What happens when good rankings stop getting search query clicks
- Over 16 months my site was shown 19,654 times for long search phrases and received no clicks from any of them
- Average position on those phrases improved from 36 to 21, so the rankings got better while the outcome got worse
- Monthly appearances for those phrases fell roughly 75% year on year
- Pew Research found long, conversational searches are the ones most likely to trigger an AI summary and AI summaries roughly halve the chance of a click
- Almost half of the phrases were from people who could never become a customer of mine, which is the part most business owners miss & where I haven’t been disciplined with my own content
- The fix is not more content creation, it is knowing which of your pages were ever attached to relevant money making phrases in the first place
The report that looked like good news
I pulled every search phrase of eight words or more that my own site appeared for over the last 16 months. That is the conversational end of search, the way people type when they are asking a question rather than firing off a two or three keyword search. This style of search is happening more and more as people use Ai assisted tools and apps to ask about things more naturally to have a virtual conversation.
The headline numbers looked healthy as the average position was improving from 36 to 21 which is fifteen places and good progress towards first page exposure if it continued.
In most circumstances that is a great reading to see in a Google Search Console report.
But, the clicks column was showing zero, so at that point it is a bit of a vanity metric.
Not surprising there weren’t any clicks from search considering none of them were in the top 3 positions but it is a wide array of 804 phrases, across 19,654 impressions over 16 months I wasn’t generating any actual traffic.
Plus the impressions were dropping;
- Between May and August 2025 my site was shown about 1,980 times a month for those phrases.
- Between April and August 2026 it was about 487 times a month.
- A drop of roughly 75%, while the positions were getting better.
I have written before about how website data without context can present a false positive.
This is a great example using my own website as a reference.
Why better positions produced worse outcomes
Two things are happening at once and they pull in opposite directions.
The first is that long, question-shaped searches are exactly the ones that now get answered on the results page instead of sending someone to a website. Pew Research Center tracked the real browsing of 900 x US adults across 68,879 Google searches in March 2025.
- Around 8% of one or two word searches produced an AI summary.
- For searches of ten words or more, it was 53%.
- Question-based searches beginning with who, what or why produced one 60% of the time.
Source: Pew Research Center, “Google users are less likely to click on links when an AI summary appears in the results“, July 2025.
The second is what happens once that summary appears. In the same study;
- Visits to a results page with an AI summary produced a click on a normal result 8% of the time, against 15% without one.
- Clicks on a link inside the summary itself ran at about 1%.
A separate Ahrefs study published in February 2026 compared 150,000 keywords that trigger an AI Overview against 150,000 that do not, using aggregated Search Console data. Position one click-through rate on the AI Overview keywords fell to a level roughly 58% below where the same keywords would have been forecast to land without one.
Source: Ahrefs study on AI Overview click-through impact, February 2026
So the long phrases I rank better for are the same long phrases most likely to be answered without a click from search. So my position was improved but with no real benefit of traffic coming to my website.
The part that actually matters for your business
Here is where I would stop worrying about AI for a moment, because the more useful finding in my own data has nothing to do with it.
I sorted all 804 phrases by what the person was actually trying to get information for:
Every search phrase of eight words or more that my site appeared for over 16 months, sorted by what the person was actually trying to do.
| What the person wanted | Share of appearances | Could they become a client? |
|---|---|---|
| Help with domain names | 31% | Almost never |
| To stop receiving unwanted emails | 13% | Almost never |
| Answers to exam and quiz questions | 4% | No |
| To hire or price a marketing consultant | 13% | Yes |
| Everything else | 39% | Mixed |
Nearly half of my long-phrase visibility comes from people solving a problem I do not get paid to solve. My three highest-appearing articles are about domain names and unwanted emails. They are genuinely useful pages and I am glad they exist as they have a part to plat when helping business owners manage their digital marketing efforts.
Though they have never produced a new client for me even if they have helped many existing ones to make strategic decisions.
That is the honest reading.
In my case it is not “AI killed my traffic”.
It is “most of that traffic was never commercially connected to anything relevant to me and now I can see it clearly because not only is the search query click number zero but impressions have fallen away.” The truth is in the data
If your own reports look busy but your phone is not ringing, run the same sort of report using Google Search Console or looking at the top pages being visited in Google Analytics.
Take your top 20 pages by appearances and ask one question about each: if a person read this page and were completely satisfied by it, would they then need to pay somebody?
Most sites have one or two pages where the answer is yes and a lot of pages where it is no. Many of those extra pages would help give a website context or relevance for their niche, but they were also very generic pieces of information. AI involvement with search doesn’t reward those pages anymore.
What I would change and what I would not
1. Do not write more of the same thing.
Google’s own guidance is explicit that producing separate content for every variation of how people might search, in order to catch more AI-generated follow-up queries, falls foul of its scaled content abuse policy. It also states plainly that you do not need to rewrite your content specially for AI systems, break it into chunks, or add llms.txt files.
Source: Google Search Central, “Optimising your website for generative AI features on Google Search“.
2. Do change what you measure.
Appearances in search were always a weak proxy for demand. They are now close to meaningless on informational topics. Track enquiries, calls, quote requests and revenue. If you are not sure how to see any of that, my walkthrough of what you can actually use Google Search Console for covers the reports worth opening and the ones worth ignoring.
3. Do check whether your commercial pages are the ones ranking.
In my data, the pages that draw the most appearances are informational while my home page & key service pages are in the top 5 for many phrases that do most likely to belong to a buyer. That is a structural problem on my site, not an algorithm problem. It is fixable and it has nothing to do with AI so I have to address my content.
4. Do keep publishing the things only you can write.
The one bright spot in the export: a case study I wrote about a 4WD accessories retailer ranks at position 1.1 for a conversational query asking “which Australian 4WD stores are strongest at content marketing”. A case study naming a real industry and a real result is being retrieved as evidence, because there is nothing generic to substitute for it. General advice about content marketing is not.
What I would want you to take from this
Keywords, impressions, rankings are all just pieces of data.
For years they were a reliable enough proxy to measure potential attention & visibility. That proxy has now changed in relation to exactly the sort of question-shaped searches most small business content is written to answer.
The businesses that will feel this least are the ones whose content was tied to a commercial outcome to begin with & who had very specific, very detailed content created for it.
The businesses that will feel it most are the ones who have spent years working on creating lots of articles with generic information hoping that it would turn into more traffic & hopefully equal more customers from getting more exposure. For many of my clients that is exactly what I helped them do and in many cases it did work, it was exposure they would have not gotten otherwise.
Although in the era of AI answering most common questions that is now shifting with those generic pieces not getting exposure. Instead what the platforms are rewarding is content that comes from first hand experience, an opinion or point of view, articles that are not generic in nature but very specific to a topic from an entity they can identify as credible.
The other scenario to consider
So far I’ve spoken about impressions or clicks for things that may not be relevant to your business.
What if the situation was directly related to income generating terms that are for your service or the product you sell?
That it a whole different scenario and obviously more concerning if you rely on it to find new customers. That type of change could be occuring for various reasons;
- Something was changed on the website in terms of URL, headings, page content or even a loss of backlinks to a page.
- Your closest competitors improved the content on their pages.
- An algorithm core update was rolled out and it determined your content didn’t fit the same requirements any longer.
- Customer behaviour changed and they don’t need what you sell as much.
- It is a seasonal shift so for example a B2B business doesn’t get much search traffic on a Saturday or Sunday or over holiday periods.
- You went through a website migration to refresh the site that looks the same but lacks the correct technical SEO structure to sustain what it was winning before.
It could be a combination of all of those things. If you want a second opinion on which of your pages were ever attached to revenue or how your content is performing, that is exactly the sort of digital audit that I do. Have a look at how I work with businesses.
