I measured five Google Business Profiles against their websites for 241 days

About once a month somebody forwards me a screenshot of Search Console with the same sentence attached: 33,000 people saw my site and 23 clicked, is it broken? For a long time I answered that honestly but incompletely — I talked about titles and rankings and competition. Then I finally did the obvious thing and put the Google Business Profile numbers next to the website numbers for the same dates, on five businesses at once. The comparison was not close, and it was not in the direction I would have guessed.

The screenshot above, by the way, is mine. This website — the one you are reading — collected 33,563 impressions and 23 clicks from Google web search between 9 January and 6 September 2026. In the same 241 days my Google listing sent 44 people to it, plus five phone calls and 367 requests for driving directions to my office.

So the tool I had been staring at was showing me roughly a third of the story, and the worse third.

Does a Google Business Profile bring more customers than a website?

On the five businesses I measured, yes — every single one. Their profiles sent 415 visits to their own websites over 241 days while Google web search sent 258, and the profiles generated a further 353 calls and 1,164 direction requests that web search cannot produce at all. The gap was widest on the businesses with the least competitive websites.

Before the table, the boring part that makes the table mean anything.

The two dashboards are not counting the same thing

Google publishes both definitions and they are not interchangeable. Search Console counts a click as "the number of times a user clicked your site from Google Search results". The Business Profile report counts website clicks as "the number of times people clicked on the website link on your Business Profile", calls as "the number of times a customer clicked on the call button on your Business Profile", and directions as "how many people asked for directions to your business".

Read those again and the shape of the problem appears. A call button and a directions request never involve your website, so no amount of website analytics will ever show them to you. They are, for a plumber or a notary, the actual event you are paying for.

That last row is the trap I nearly fell into myself. Google states that a profile view counts "unique visitors who viewed your profile" and that "a person can only be counted once a day". Search Console impressions have no such rule. So comparing "1,154 profile views" against "33,563 impressions" is comparing two different units, and any percentage you build from the pair is fiction. I only compare the outcome columns below — clicks, calls, directions — because those are raw event counts on both sides.

What the five businesses actually did

Five businesses whose Google listing I have access to, all read over the identical window of 9 January to 6 September 2026. Two Toronto-area service businesses, a paralegal, a notary and an art gallery.

The notary is the row worth staring at. That website earned 148 clicks from all of Google web search in eight months, which is respectable. In the same period the listing sent 277 people to the site and — this is the number that matters when you sell a fifteen-minute appointment — 313 people pressed the call button. If that business judged its marketing by Search Console alone, it would be looking at less than a fifth of what Google actually delivered.

My own row is the most embarrassing and the most instructive. 33,563 impressions is a lot of visibility. It converted into 23 clicks, a rate of 0.07%. The listing, from a fraction of that exposure, did nearly twice as well and threw in 367 direction requests. I sell websites for a living and my own website is the weakest channel I own. That is not false modesty, it is what the export says.

How I know these aren't the same clicks counted twice

This was my first objection to my own data, and it has a clean answer. If Search Console were quietly including the clicks that arrive via the Business Profile, then its total would have to be at least as large as the profile's subtotal. It isn't.

My profile reported 44 website clicks in a window where Search Console reported 23 clicks for the entire site, from every query. A total cannot be smaller than one of its parts. The art gallery shows the same thing — 5 profile clicks against 1 total — and so does the construction company, 2 against 0. Three independent cases, all pointing the same way: these are different clicks, and if you only read one dashboard you are missing the other set entirely.

So where did 50,000 impressions go?

Into searches that were never going to click. I widened the same export to every property I can read — ten sites, 13,847 real searches after stripping out site: operator queries, which are people auditing a site rather than shopping. Those searches produced 335,070 impressions and 985 clicks. 97.6% of the queries never produced a single click.

The pattern inside that is the useful bit, and it is not the one the SEO industry sells.

Short generic terms — the ones an agency puts in a proposal because they have impressive search volume — accounted for 32% of every impression these sites earned and returned a click rate of 0.21%. The business's own name did 8.73%. Thirty-four brand searches out of 13,847, less than a quarter of one per cent of the queries, delivered 18.6% of all the clicks.

Two honest corrections to the folk wisdom while I am here. Longer searches did better, but only mildly: 0.21% for one or two words against 0.38% for six to eight. And phrasing something as a question did almost nothing on its own — 0.27% versus 0.23%. What actually earns is specificity, not word count and not a question mark. "Near me" searches were the local exception at 1.18%, five times the non-brand average.

There is one more explanation for those zeroes, and it is the reason this article exists. "Roofing company" sat at an average position of 1.2 with 1,072 impressions and no clicks. Position one, nobody clicking, is normally read as a broken title. On a local search it usually isn't: the map pack sits above the blue links and the customer taps a business card there instead. Search Console cannot see that click. The profile report can. I have written separately about the numbers in Search Console that look like problems and aren't, because that one costs people real money in pointless rewrites.

What this does not mean

I would rather say this myself than have it said back to me.

It does not mean skip the website. Every one of those 415 profile clicks was somebody tapping a link and arriving on a real page. A listing with nowhere to send people is a phone number with a photo. And a website does work the profile structurally cannot: it answers the eleven questions a customer has before they will phone anyone, which is the whole argument for what actually belongs on a service-business homepage.

It is five businesses, not an industry study. One agency, mostly Ontario, mostly local services, all with profiles I look after. A national e-commerce brand would produce the opposite table. The pattern was consistent across all five, which is why I am publishing it, but consistency across five is a hint and not a law.

The websites here are young. Most of the pages in that impression count were published in the last few months, and new pages take about two days to appear and around three weeks to earn a first click. Some of this gap will close on its own. It will not close to zero — the notary's calls are not waiting on anybody's crawl budget.

What I do with this on a Monday morning

The measurement changed my own routine, so here is the version I follow now.

And the habit I dropped: judging a local business's marketing by its website traffic. For the notary, the website is a supporting document. The listing is the shop window, and I now spend the monthly Google Maps work accordingly — categories, photos, hours and reviews first, blog posts second. For the gallery it is the reverse, because people who buy paintings want to look at them for a while before they walk in.

If you are choosing where to put a limited budget: get the listing complete and verified before you spend anything on content. It is free, it took me an afternoon per business, and on this sample it out-performed every website it was attached to. Then build the site the listing points at properly, which is what I charge for and, given the table above, the part I should probably talk about less than the free part.

Frequently asked questions

Sources, both read in full on 8 September 2026: Google — Search Console performance report metrics · Google — Business Profile performance metrics. Everything numeric is my own measurement, not an industry study: Search Console query data for ten verified properties and Business Profile performance data for five profiles I manage, both covering 9 January to 6 September 2026, pulled read-only on 8 September 2026. Ninety-nine site: and inurl: operator queries (1,938 impressions, 0 clicks) were excluded as audits rather than customers. Two further properties could not be read and were left out entirely rather than counted as zeroes.