{
"title": "The Ontario Practice Website Index",
"edition": "2026-10b",
"collectedFrom": "2026-10-08T19:27:07.960Z",
"collectedTo": "2026-10-18T02:40:06.353Z",
"n": 187,
"nBySector": {
"law": 65,
"clinic": 45,
"accounting": 77
},
"methodology": [
"2026-10-08 — Sampling frame. A convenience sample, not a random one: practices listed with a website in the public member directories of four Ontario business associations — London Chamber of Commerce, Ottawa Board of Trade, Barrie Chamber of Commerce (all GrowthZone directories) and the Hamilton Chamber of Commerce (WordPress directory). Listing pages were filtered by sector keywords on the listing name and description, de-duplicated by domain, and national or multinational networks were removed. Each sector was capped at 30 sites, taken in SHA-1(domain) order, which spreads the cap across cities without hand-picking.",
"2026-10-08 — Politeness. Every directory page and every firm host was checked against robots.txt (our token \"UXProPracticeIndex\", else \"*\") before any request; disallowed hosts were dropped, and a robots.txt that failed with a 5xx or network error was treated as \"disallow all\". Requests were limited to about one per second per host and identified as \"UXProPracticeIndex/1.0 (+https://uxpro.ca/research/ontario-practice-website-index)\". Per site we fetched only the homepage, /llms.txt and at most one contact or booking page. No form was ever submitted.",
"2026-10-08 — Verification. A site stayed in the sample only if its own homepage text reads as a practice in the sector (e.g. lawyer/law firm/barrister; patient/dental/clinic/physiotherapy; accountant/CPA/bookkeeping/tax). Ontario location comes from the directory listing's address (Ontario only) and is cross-checked against an Ontario address cue on the site; rows without that cue say so in notes. Bot challenges, unreachable sites and non-practices were excluded; the exclusion list is kept with the scripts.",
"2026-10-08 — Performance. Each homepage was run through Lighthouse 13.5 (npx), performance category only, mobile form factor, default simulated throttling, attached to a fresh headless Chrome 154 profile per site. Page weight is total-byte-weight; LCP is largest-contentful-paint. Median Lighthouse run time was 14 s per site. Lighthouse's LCP is a simulation from one run, so single-site values are noisy; the medians are the point.",
"2026-10-08 — Markup and accessibility. In the same Chrome, a separate tab emulated a 390x844 phone (DPR 2, touch, mobile UA), loaded the homepage, scrolled it top to bottom, and read the rendered DOM for tap-to-call links, the viewport meta, JSON-LD and microdata types, , ![]()
alt coverage, the price heuristic, forms, and text contrast. The contrast check reuses the alpha-compositing and WCAG luminance logic of the studio's own contrast gate, with WCAG 2.1's standard large-text thresholds.",
"2026-10-08 — Statistics. Medians and shares are computed over the sites where a metric could be measured; each summary metric reports its denominator (n) and the number of nulls.",
"2026-10-18 — Edition 2 frame. Same kind of source and the same rules: the public member directories of the Mississauga Board of Trade, Markham Board of Trade, Richmond Hill Board of Trade and Milton Chamber of Commerce, filtered with the same sector keywords, de-duplicated against every edition-1 candidate, national and multinational networks removed (now also PKF and Kreston member firms). Each new directory was capped at 30 sites per sector in SHA-1(domain) order. The 17 edition-1 candidates held back only by edition 1's cap were added. Four directories near our base could not be used and are listed with the reason: Greater Kitchener Waterloo Chamber of Commerce, Cambridge Chamber of Commerce, Guelph Chamber of Commerce, Toronto Region Board of Trade.",
"2026-10-18 — Edition 2 measurement. The same scripts (scripts/practice-index/measure.mjs, unchanged audit code), Lighthouse 13.5.0 with the same flags (performance only, mobile form factor, default simulated throttling), headless Chrome 155, one Chrome at a time through a machine-wide lock; two sites were in flight only for the plain-HTTP steps. Edition-1 rows were not re-measured. Every row carries its own measuredAt (UTC) and an addedIn edition, so either edition can be analysed on its own."
],
"limitations": [
"Convenience sample drawn from the member directories of 8 Ontario chambers of commerce and boards of trade (London, Ottawa, Barrie, Hamilton, Mississauga, Markham, Richmond Hill, Milton); members of business associations may differ from practices that are not members. Results do not generalise to all Ontario practices.",
"Kitchener, Waterloo, Cambridge, Guelph and Toronto have no rows of their own (or too few to report): the Greater Kitchener Waterloo and Cambridge chambers serve their directories through a widget their robots.txt disallows, Guelph publishes its directory only as a flip-book PDF, and the Toronto Region Board of Trade has no public member directory. We did not substitute another source.",
"The Law Society of Ontario, RCDSO/ODA and CPA Ontario finders were not used: they are search-driven applications rather than crawlable listings with website links, so they could not supply a frame politely within the time budget.",
"Two collection windows from a single machine and network: edition-1 rows on 2026-10-08 (Chrome 154), edition-2 rows from 2026-10-18 (UTC) with Chrome 155. One Lighthouse run per site; simulated LCP varies run to run and with machine load, so compare medians, not single sites, and treat small differences between editions as noise.",
"City is the city of the directory listing's address. A GTA board's members may list an office in a neighbouring municipality (Thornhill, Unionville, Woodbridge), which the dataset records under its own name and does not fold into a larger city.",
"Sites behind bot protection that served a challenge to headless Chrome were excluded, which biases the sample slightly away from heavily protected sites.",
"Price, sensitive-form and sector checks are text heuristics. The price rule is deliberately conservative (it will miss prices written without '$' or on inner pages). The sensitive-form flag reads field labels only and cannot see forms embedded in third-party iframes (e.g. online-booking widgets).",
"/llms.txt presence mostly reflects SEO-plugin defaults: of the files spot-checked, most were generated automatically by WordPress SEO plugins (Yoast SEO, All in One SEO) rather than written by the practice.",
"Contrast is checked on text over solid colours only; text over images or gradients is skipped, so failures are a floor, not a ceiling.",
"Clinic includes dental, medical, physiotherapy, chiropractic, foot-care, hearing, mental-health and similar private practices found in the 'health' categories; the mix is not balanced across professions."
],
"fields": [
{
"key": "domain",
"label": "Domain",
"unit": null,
"description": "Registrable hostname of the practice website (www. stripped). No person or firm names are stored."
},
{
"key": "sector",
"label": "Sector",
"unit": null,
"description": "law | clinic | accounting, from the directory category and the firm's own homepage text."
},
{
"key": "city",
"label": "City",
"unit": null,
"description": "City from the directory listing address, where given."
},
{
"key": "https",
"label": "HTTPS",
"unit": "boolean",
"description": "https:/// (or https://www./) loads with a 2xx and the final URL is https."
},
{
"key": "httpRedirectsToHttps",
"label": "HTTP redirects to HTTPS",
"unit": "boolean",
"description": "Requesting http:/// ends, after redirects, on an https URL with a 2xx."
},
{
"key": "viewport",
"label": "Viewport meta",
"unit": "boolean",
"description": "Rendered DOM has whose content sets width=."
},
{
"key": "tapToCall",
"label": "Tap-to-call",
"unit": "boolean",
"description": "Rendered mobile homepage has at least one ."
},
{
"key": "schemaTypes",
"label": "Schema types",
"unit": "list",
"description": "@type values from all JSON-LD blocks (recursively, incl. @graph) plus microdata itemtype names, on the rendered homepage."
},
{
"key": "llmsTxt",
"label": "/llms.txt",
"unit": "boolean",
"description": "GET /llms.txt returns 200, a non-HTML content type, a non-empty body that does not start with '<', and was not redirected elsewhere (rules out soft-404s)."
},
{
"key": "pricesVisible",
"label": "Prices on homepage",
"unit": "boolean",
"description": "Conservative heuristic: a $ amount between $1 and $9,999 in visible homepage text with a fee word (fee, flat fee/rate, price, pricing, cost, rate, starting at, per hour/session/visit/month/return, consultation, assessment, package) within 60 characters, and no brag/limit word (recovered, settlement, award, million, limit, up to, claim, value, save, worth, grant, rebate, credit, refund, compensation, damages, deposit, gift, prize, win/won) in the same window."
},
{
"key": "formFound",
"label": "Contact form found",
"unit": "boolean",
"description": "A