Click Track Data & Marketing
San Diego harbour at dusk
Original research · 60 businesses

Impressions Are Not Demand

We pulled Search Console data across 60 local businesses. Nearly three million impressions produced thirty-two thousand clicks, and the sites with the most impressions converted them worst.

60
businesses
verified Search Console access
2,913,960
impressions
90 days ending 7 September 2026
32,005
clicks
1.10% pooled
1.08%
median site CTR
close to pooled, unlike position data

Quick Answer

An impression is a record that Google considered showing your site. It is not evidence that anyone wanted what you sell. Across 60 local business properties over 90 days, 2,913,960 impressions produced 32,005 clicks. Seventeen of those sites came in under 0.5%, and between them held 40% of all impressions but only 11% of the clicks. The relationship between impression volume and click rate was negative: the 15 sites with the most impressions converted at 0.85%, the 15 with the fewest at 2.74%. If your impressions are climbing while your leads are flat, that is the ordinary case in this data, not a fault.

Forty percent of the impressions, eleven percent of the clicks

Sites grouped by their own click-through rate. The two lowest bands hold more impressions than any other group and almost none of the result.

Site-level click-through rate distribution across 60 businesses
Site CTR bandSitesImpressionsClicks
Under 0.25%8399,545709
0.25–0.5%9763,6362,921
0.5–1%11536,8834,406
1–2%13946,54013,228
2–5%11191,9965,392
Over 5%875,3605,349

The sites with the most impressions converted them worst

This is the part that should change how the metric is read. If impressions tracked demand, the busiest sites would convert at least as well as the quiet ones. They did the opposite.

0.85%
The 15 sites with the most impressions
2,225,948 impressions · 18,991 clicks
2.74%
The 15 sites with the fewest impressions
23,612 impressions · 648 clicks

Across all 60 sites the rank correlation between impression volume and click rate was −0.37. Not a strong relationship, and pointing the wrong way for anyone treating impression growth as progress.

What we cannot tell you yet

We can demonstrate the pattern across 60 businesses. We cannot demonstrate its cause. Three mechanisms are plausible, we have ruled none of them out, and we are not going to pick the most quotable one and present it as a finding.

01

Answers given on the results page

One site's blog post answering a routine measurement question drew 71,798 impressions and 57 clicks at an average position of 6.7. The question has a one-line answer, and a searcher who reads it in a featured snippet or AI Overview has no reason to click. This fits the informational pages in the sample well.

02

Map pack impressions where the click goes elsewhere

A local business appearing in the map pack records an impression, but the searcher may call, tap directions, or open the Google Business Profile. None of those register as a click to the website. We tested this and could not confirm it: on the sites where we expected it, impressions were spread across interior service pages rather than concentrated on the homepage as this would predict.

03

Impressions that were never human

Rank trackers, scrapers and prospecting tools generate impressions. One site in the sample ranked first for meme and pop-culture image searches entirely unrelated to its business. We can see this is happening somewhere in the data. We cannot size it.

The number we are not publishing

We built a click-through rate curve by position and then declined to publish it. Position-one rates in this sample ran from effectively zero to 30.4% depending on the business. Pooled, that is 4.06%; the median site sits at 0.7%; and one property accounted for more than a quarter of all position-one impressions. A single headline figure would have been memorable, quotable and wrong. Any CTR-by-position curve you currently rely on is exposed to the same problem.

What to do with this

  1. 01Stop reporting impressions as growth. An impression is a record that Google considered showing you, not evidence anyone wanted what you sell. Rising impressions with flat leads is the normal case in this data, not an anomaly.
  2. 02Judge a page by the queries it wins, not the volume it accumulates. A page ranking for one question a buyer asks is worth more than one ranking for a thousand questions nobody follows up on.
  3. 03Read position with intent attached. Position one on an informational query and position one on a hiring query are not the same result, and in this sample they did not behave remotely alike.
  4. 04Measure at the end of the funnel. Booked jobs and qualified leads are the only numbers in this chain that cannot be inflated by something that is not a customer.
How we attribute leads to their sourceWhat SEO costs in San Diego

Method

Google Search Console data across 60 distinct business properties where we hold verified access, covering the 90 days ending 7 September 2026, pulled on 9 September 2026.

Where a business had several property variants — www, non-www and domain properties all describing one site — we counted the variant carrying data once rather than summing duplicates. Sixty businesses came from seventy-two raw properties this way.

Query-level figures cover roughly two thirds of impressions, because Search Console withholds low-volume queries. Query totals therefore do not reconcile exactly against site totals, and we have not adjusted for the gap.

All figures are anonymised. Businesses are described by industry where an example is useful and never named, and no per-client figure is published that could identify one.

Want to know what your impressions are actually worth?

We will look at your own Search Console data the same way, and tell you which of your traffic is demand and which is noise. You keep the analysis either way.

Impressions, clicks and what they mean

This is far more common than it looks. Across 60 local business properties we pulled over 90 days, 2,913,960 impressions produced 32,005 clicks, a rate of 1.10%. Seventeen of those 60 sites came in under 0.5%, and between them they held 40% of all impressions but only 11% of the clicks. Several explanations are plausible: the query is answered on the results page so nobody needs to click, the impression came from the map pack where the action is a phone call rather than a visit, or the impression was never a person. We can demonstrate the pattern across the sample. We cannot yet tell you which mechanism dominates.
Not necessarily, and treating it as one wastes money. CTR is a ratio, and in this data the denominator moves far more than the numerator. If impressions triple while clicks hold steady, CTR collapses without anything about your site getting worse. Before treating low CTR as a defect, check whether clicks actually fell. If clicks are flat or rising while CTR drops, nothing broke.
Not on this evidence. The relationship runs the other way. The 15 sites with the most impressions in our sample converted at 0.85%, while the 15 with the fewest converted at 2.74%, and the rank correlation between impression volume and click rate across all 60 sites was negative at -0.37. Impressions grow when a site becomes eligible for more queries, which is not the same as becoming relevant to more buyers.
We deliberately are not publishing that number, because our own data would not support it honestly. Position-one click rates in this sample ranged from effectively zero to 30.4% depending on the site. Pooling them gives 4.06%, while the median site sits at 0.7%, and a single property accounted for more than a quarter of all position-one impressions. A single headline figure would be memorable and wrong. Any published CTR curve you rely on is subject to the same problem.
Google Search Console data across 60 distinct business properties where we hold verified access, covering the 90 days ending 7 September 2026, pulled on 9 September 2026. Where a business had several property variants such as www, non-www and domain properties, we counted the one with data once rather than summing duplicates. Query-level figures cover roughly two thirds of impressions because Search Console withholds low-volume queries, so query-level totals will not reconcile exactly against site-level totals. All figures are anonymised and no client is identified.
Booked jobs and qualified leads, tied back to the channel that produced them. Every metric earlier in the chain can be inflated by something that is not a customer: impressions by tools and irrelevant queries, clicks by curiosity, form fills by spam. Attribution to a booked job is the only point in the sequence where the number cannot be padded by anything except real demand.

Ready to Dominate AI Search Results?

Book a free strategy call and we'll show you exactly where your brand appears (and doesn't) in ChatGPT, Google AI Overviews, and Perplexity.

“They set up automation for every pipeline stage, so nothing falls through the cracks. David and his team are genuinely invested in our success.”

— Christopher Whitney · Google Review