My traffic dropped after AI Overviews. Where do I start?
Separate a click-through problem from a ranking problem, a demand problem, and a broken-tracking problem before you pay anyone to fix the wrong one.
If your pages still rank where they always did and the traffic has quietly gone, nothing is broken on your site. Something changed above it.
Ask Google, ChatGPT, or Perplexity a question today and you get an answer rather than a list. That answer names a small number of businesses. If yours is not one of them, sitting tenth or fifth or second on the page underneath it is worth a great deal less than it used to be.
Search still works. The click no longer follows it, because an answer sits above your listing and names a handful of companies. So the question worth asking stopped being where do I rank. It is now whether you get named, and who gets named instead of you. That is measurable, and we measure it free.
The scan is a diagnostic, not a guarantee. It reports what the engines actually said when we asked them, and who they named. Nobody can promise an AI will recommend a particular business, and anyone selling you that promise cannot deliver it.
The clearest case we have is a content site we own, so we have the full search record for it rather than a client’s partial view.
One page on that site held its Google position for a year. It averaged 5.7 at the start and 5.5 at the end, which is the kind of stability an owner would normally be pleased with.
Across the same year, the share of people who saw it and clicked went from 1.55% to 0.00% while site-wide daily impressions fell from 3,672 to 432.
Nothing about the page got worse. What changed is how much being on the page is worth, because the answer above it had started doing the choosing and that answer did not name us.
A ranking report can look healthy for a year while the business it belongs to stops receiving visitors.
Before changing anything, get the measurement. We ask real answer engines the questions your buyers would type, count how often your business gets named, record who got named instead, and collect the exact pages the engine used as its sources.
The report lands within one business day. There is no signup, no card, and no sales sequence behind it.
Twelve pieces follow, in three groups. Read the one that matches the decision in front of you, because nobody needs all twelve.
Start here if the numbers moved before you did anything. Half of this work is ruling out the ordinary explanations before you spend on the new one.
Separate a click-through problem from a ranking problem, a demand problem, and a broken-tracking problem before you pay anyone to fix the wrong one.
Six reasons an engine names somebody else, and how to work out which one is yours before you act on a guess.
Test the engines yourself, track citations over time, and build a baseline that later numbers can be compared against.
This is the work itself. It overlaps heavily with fundamentals you may already be paying for, which is worth knowing before anyone sells you a second retainer.
The work that makes a business easy for an answer engine to understand, retrieve, trust, and repeat back to a buyer.
Being named and being cited are two different scoreboards. Only one of them sends anybody to your website.
There is no submission form and there never was. Seven things that make a business clearer, corroborated, and worth citing.
What Google’s own published guidance says about qualifying, and how to measure afterward whether any of it worked.
Three overlapping disciplines in plain English, and which numbers actually tell you whether any of them is working.
The market here runs from a small monthly subscription to a full engagement, and the gap between them is mostly about who does the work rather than who sees the problem.
From a software subscription to a done-for-you engagement, with the things that actually drive the number either way.
Seven of them compared on engine coverage, the workflow they expect from you, and published pricing.
Who is credible, judged on execution depth, public evidence, and how clearly they are willing to measure themselves.
A tool shows you the gap. Closing it is a different purchase entirely, and your own capacity decides which one you need.
The measurement is free and takes a day. Everything above is what you do with it afterward.
Some readers arrive here evaluating a spend rather than diagnosing a drop. That is a different decision with different math, and it deserves its own answer.
Buying a sponsored placement does not change what the assistant recommends. The two things sit in the same interface and are otherwise unrelated, which is the most expensive misunderstanding in this category right now.
Yes. One engine, six questions a real buyer would ask, and a report inside one business day. No signup, no password, no card, and no sales sequence afterward. If it turns out you are already being cited well, we will tell you that too.
How often the engine named your business, which competitors it named instead, the exact pages it cited as its sources, and the specific questions you did not appear on at all.
No, and neither can anybody else. What is possible is measuring where you stand today, seeing who is being cited in your place, and working on the gap between the two.
No. The technical and editorial fundamentals that earn a ranking are largely the same ones that make a page worth citing in an answer. The work is additive rather than a replacement, which is also why anyone quoting you a second full retainer should be asked what is genuinely new in it.
No. A sponsored placement is labeled, appears below the answer, and does not influence what the assistant recommends. If advertising is what you are evaluating, the ChatGPT ads hub covers it properly.
Longer than a ranking change, and less predictably. Engines refresh what they know at their own pace. That is why the free scan repeats on day 21 and again on day 60, so you can watch the difference rather than guess at it.
They are used to run your scan and to send you your report. We do not publish research drawn from the businesses that request one, and we do not add you to a marketing sequence.
We will ask the engines six questions your buyers would ask, and send you what came back. It costs nothing and it takes a day.
The scan reports what answer engines returned at the time they were asked. Answers vary between runs and between engines, and no provider discloses how its recommendations are produced. NAMYNOT cannot guarantee that any engine will name, cite, or recommend a particular business.