deleting informational content that still builds entity understanding
• Blocking AI search crawlers out of frustration, which removes you from answers altogether
• Trusting a single third-party study as universal truth

Common questions

Should we block AI crawlers to protect our traffic? Blocking AI search crawlers such as OAI-SearchBot removes your pages from those engines' answers, so you lose the mention and the citation as well as the click (OpenAI). Blocking training crawlers is a separate decision with different trade-offs.

Do featured snippets still matter? Yes. Google still shows them for many queries (Google). Our view: the clear, direct answers that earn a snippet are also easy for AI answers to reuse.

Will people stop clicking altogether? Not for everything. People still click to buy, sign up, compare in detail, check a price or use a tool. The share of quick fact-finding visits is what's shrinking.

Is this only a Google issue? No. Assistants such as ChatGPT, Perplexity and Copilot answer questions directly too, and as our data shows, many of their answers carry few or no links. Google is simply where most search traffic, and so most of the measurable loss, still sits.

How do we know if AI answers are behind a drop? Check whether impressions held steady while CTR fell, and whether the queries now show an AI answer. If impressions fell too, look at rankings and demand first.

Checklist

• Query types mapped by click risk
• AI visibility tracked
• Branded search and self-reported attribution tracked
• Content mix shifted toward visit-worthy assets
• Direct channels developed

Related

• Is SEO dead?
• Why AI answers differ every time

Sources

• Google Search Central, Featured snippets and your website
• Google, Google I/O 2024: New generative AI experiences in Search
• OpenAI, Overview of OpenAI crawlers

• Google Search Console Help, Performance report
• Google Analytics Help, Custom channel groups
• Google Merchant Center Help, Product data specification