AI Consulting

Your traffic didn’t fall because your SEO got worse.

The number of people who saw your link held up. The number who clicked it did not. That is not an agency failing, and no amount of the work you have been buying will fix it. Here is how to check your own two numbers in ten minutes.

The result page now answers the question. The click was the only thing you were counting.

clicks ÷ impressions

MeasuredGoogle’s own definitions: an impression is your link being SEEN, a click is somebody arriving. They are separate metrics and they stopped moving together.

Your own number is sitting in a report you already have.

10 minutes

AssumedHow long it takes to put impressions against clicks for the same twelve months in Search Console. No agency, no tool, no login you do not already own.

We cannot tell you how far yours fell, and neither can anyone quoting you a market-wide figure.

not measured

Not measuredWe have not measured discovery decline across Indian owner-led businesses. The market-wide percentages in circulation are vendor-published and not reproducible from anything you can check.

For thirty-one years, British law required a man to walk ahead of every self-propelled vehicle carrying a red flag.1The stated reason was safety. The actual effect was to make a new thing behave like the old thing, at the old thing’s speed, until 1896.

Nobody in 1865 said “let us hold this back.” They said “let us regulate it sensibly,” and what came out was a machine walking at the pace of a man. That is what an incumbent does with a new channel: not reject it, reshape it into the familiar one. Older technology rarely disappears when something better turns up. It stays, and somebody keeps paying for it.2

Which is roughly where a great many businesses are with their website. They are still buying an asset built to catch clicks from a page that has, quietly, stopped handing them out.

Two numbers that used to move together

Google reports two separate things about your site, and it defines them differently on purpose.3

  • An impression is how often someone saw a link to your site. Saw it. That is the whole definition.
  • A click is how often someone actually came to you.

For most of the last twenty years these two moved together, near enough that nobody had to think about the difference. More people seeing you meant more people arriving. Everything the industry sold you — rankings, keywords, content, position — was sold on that assumption, and the assumption was fine.

What changed is not your ranking. It is what sits between the two. When the result page answers the question where it stands, the person has what they came for. They saw your link. They had no reason to touch it. The impression happened. The click did not.

You did not get worse at being found. You got worse at being visited, and those were never the same thing — you just never had to tell them apart before.

how manytwelve monthsIMPRESSIONS — they saw youCLICKS — they camethis gapis the whole storysame site · same pages · same work being paid for
The shape to look for in your own report. If the top line holds and the bottom line falls away, the problem is not your ranking. If both fall together, it is, and that is a different paper.

Why the invoice kept arriving

Here is the uncomfortable part, and it applies to people in my line of work as much as to anyone.

An agency doing search work is paid for producing rankings and pages. Both of those things still happen. The rankings can hold. The pages get written. Every number on the report can be green while the only number that ever mattered to you — people arriving — goes down. It is not usually a con. It is a service that was defined against a step in the chain rather than against the end of it, and the chain changed underneath everybody at once.

Which is why “fire the agency and hire a better one” so often changes nothing. The second agency is measured on the same step.

The thing worth noticing about what still works

When the top of the funnel thins out, the visits you keep are the ones that were never really search in the first place: somebody who already knew your name and typed it, somebody a friend told, somebody from a message or a map listing or a photograph.

Those visitors behave completely differently. They arrive already half-decided. That is worth knowing before you spend another rupee trying to replace the discovery traffic with more of the same, because the honest reading of that pattern is often: stop buying reach, start being worth repeating.

someone with a questionthe page answers ita few still arrivesomeone who knows your namethey type it, or someone sent themthey arrive, half-decidednothing in the middle to squeeze
Two ways in. The left one is being squeezed by something you do not control. The right one is not, and it is where the people who actually buy have been coming from for a while.

What we are not going to tell you

There are percentages going around. Traffic down this much, zero-click up that much, a decimal point on it to make it look counted. We are not going to add one, because we have not measured this across Indian owner-led businesses and we are not going to repeat a number we cannot reproduce.

More to the point, a market-wide figure would be useless to you even if it were sound. Your business is one site, in one town, in one trade. What you need is your own two lines, and you can have them today.

Ten minutes, on your own site

  1. Open Search Console. If you have never logged in, whoever built the site can add you in two minutes. You should own this login, not your agency.
  2. Performance report. Turn on both boxes— total clicks and total impressions — and set the range to the longest you can.
  3. Look at the shape, not the numbers. Impressions holding and clicks falling means the result page is answering for you. Both falling together means a ranking problem, which is the ordinary kind and is fixable in the ordinary way.
  4. Now split it. Filter queries containing your business name against everything else. The first group is people who already knew you. The second is discovery. Watch them separately from here on, because they are two different businesses and one of them is being squeezed.

Do that before you buy anything. If the two lines are still moving together, this paper is not about you and you should keep doing what you are doing.

How this paper was made

This paper makes one structural claim and gives you a way to test it on your own site. The structural claim is that impressions and clicks are separately defined and separately reported, which is documented by Google and cited below, and that a page which answers a question in place has no reason to send the click.

The size of the effect on YOUR site is a measurement only you can make, and the instructions for making it are in the last section. We have not run that measurement across a population, so this paper contains no market-wide figure. If you have seen one quoted at you with a decimal point in it, ask who counted, on which sites, over which months, and whether you could reproduce it.

No client data appears here. No business is named or implied.

On the date at the top of this page. This paper is dated 29 June 2026 because that is its slot in the series. The writing and the working were done on 26 August 2026, when the series was compiled and released together. We would rather say that here than have you find it in the page history.

References

  1. Parliament of the United Kingdom (1865). Locomotive Act 1865 (28 & 29 Vict. c. 83). Repealed by the Locomotives on Highways Act 1896. The red flag law: a man on foot ahead of every self-propelled vehicle, for thirty-one years.
  2. Edgerton, D. (2007). The Shock of the Old: Technology and Global History since 1900. Profile Books. Old technology does not vanish when a better one arrives. It lingers, and it is paid for.
  3. Google (2026). Search Console Performance report: impressions, clicks and position. Google Search Central Help. The definitions this paper leans on. An impression is the link being seen; a click is somebody arriving.