An accounting firm website needs three things before anything else: it loads quickly on a phone, every visitor can use it, and nothing stops search engines and AI tools from finding and reading its pages. In September 2026 I tested 824 Ontario accounting firm websites on those three. Of the 741 I could measure on all three, 6 passed every one, and two of those are single-page business cards.
Content matters too: a page for each service, the people behind it and a clear way to get in touch. This article covers both, starting with the three basics, because that’s where the data is. You don’t need to be technical to follow it. Where a term needs explaining, I explain it.
What the benchmark found
I found the firms through Google listings in 100 Ontario cities and towns and tested each homepage between September 16 and 23, 2026. Every figure below, with its base and the full method, is in the 2026 benchmark report.
| Check | What it means for a visitor | Sites that met the standard |
|---|---|---|
| Speed | The page appears quickly on a phone, not just on an office computer | 52 of 791 (6.6%) |
| Accessibility | The text is readable, and nothing a free checker can detect blocks people with low vision or who use a screen reader | 63 of 791 (8.0%) |
| Search | Nothing stops search engines and AI tools from finding and reading the site’s pages | 633 of 751 (84.3%) |
| All three | 6 of 741 |
Share of sites that meet each standard
Horizontal bars on a 0 to 100% scale. Speed: 6.6%, 52 of 791. Accessibility: 8.0%, 63 of 791. Search: 84.3%, 633 of 751. All three: 0.8%, 6 of 741.
The totals differ because not every site could be measured on every check.
Most of these sites look fine to the people who own them. They load on the office computer, the phone number is there, and clients who already know the firm find it. That’s the trouble. What these sites get wrong doesn’t show up for the owner. It shows up for a stranger: the prospect on a phone staring at a half-loaded page, the older client who can’t read the pale grey text, the search engine sent to pages that no longer exist. None of them calls to tell you. They go to the next firm on the list.
No report will show you that, so it helps to put a number on it. Use your own figures, but say a typical small-business client pays you $3,000 a year and stays for eight years. That’s $24,000 from one client. A site that quietly turns away one of those a year is a $24,000 problem nobody sees.
The standard I measured against is mine, the one I hold my own builds to. Every part of it is measured with free public tools, Google’s speed test and WebAIM’s accessibility checker, so anyone can run the same checks.
Does an accounting firm website need to be fast?
Yes. A prospect who searches for an accountant and taps your listing is one tap away from the next firm in the results. If your page is still blank, the back button is right there. Google also says its ranking systems use loading measures taken from real visitors, though relevance always comes first.
Google’s free speed test, PageSpeed Insights, loads a page the way a typical phone would and scores it out of 100. Google calls 90 or more good. My standard is 90 or more on a phone and 95 or more on a computer.
The median phone score in the benchmark was 62, which means half the sites scored lower than that. Only 62 of the 791 sites scored 90 or more on a phone.
Two things make speed harder to judge than it looks:
- The score moves. Run the test twice and you can get two different numbers. On 366 of 756 sites, the score moved more than 10 points between runs on the same day. One site scored 93, then 83, then 41 on a computer. The page itself was the same size each time. What changed was how long add-ons from other companies kept the browser busy. If you test your own site, run it three times and take the middle score.
Same page, same day: 93, then 83, then 41
Three vertical bars of one site's desktop score, in run order, with a dashed line at 95, the desktop standard. First run: 93, add-ons 0.7 seconds. Second run: 83, add-ons 1.0 seconds. Third run: 41, add-ons 2.1 seconds.
- The test is a simulation, not your visitors. Google also collects loading times from real people using its Chrome browser, but only for sites with enough visitors. It had enough for 81 of the 824 firms. For the rest, the test is the only measurement there is, and a test score isn’t what any one visitor experiences.
What slows these sites down shows up in the test’s own findings. Four each appear on more than 80% of the 791 sites:
- The main content is slow to appear. The heading or main image at the top of the page takes a long time to show up. The visitor’s first impression of your firm is a blank or half-built page.
- Heavy code from other companies. Chat windows, booking tools, maps, video players and tracking tools each load their own code, and the visitor’s phone has to work through all of it. The page can look ready while taps do nothing yet. The add-ons installed to win enquiries can be what’s slowing the page down.
- Images larger than they need to be. A photo saved at full camera size, shown in a space the size of a postcard. The visitor’s phone downloads all of it, which hurts most on a weak mobile signal.
- Files that hold up the page. Some files have to finish loading before the browser will show anything at all. Until they do, the visitor sees a white screen.
It’s rarely one big problem. It’s several at once.
Does an accounting firm website need to be accessible?
An accessible website is one people with a disability can use: someone with low vision who enlarges the text, someone who can’t use a mouse and moves through the page with the keyboard, someone who is blind and uses a screen reader, software that reads the page aloud. That’s not a small group. Statistics Canada’s 2022 Canadian Survey on Disability found that 27% of Canadians aged 15 and older have a disability that limits their daily activities, and 40% of those 65 and older. Some of them are business owners choosing an accountant. Some are your own clients, getting older. And the fixes help everyone: text that’s easy to read is easier for anyone on a phone in bright daylight.
How I tested it. I ran each homepage through WAVE, a free checker from WebAIM. WAVE reads the page’s code and flags problems a machine can detect, such as text too faint against its background, form boxes with no name, and images with no description. I counted two kinds of WAVE result: errors and contrast errors. My standard is zero of each. A checker only finds what’s measurable. It can’t tell whether the page actually makes sense to someone using it, so passing WAVE is a starting point, not proof a site is accessible.
Two failures were common:
- Text that’s too faint: 613 of 791 sites (77.5%). Usually light grey text on a white background. It looks clean in a design file. For a visitor with low vision, or anyone reading on a phone outdoors, it means squinting or giving up.
- A form field with no label: 227 of 791 sites (28.7%). A form field is any box a visitor types into: a search box, a newsletter sign-up, an enquiry form. The label is the name attached to that box in the page’s code. Without it, a screen reader announces the box with no name, so a blind visitor can’t tell what to type. If it’s your enquiry form, that visitor can’t reach you through it.
Many sites are close. 195 of the 791 are within five issues of the standard.
195 sites are within five issues of the bar
Five vertical bars of WAVE errors plus contrast failures per homepage, 791 sites. None issues: 63 sites (meets the standard). 1 to 5 issues: 195 sites (within five issues of the bar). 6 to 20 issues: 321 sites. 21 to 50 issues: 167 sites. More than 50 issues: 45 sites.
Try this yourself. Open your homepage on a computer, click in the address bar, then press the Tab key over and over. Each press should move a visible outline to the next link, button or form box, so you can always see where you are. If the outline disappears, or you can’t reach the menu or your enquiry form, a visitor who can’t use a mouse can’t get there either.
The law, stated accurately: in Ontario, the Accessibility for Ontarians with Disabilities Act (AODA) requires the public websites of private organizations with 50 or more employees to meet WCAG 2.0 Level AA, the international standard for accessible websites. Many accounting firms are smaller than that, so the legal requirement may not apply to them. The visitors it’s meant to protect visit small firms too.
A checker finds what can be measured. Whether the whole site works with only a keyboard, and whether the descriptions of images make sense, needs a person to test.
Does it need HTTPS?
Yes. HTTPS is the secure version of a web address: the padlock in the browser, and the “s” in https://. When a page doesn’t use it, Chrome shows “Not secure” next to the address. For an accounting firm, that warning sits next to your name at the moment a prospect decides whether to trust you with their financial details.
In the benchmark, 49 of the 734 sites where both connection checks could be observed send visitors to a page without it, about 1 in 15. 30 of them have a secure version but don’t send visitors to it automatically. The other 19 have no secure version at all.
Check yours: type your web address into a browser starting with http:// instead of https://. If it doesn’t switch to https:// by itself, that’s the first thing to fix.
Can search engines find your pages?
Search engines find a site’s pages two ways: by following links, and by reading the site’s sitemap, a list of its pages that the site publishes for them. The check also looks for settings that tell search engines to stay away.
This is the check most sites passed: 633 of 751. Of the 118 that failed, the most common reason was a sitemap that lists pages that no longer load (43 sites). 39 show their text only after code runs (Google waits for that code; many AI tools don’t), and 25 had a sitemap with nothing in it. A broken sitemap won’t hide a site on its own, but it sends search engines to pages that aren’t there, and it’s a sign nobody has looked in a while.
Can AI tools read your site?
AI assistants and AI search tools answer questions about businesses by reading websites, the same way search engines do: they read the text on the page. When someone asks one to recommend an accountant nearby, your own website is one of the places it draws on, and it can only use what it can read. Two things can get in the way:
- the site blocks programs named by AI companies, in a small file called robots.txt (some of these only collect training data, which doesn’t stop an AI tool reading the page)
- the page’s text only appears after code runs, which some of these programs don’t wait for
Both were rare. 3 of 751 sites block a named AI program, and 39 show their text only after code runs. On 709 of the 751, neither problem shows up. That doesn’t prove an AI tool uses a site, only that nothing on the page stops one.
You may hear that AI tools need structured data, a summary of the firm hidden in the page’s code. Google says its AI features don’t need any special code. Microsoft says its Copilot assistant uses it, and for other AI tools there’s little evidence either way. It’s still worth having, because Google and Bing use it to identify a business, but it isn’t what decides whether an AI tool can read your site. 313 of the 751 sites have none.
Check yours: open yourfirm.ca/robots.txt in a browser, using your own address. Look for a line naming GPTBot, ClaudeBot or PerplexityBot followed by Disallow: /. If you find one, that AI tool is being turned away. (A Disallow: line with nothing after it lets the tool in.) Also look for User-agent: * followed by Disallow: /. That turns away every search engine and AI tool. Four sites in the benchmark had it. That may be deliberate, so ask whoever runs your site.
What should the content include?
The benchmark tested three technical checks on homepages. It didn’t grade content. This list comes from 22 years of building websites, not from the benchmark:
- A page for each service you want to be hired for. Personal tax, corporate tax, bookkeeping, payroll, advisory. A page each gives every service its own place to be found and described.
- Who you work with. Industries, business sizes, personal or corporate clients. “We serve individuals and businesses” tells a prospect nothing.
- The people. Real names, real photos, designations, and what each person handles.
- Where you work. Your city, and whether you take clients from across Ontario.
- How to start. A phone number visitors can tap to call, a short form, and one sentence on what happens after they get in touch.
- How fees work. You don’t have to publish prices. Saying whether you charge a fixed fee, hourly or monthly answers the question every prospect has.
- A clear link to your client portal or secure upload, if you have one, where clients can find it in April.
- A privacy policy. Canada’s federal privacy law (PIPEDA) governs how businesses handle personal information, and a firm that collects financial details should say plainly what it does with it. If you send a newsletter, you also need consent under Canada’s anti-spam law (CASL).
What do the fastest sites have in common?
Less than you’d hope. 62 sites scored 90 or more on a phone, and 21 of them are business cards: single-page sites with little behind them. A page with little on it is fast.
The other 41 are spread across places of every size, 36 towns and a dozen different website systems, and nothing obvious explains them. Being fast didn’t carry over to the other checks either: 7 of the 41 meet the accessibility standard, and 4 pass all three.
Of the 741 sites measured on all three checks, 6 passed everything. Two are business cards. The standard is reachable, and very few sites reach it.
Questions to ask whoever runs your website
You don’t have to run these checks yourself. Send these five questions to whoever looks after your site. Each has a free tool behind it, so they can show you the answer.
- Speed: “What does our homepage score on a phone and on a computer in PageSpeed Insights? Run it three times and give me the middle score.” My standard is 90 on a phone and 95 on a computer.
- Accessibility: “How many errors and contrast errors does WAVE find on our homepage?” The standard is zero of each.
- Security: “If someone types our address with
http://instead ofhttps://, does it switch over automatically?” - Search: “Does every page listed in our sitemap load?”
- AI tools: “Is anything on our site blocking AI tools like ChatGPT from reading it?”
If you’d rather look yourself, each tool is free and needs no account: PageSpeed Insights and WAVE. The full list, with the exact standard for each check, is in the benchmark report.
Then ask again in three months. A software update, a new chat window or a redesigned form can undo a passing score without anyone noticing. That’s why the bar I hold sites to is re-checked every 90 days, not once at launch.
What this doesn’t cover
- Homepages only. The other pages on each site weren’t tested.
- A September 2026 snapshot. Sites and standards both change.
- Automated checks only. Anyone can repeat them, but they’re not a full accessibility audit, and nothing here is legal advice.
- No business outcomes. The benchmark measured the websites, not enquiries or revenue. The $24,000 example is arithmetic on sample figures, not a finding.
The full method and every base are in the benchmark report. If you’d rather someone else ran these checks, here’s how I work with accounting firms.
Updates
9 November 2026. Seven sites were counted as failing search because pages listed in their sitemap returned errors. Those errors were the sites’ servers refusing my scanner (too many requests, or access denied), not pages a visitor couldn’t open, so those sites now pass search. Search moves from 626 to 633 of 751. Sites passing all three, the scan dates and every other figure are unchanged.
