First place in Google that gets no clicks. How should SEO be measured today?

Publishers in the Chartbeat network lost 40 percent of their Google traffic in a year, while Search Console can at the same time show their pages in position 1. We looked into why this happens and what to report to management instead of rankings.

Google traffic is falling, and faster and faster

Chartbeat, a company selling analytics tools to publishers, published the “2026 Publisher Playbook” report. According to it, traffic from the Google search engine across its clients’ network fell by 40.2 percent between July 2025 and July 2026. A year earlier, the decline was 21.9 percent. Google Discover (the recommended content channel in the Google app) accelerated even more sharply: from 6.6 to 34.3 percent. In July 2024, the search engine delivered about 9 percent of page views in this network; by July 2026, only 5 percent.

These figures should be treated cautiously. The Chartbeat network consists mainly of news sites, the report does not give the number of websites, and Search Engine Journal points out that it is unclear whether the yearly declines concern the number of page views or the channel’s share. The direction is, however, confirmed by Graphite, another analytics company focused on SEO (search engine optimization). In data from January 2026, cited by Search Engine Land, Google traffic among publishers fell 38 percent year over year.

The same data shows that total organic traffic in the US lost only 2.5 percent, and the biggest sites even grew slightly. The declines are concentrated among mid-sized sites and publishers. The brands performing best are those people visit directly, without going through a search engine.

Why “position 1” in an AI summary can be misleading

Google Search Console, the free tool for tracking a site’s visibility in the search engine, counts positions in AI Overviews — answers generated by AI above the results — in a simplified way. The entire summary block gets one position. Because it usually sits above the organic results, every link inside it is reported as position 1.

The same applies to the highlighted card under the first sentence of the answer and the link hidden under the “Show more” button. The chance that someone clicks one or the other is completely different, but in the report both placements look identical. John Mueller, Search Advocate at Google, admitted on Reddit that tracking traditional rankings for AI features is hard to do in a useful way.

Average position adds another problem. Search Engine Journal gives this example: a page is in position 1 when an AI summary appears, and in position 19 when it does not. If both cases happen equally often, the report will show a neat average of 10, which corresponds to nothing users actually saw.

Impressions counted automatically, clicks getting rarer

The second trap concerns impressions. A link in the default visible content of the summary gets an impression the moment the page loads, even if the user read the first sentence and left. Links behind “Show more” get none at all until someone expands the section.

Impressions counted this way distort CTR, the share of impressions that end in a click. Seer Interactive, an SEO agency, recorded a drop in CTR for queries with an AI summary from 1.76 to 0.61 percent. According to Search Engine Journal, this drop is largely due to Search Console counting impressions automatically, which inflates the denominator.

User behavior is also changing, however. In March 2025, Pew Research Center analyzed real web browsing by 900 adult Americans. When an AI summary appeared, a traditional result was clicked in 8 percent of searches; without a summary, in 15 percent. A link in the summary itself was clicked in 1 percent of cases. At that time, summaries appeared in about 18 percent of searches, and according to data cited by Search Engine Journal, now in 21 percent, more often for question-based queries.

SEO tool vendors report even bigger declines. Ahrefs estimated in December 2025 that when an AI summary is present, the CTR of the page in the first results position is on average 58 percent lower. In February 2026, Sistrix pointed to a drop in CTR for the first position from 27 to 11 percent. These are company estimates whose methodology cannot be independently verified.

More links in AI answers do not necessarily mean more visits

Since January 2026, Google has several times announced more links in AI answers. On September 21, 2026, Peec AI, a company analyzing AI-based search, showed a chart according to which the share of summaries with an external link in the answer text rose from almost zero before September 11 to 26.2 percent on September 20. The chart does not provide sample size, markets, or devices, so it is unclear whether the change results from a Google decision or a change in how the data was collected.

On the same day, SEO practitioner Gagan Ghotra published a recording in which highlighted text in the summary led not to a website, but to AI Mode, the AI conversation mode in the search engine. He described it as a Google test. The recording concerns one search, and it is not known how often Google shows such a link. If Search Console treats it as a query refinement (a link that launches a new search), the page will get neither a click nor an impression; Google’s documentation does not settle the matter.

The new generative AI report in Search Console, made available to all websites on August 31, 2026, shows only impressions in AI Overviews and AI Mode. Clicks from AI features are mixed into the general search report.

With Google’s current tools, it is therefore impossible to determine how many visits being present in AI summaries brings. It is, however, possible to check what is happening on your own site.

What to report to management instead of rankings

Search Engine Journal proposes treating visibility in AI as the simple question “is the brand cited,” and assessing effectiveness using metrics not distorted by the way impressions are counted. On this basis, we compiled the following metrics for a management report:

• search-engine visits and their year-over-year change, broken down by page,
• inquiries, sign-ups, and sales attributed to organic traffic,
• brand presence in AI summaries as a yes-or-no answer, without average position,
• direct traffic and the number of loyal readers or customers.

The last point is well illustrated by Chartbeat data. Loyal readers, meaning visitors who visit a site at least 8 of the last 16 days, generated an average of 14.98 billion page views per month in 2024 and 15.02 billion in 2026. However, returning readers fell by 8 percent and new readers by 14 percent, so the stability concerns only the most loyal group. The share of direct traffic also grew (from 13.5 to 15.9 percent), as did dark social, meaning links sent in private messages (from 7.1 to 11.3 percent).

Chartbeat advises treating SEO as one of several channels and shifting resources toward audiences the company has more influence over. That is a recommendation from a company selling analytics tools, but our conclusion is similar: position in Search Console has ceased to be a reliable measure of success. Management that receives a report with “first places” may fail to notice that customers from search are disappearing. It is safer to show visits, conversions, and traffic the company built itself.