- Across 44 GTA service businesses that already rank, the average site health score was 76 out of 100, median 79.5, range 35 to 96. Only 4 of 44 cleared 90.
- Every single site had a phone number and an address on the homepage. Zero failures on the human basics. All the damage was in the machine-readable layer.
- 34% do not link their Google Business Profile from the homepage. I first measured this at 61% and my own checker was wrong, which is covered below.
- Schema was the weakest category by a distance: 62 average against 77 to 83 for everything else the scanner measures.
- The most-missed signal of all, FAQPage schema at 82%, is the one I would now tell you to skip. Google removed FAQ rich results in May 2026 and a controlled study found schema does not lift AI citations.
- Two of my own numbers overstate the problem: the scan reads homepages, and FAQ and breadcrumb schema belong on interior pages. Treat those two rates as an upper bound, not a verdict.
- These businesses already show up in search, so every failure rate here is a floor. The wider market is worse.
Most articles telling you your website has problems were written by someone who never looked at your website. So I looked at 44 of them. Real Toronto-area service businesses, six trades, all of them found the way a customer would find them. Average score: 76 out of 100.
Here is the part that surprised me. Not one of the 44 was missing a phone number or an address on the homepage. The visible, human stuff was handled. What was broken sat underneath, in the layer that decides whether a machine can understand the page before a person ever sees it. And the single most common failure turned out to be a signal I no longer think you should bother fixing, which I will get to.
How I ran this scan
I used the same engine that powers the free audit tool on this site, run as a batch instead of one URL at a time. That matters for a reason worth stating plainly: you can reproduce any number in this article yourself, on your own site, in about thirty seconds.
- Sample
- 44 GTA service businesses across six trades: HVAC (8), plumbing (7), electrical (9), landscaping (7), cleaning (8), kids' party entertainment (5). Found through customer-style searches like "furnace repair Etobicoke" and "plumber Scarborough". Directories and national franchises excluded.
- What was scanned
- Homepage HTML plus robots.txt and sitemap, on 20 July 2026. Roughly 25 signals across six categories, scored out of 100. No interior pages, no field performance data.
- The bias, stated openly
- These businesses already show up in search. That is how I found them. So this is not the market average, it is the visible end of the market, and every failure rate below is a floor rather than a ceiling.
- Publication rule
- Aggregate numbers only. I am not naming or shaming individual small businesses, and no site is identified anywhere in this article.
What does the average local business website score?
Seventy-six out of a hundred, with a median of 79.5 and a spread from 35 to 96. Four sites out of 44 scored 90 or better. Twelve scored below 70, and five of those were below 60.
The spread is what I keep looking at. These businesses compete for the same customers in the same city, and 61 points separate the best site from the worst. Not 61 points of design taste. Sixty-one points of whether the page is legible to the systems that decide who gets shown.
Average site health score, by trade
Out of 100, across 44 GTA service businesses. Switch between the whole sample and each trade.
Where do the failures actually cluster?
In the schema and structured-data layer, and nowhere near the visible design. The scanner groups its checks into six categories, and five of them averaged between 77 and 83. Schema averaged 62.
That is the whole story of this dataset in one comparison. These sites load, they work on a phone, they say who they are, they list a number. Then they stop, right at the point where a page has to explain itself to something that is not a human being.
Average score by category, and where mine land
Six categories, scored out of 100. Toggle to see how RMCM's own two client builds compare on the same checks.
Here is the failure table in full. I have added a column that the raw data does not give you, which is whether the signal still earns anything in August 2026, because two of the most-missed items on this list stopped paying out earlier this year.
| Signal | Failing | Does it still pay? |
|---|---|---|
| No FAQPage schema | 82% | Not for Google. Rich result removed May 2026 |
| No BreadcrumbList schema | 66% | On interior pages, yes. On a homepage, no |
| Title tag outside 48–65 characters | 55% | Yes. It is your headline in search |
| Image alt text below 85% coverage | 45% | Yes. Accessibility and image search |
| No preload or preconnect hints | 41% | Minor. Speed nicety, not a fix on its own |
| No Google Business Profile link | 34% | Yes. Connects site to your map listing |
| Wrong H1 count (zero, or more than one) | 32% | Yes. The clearest structure signal you have |
| No og:image (blank link previews) | 32% | Yes. Every share and text message |
| No LocalBusiness-type schema | 25% | Yes. Corroborates your business details |
| No robots.txt | 14% | Minor on its own |
| No structured data of any kind | 11% | Yes. Five sites are invisible as an entity |
| No meta description | 11% | Yes. Google writes one for you otherwise |
| No canonical tag | 7% | Yes, quietly. Duplicate-content insurance |
| No XML sitemap | 2% | Yes, and almost everyone has one |
Failure rate by signal, ranked
Percentage of the 44 sites failing each check. Filter to the signals that still earn something in 2026.
The finding I have to walk back
The biggest number in my own dataset is the one I am least excited about. 82% of these sites have no FAQPage schema, and when I started this scan I would have called that the headline. It is not, because Google killed the payoff ten weeks before I ran it.
On 7 May 2026 Google stopped showing FAQ rich results in search, for every site, with no phased rollout and no replacement. By 15 June it had deleted the documentation entirely, and Search Console API support for the feature ends this month. FAQ is no longer listed in Google's structured data gallery at all. The markup is still valid and it will not hurt you. It just does not do anything in Google any more.
The obvious comeback is that FAQ schema now helps AI answer engines instead. I believed that too, until Ahrefs tested it properly. They tracked 1,885 pages that added JSON-LD schema against 4,000 matched control pages and measured citations before and after. AI Overviews went down 4.6%. AI Mode moved 2.4% and ChatGPT 2.2%, both indistinguishable from zero. Adding schema did not make pages get cited more.
One caveat matters, and Ahrefs raised it themselves: every page they studied already had over 100 AI Overview citations. So the study shows schema will not push an already-visible page higher. It does not prove schema is useless for a business nobody has heard of, and that is a different question nobody has answered yet.
So my honest position, having generated the stat myself: skip FAQPage markup, and write the actual questions and answers on the page instead. The retrieval tests in that same study found AI systems pulled visible HTML and ignored hidden markup. Your customers ask you the same six questions every week. Put those on the page in plain words. That is worth more than a JSON block that Google stopped reading in May. There is more on which schema types still earn their place in our guide to local business schema.
What this scan gets wrong
Two of the numbers above overstate the problem, and you should know that before you quote them at anyone. The scan reads homepages. FAQPage and BreadcrumbList schema do not belong on a homepage.
I found this by running the scanner at my own work. One of the sites I built scores badly on schema here, and when I went looking, the schema was all there: FAQPage on its /faq page, BreadcrumbList on 56 of its 62 pages. The homepage carries neither, deliberately. A breadcrumb marks a page's position in the site hierarchy, and the homepage is the top of the hierarchy, so there is nothing to mark. Putting one there would be noise.
Which means the 66% breadcrumb figure is not 66% of businesses making a mistake. It is 66% of homepages doing something reasonable, counted as a failure by a check that should have been scoped to interior pages. The 82% FAQ number has the same flaw on top of the deprecation problem. Treat both as an upper bound on a homepage, not a verdict on a website.
Then I found a second problem, and this one was a plain bug in my own checker. It looked for a Google Business Profile link by searching for a maps.google URL or the text place_id. Plenty of businesses link their profile a different way: the "leave us a review" link Google gives you is search.google.com/local/writereview?placeid=, with no underscore in placeid, so my check sailed straight past it. Short links like g.page and maps.app.goo.gl were invisible to it too.
That one hurt, because it was the headline recommendation. I originally had this at 61% of businesses not linking their profile. I re-checked all 44 homepages on 6 August with the pattern corrected, and the real number is 34%, re-measured on 6 August. Eleven businesses were linking their profile the whole time and my tool called them failures. The fix is now live in the audit tool, and the table above shows the corrected figure.
The rest of the table holds up, because those checks belong on a homepage and none of them depend on the two patterns I got wrong: the H1, the title tag, the alt text, the business schema. That is where I would still put your afternoon.
Want to know where your own site lands?
Run the same scanner I used for this study. Free, about 30 seconds, no call required.
SCAN MY WEBSITEWhich trades are ahead, and which are behind?
Cleaning companies had the best websites in this sample at an average of 84, and HVAC had the worst at 68. That ordering held even though the trades are competing for similar budgets and similar customers, which tells you the gap is habit rather than money.
Cleaning was the only trade where every single site had proper business schema, and it produced the highest score in the whole study at 96. Plumbing had the strongest ceiling, with three of seven sites scoring between 91 and 95, which fits a trade where an emergency call is worth enough to justify a real website. HVAC had both the lowest average and the lowest single score in the sample, a 35.
The party entertainment sites had the most broken structure of any trade: four of the five had a wrong H1 count, and four of five had no link to their Google Business Profile. That is a small sample and I would not lean hard on five sites, but it matches what I see in that niche generally.
Score range by trade
The bar spans each trade's lowest and highest scoring site. Switch to see the average alone.
What should you check on your own site today?
Three things, in this order, and you can do all of them in an afternoon without a developer.
First, count your H1 tags. Right-click your homepage, choose View Page Source, and search for <h1. You want exactly one. In this sample 32% got that wrong: five sites had none at all and nine had more than one. The H1 is the single clearest statement of what a page is about, and having zero or four is a way of declining to make it.
Second, link your Google Business Profile from your homepage. A third of these businesses had not. Add your Maps listing or your review link to the footer. It connects the two things a local buyer checks, and it takes five minutes.
Third, read your own title tag in a search result. Google your business name and look at the blue link. Is it under 65 characters, does it say what you do, and does it name your city? Over half of these sites had a title outside the useful range. It is the first thing anyone reads about you, and rewriting it costs you one coffee.
What I would not spend the afternoon on: chasing a perfect score. Only 4 of 44 businesses cleared 90 and plenty of them are doing fine. Most of the remaining points come from signals with small or fading returns, and one of them, as covered above, is now worth nothing at all in Google. Fix the structural failures and go back to running your business. If you want the fuller checklist, we cover it in what every local business website needs and what happens when Google cannot read your website.
For context on why any of this matters more than it did two years ago: 45% of consumers now use AI tools to find local businesses, up from 6% the year before, which makes AI the third most common discovery channel behind Google and Facebook (BrightLocal, 2026, 1,002 US adults). More systems are reading your page and fewer humans are browsing to it. A page a machine cannot parse quietly drops out of more conversations than it used to.
Frequently asked questions
What score should a local business website actually aim for?
Does a higher audit score mean I will rank higher?
My site scored well but the phone still isn't ringing. What now?
Do I have to fix all of these at once?
Can I run this same scan on my own website?
So where should you start?
Scan your own site before you believe anything about it. You have an opinion about your website, and this study exists because opinions about websites are usually wrong in a specific direction: owners assume the problem is how it looks, and it almost never is. Every one of these 44 businesses got the visible part right and most of them stopped there.
Then fix the boring three. Get to exactly one H1, rewrite the title tag, link your Google Business Profile. None of that is interesting and none of it will impress anyone, which is exactly why a third to a half of your competitors have not done it. The work that separates the top of this sample from the bottom is not clever. It is just unglamorous enough that people skip it.
I will be straight about my own numbers too. Run this scanner at the two sites I built and it returns 85 and 87, above every trade average here but not perfect, and part of what it marks them down for is the breadcrumb and FAQ schema sitting on their interior pages where it belongs. The 90-plus figures in our case studies come from a fuller manual review, which is a different instrument reading a whole site rather than one page. You should be suspicious of anyone who only ever shows you the flattering number, including me, which is why both are in this article. If you want the same scan on your own site, the free audit runs these checks, and if the result raises more questions than it answers, that is a good reason to talk.
