Open Search Console for almost any content site and you will probably see the same picture: impressions flat or climbing, clicks sliding down. It looks like a tracking error. Usually it is not. More of the answer now sits on the results page itself, and fewer people leave it.
That puts anyone who reports on SEO in an awkward spot. If organic traffic is your headline number, you are explaining a decline that may have little to do with the quality of your content. This guide covers the zero-click search metrics worth putting in its place, and how to approach measuring SEO success in 2026 without pretending that clicks have stopped mattering.
What zero-click does to your numbers
Zero-click behavior is older than AI. Featured snippets, knowledge panels, maps and weather boxes have been answering queries on the results page for years. AI Overviews added a fuller, synthesized answer on top of that, and the effect shows up in the data.
Pew Research Center tracked the browsing of 900 US adults in March 2025. Users clicked a traditional search result in 8% of visits to results pages with an AI summary, compared with 15% of visits to pages without one. Ahrefs analyzed 300,000 keywords using December 2025 data and estimated that the presence of an AI Overview was associated with a click-through rate for the top organic result roughly 58% lower than the modeled CTR without an AI Overview (Ahrefs study).
Take those numbers as a direction, not a forecast. The Ahrefs figure is an association measured against a modeled baseline, not proof of cause. The methods differ, Google has disputed Pew’s reading, and the effect on your site depends on your mix of queries. Informational questions get hit hardest. Branded, transactional and local searches tend to hold up better.
Why the old dashboard misleads you
- Clicks fall while influence does not. Someone reads an AI answer that cites you, remembers the name, and searches for you directly a week later. The click you get is branded, and the first touch never shows up anywhere.
- Rank no longer equals visibility. Position one under a large AI answer is a different asset than position one used to be.
- Averages hide the split. A sitewide CTR blends branded and non-branded queries, informational and commercial ones. A healthy segment can mask a collapsing one.
- Last-click reports ignore everything before the click. If your team is judged on them, it will drift toward content that is easy to click, not content that builds a reputation.
The metrics that still matter in 2026
Nothing in this table is new. What changes is how much weight each line carries and how carefully you segment it.
| Metric | What it tells you now | Where to find it | Caveat |
|---|---|---|---|
| Organic clicks | Demand you actually capture | Search Console, GA4 | Falls with zero-click even when content is fine |
| Click-through rate (CTR) | How well your result competes on the page | Search Console | Only useful split by query type and page |
| Impressions | Whether you are still shown | Search Console | Includes impressions inside AI features |
| Branded search demand | Awareness and recall | Search Console branded queries filter, Google Trends | Also moves with campaigns and PR |
| Engagement rate | Whether arriving visitors find what they need | GA4 | Compare organic landing pages, not the whole site |
| Organic conversions and revenue | The result the business cares about | GA4, CRM | Attribution gaps on long journeys |
| Share of voice (SoV) | Your visibility against named competitors | Rank and AI tracking tools, manual tracking | Depends on the keyword or prompt set you choose |
| Generative AI impressions | Whether your pages appear in generative answers | Search Console generative AI report | AI Overviews and AI Mode combined; impressions only, no clicks or CTR |
| Brand mentions | Whether people and publishers talk about you | Alerts, monitoring tools, manual search | Quality varies a lot |
Organic clicks and CTR
Keep both, but stop reading them at site level. Split queries into branded and non-branded using the branded queries filter in Search Console, then into informational and commercial intent. When clicks drop and impressions hold, the next question is which queries moved. Often it is a cluster of how-to and definition pages, and the fix, or the decision to accept the loss, is different for each cluster.
Engagement rate and conversions
If fewer people arrive, the ones who do may be further along in their decision. Check whether that is true for you. Look at GA4 engagement rate and conversion rate for organic landing pages, compared with the same pages a year ago. In some accounts the smaller audience converts better. In others it simply is smaller. Both are useful to know before you decide how much a lost informational click was worth.
AI Overviews tracking: what you can measure today
The Search Console generative AI report
On June 3, 2026, Google launched dedicated generative AI performance reports in Search Console (Google’s announcement). The Search report covers AI Overviews and AI Mode, and there is a separate one for Discover. At launch it showed impressions by page, country, device and date, without clicks or CTR, and its data starts from May 18, 2026. Rollout was staged and the report was labelled beta, so check whether it is available for your property.
Two things to keep in mind when reading it:
- AI Overviews and AI Mode are grouped together. You cannot isolate one from the other.
- These impressions were already counted in your main performance report. The new view separates them, it does not add them.
The most practical use is page-level. List the pages with the most generative AI impressions, then compare their clicks and CTR in the standard report. Pages that are shown often but rarely clicked are your zero-click exposure. Pages with both are the ones to protect.
Build a simple citation tracker
Search Console tells you where you appear. It does not tell you what the answer said about you, or who else was cited. For that, build a small tracker, in a spreadsheet or in a dedicated tool.
- Pick 30 to 50 queries or prompts that reflect how customers actually ask questions in your category. Include a few branded ones and a few competitor comparisons.
- For each one, record whether an AI answer appears, whether your brand is cited or named, which URL was used, and which competitors showed up.
- Run the same list on a fixed schedule, monthly for most teams, from the same country and device type.
- Note whether the description of your brand is accurate. A citation that gets your pricing or features wrong is worth flagging.
AI answers vary from run to run and between users. Treat each check as one sample, and read the trend over several months, not any single result.
Brand visibility in search: measure what clicks cannot see
Brand visibility in search is the part of your performance that happens before, or instead of, a visit. It is harder to measure, so most teams skip it. That is a mistake, because it is where a lot of zero-click value lands.
Branded search demand
Branded impressions and clicks in Search Console, branded interest in Google Trends and direct traffic all point to the same question: are more people looking for you by name? If your content is cited often in AI answers and branded demand rises over the next few quarters, that is a reasonable sign the visibility is turning into awareness. If branded demand is flat, the citations may not be doing much. Neither reading is proof, but both are better than guessing.
Share of voice and impression share
Share of voice (SoV) answers a simple question: out of all the visibility available on a set of queries, how much is yours? Define the set first, then count.
There are two ways to measure it, and they should not be mixed in one calculation.
- Organic SoV is the share of keywords in a defined set where you hold a top-three ranking.
- AI citation share is your share of brand mentions or citations out of all citations recorded across a defined set of prompts.
\text{Organic SoV} = \frac{\text{keywords where you rank in the top three}}{\text{keywords in the set}} \times 100
\text{AI citation share} = \frac{\text{your citations}}{\text{all citations recorded across the prompt set}} \times 100
Example: you rank in the top three for 18 of 60 tracked keywords, so your organic SoV is 30%. Separately, across 50 tracked prompts your brand is cited 12 times and all brands combined are cited 80 times, so your AI citation share is 15%. Run both against two or three named competitors, not the whole market, and keep the sets stable so the trend means something.
Impression share is a related idea from Google Ads: your impressions divided by the impressions you were eligible to receive. Organic search has no official equivalent, but it is worth checking on your branded paid campaigns. A low impression share on your own brand terms means you are leaving branded demand to others.
Brand mentions and unlinked brand references
AI systems and search engines pick up on how often, and in what context, your brand appears across the web, whether or not there is a link. So count both linked and unlinked brand references.
- Set up alerts for your brand name, product names and the names of key people.
- Search your brand name with your own domain excluded and scan the first few pages for mentions without links.
- Log the source, the context and whether the tone is positive, neutral or negative.
- Weigh by source quality. Ten mentions on forum threads are not the same as one in an industry publication.
For unlinked mentions from sources that matter, a short, polite outreach asking for a link is often enough. Do not make that your only goal, though. The mention itself has value.
Where Generative Engine Optimization fits
Generative Engine Optimization (GEO) is the practice of making your content more likely to be cited in AI-generated answers. In practice, this is less a separate optimization discipline and more about improving the clarity, usefulness, authority and discoverability of content that AI systems can retrieve and reference. Google’s guidance says that fundamental SEO best practices remain the foundation for appearing in AI Overviews and AI Mode, and that no special techniques are required, so be careful with anyone who sells GEO as a secret formula. Google’s own guidance on optimizing for generative AI search is a better starting point than most third-party checklists, and much of it overlaps with good SEO: clear, accurate, well-structured content from a source that is demonstrably expert on the topic.
What matters for measurement is that GEO gives you a few new things to count:
- Citation frequency: how often you are named or linked in AI answers for your tracked prompts.
- Citation share: your portion of all citations recorded for the prompt set, the AI version of share of voice.
- Accuracy: whether what the answer says about you is correct.
- AI referral traffic: visits from AI assistants. In GA4, group sources such as chatgpt.com, perplexity.ai, gemini.google.com and copilot.microsoft.com into one custom channel. Expect an undercount, because some of these visits arrive without a referrer and land in direct traffic.
A reporting setup that holds up
One way to avoid a 40-slide report is to organize everything into three layers and keep each one short.
Layer 1: Visibility
- Impressions, including generative AI impressions where available
- Organic share of voice and AI citation share on your tracked sets
- Citation frequency and accuracy
- Branded search demand
Layer 2: Engagement
- Clicks and CTR by query type
- Engagement rate on organic landing pages
- AI referral sessions
Layer 3: Business results
- Organic conversions and assisted conversions
- Pipeline or revenue influenced by organic
- Conversion rate of organic visitors against other channels
Read them top to bottom. If visibility is up and results are flat, you have a conversion or targeting problem. If visibility is down, fix that first. If visibility is up and clicks are down, you are seeing zero-click at work, and the decision is whether the exposure is worth it.
Mistakes to avoid
- Reporting a sitewide CTR as if it were one number with one cause.
- Blaming every traffic decline on AI Overviews. Check for ranking losses, seasonality, technical problems and algorithm updates first.
- Changing your tracked prompt set every month, which makes any trend meaningless.
- Treating the generative AI report in Search Console as a traffic report. It counts impressions.
- Chasing mentions in low-quality places because the count looks good.
- Dropping informational content altogether. Some of it feeds the brand awareness and citations you are now trying to measure.
Where to start this week
- Split your Search Console data into branded and non-branded, and compare the last six months with the same period a year ago.
- Check whether the generative AI report is available for your property, and export the pages with the most AI impressions.
- Write down 30 prompts your customers might ask, and run them once to get a baseline for citations and share of voice.
- Add engagement rate and conversions for organic landing pages next to clicks in your monthly report.
The numbers will not be perfect, and the tools will keep changing. A consistent, honest set of measurements, even a rough one, will tell you more than a traffic chart that nobody trusts anymore.
