They measure youbefore youarrive

A1. The research question

When a homeowner has a leak, a dead furnace, a storm-damaged roof, a tripped panel or a broken washer and looks up an independent contractor online, usually on a phone, can they find out how to reach it and when it is open, whether it works in their area, whether it comes in an emergency, roughly what a visit costs, whether it is licensed, whether they can book without calling, whether there is financing, whether the work is guaranteed and whether neighbours vouch for it?

Population. Independently owned US home-service contractors with one to three locations and their own website, in four groups: plumbing and HVAC, electrical, roofing and exteriors, and appliance repair and handyman services. These are the trades businesses a small or mid-size web agency can realistically win as clients.

A2. Who was studied

The sample was built from local search listings, city by city, then screened so that every site in it is one an independent agency could realistically win.

Source	What it contributes	Sites
Local search, plumbing and HVAC	Independent plumbing, heating and air-conditioning contractors found in local search results, city by city	77
Local search, electrical	Independent electrical contractors and electricians found in local search results, city by city	42
Local search, roofing and exteriors	Independent roofing, siding and exterior contractors found in local search results, city by city	42
Local search, appliance repair and handyman	Independent in-home appliance repair services found in local search results, city by city; in two cities where appliance repair ran short, independent handyman and home-repair services	37

From list to sample	Sites
Candidates listed	212
Removed at screening: the site refused automated visitors (25)	−25
Removed at screening: no response, or no page served to automated visitors (7)	−7
Removed at screening: parked or placeholder (2)	−2
Removed at screening by hand: offices in two states, no address, a generic keyword domain (3)	−3
Removed at editorial review: mostly commercial or industrial work (3), offices in more than one state (3), a site under maintenance (1), retail stores (2)	−9
Cohort	166

Region (US Census)	Plumbing and HVAC	Electrical	Roofing and exteriors	Appliance repair and handyman	All
Northeast	11	9	10	8	38
Midwest	12	11	12	10	45
South	15	14	13	11	53
West	6	8	7	8	29
All regions	44	42	42	37	165

A3. How the sites were read

Pages. The home page, then for each of nine topics (services, service area, prices and specials, financing, warranty, about, reviews, booking or estimate, contact) the best-matching link on the home page.

Speed and accessibility. Google PageSpeed Insights (Lighthouse) on a mid-range phone profile and on desktop, plus Chrome UX Report data from real visitors over the previous 28 days where Google publishes it for the site.

Crawler conduct. One request at a time to each site, at least 400 ms apart, at most ten pages per site; robots.txt is read and obeyed for every page requested; a site that refuses is left out and never retried. The crawler identifies itself as Mozilla/5.0 (compatible; ClikitiResearch/1.0; +https://clikiti.com).

A4. What was measured

9 questions homeowners ask before they let a tradesperson into the house, answered from what each site publishes, and 4 technical checks against published standards. A check passes only when the rule below is met; anything the crawler cannot confirm counts as not answered.

Question	Counts as answered when	Standard
How do I reach you, and when are you open?Tappable phone plus opening hours	The home page has a tappable phone link, and the home, contact or about page states opening hours (a day and a time range) or 24/7 availability.	Clikiti homeowner question 1 of 9
Do you work in my area?Service area named: towns, ZIP codes or an area page	The site names where the business works: a service-area page linked from the home page, a statement such as "serving the greater Portland area" or "the following areas:" with places, or a list of ZIP codes.	Clikiti homeowner question 2 of 9
Can someone come tonight?Emergency, 24/7 or same-day service stated	Emergency, 24/7, after-hours or same-day service is offered (a warning about emergencies, or "24/7" for answering questions online, does not count).	Clikiti homeowner question 3 of 9
Are you licensed?Licence number shown as text	A state licence or registration number is shown as text (for example "License #M-116444" or "Registration # 13-115186"); a bare "licensed and insured" does not count.	Clikiti homeowner question 4 of 9
Can I book without calling?Book or request an estimate online	A link, button or embedded scheduler lets the visitor book service or request an estimate, quote or inspection online (phone and e-mail links and call-back requests do not count).	Clikiti homeowner question 5 of 9
Can I spread the cost?Financing offered	Financing is offered: a statement such as "financing available" or "flexible financing", 0% APR terms, or a named lender.	Clikiti homeowner question 6 of 9
What if it fails again?Warranty or guarantee stated with its terms	A warranty or guarantee on the work is stated with its terms: days, months or years, or lifetime (a bare "satisfaction guaranteed" does not count).	Clikiti homeowner question 7 of 9
What will the visit cost?A price shown: a service call, a fee, a rate or a job	A dollar price is shown for a service call, a diagnostic or trip fee, a rate, a maintenance plan or a named job, including a typical price range (discounts, rebates, savings, referral rewards and financing amounts do not count).	Clikiti homeowner question 8 of 9
Do my neighbours vouch for you?Reviews marked up, in a widget or quoted	Customer reviews are marked up for search engines (Review or AggregateRating structured data), shown through a review widget, or at least two customer testimonials are quoted.	Clikiti homeowner question 9 of 9
Technical checks
Loads in under 2.5 s on a phoneLoads in under 2.5 s on a phone	Real visitors on phones see the main content within 2.5 seconds (Chrome UX Report, 75th percentile). Where Google publishes no visitor data, the phone lab test is used.	Core Web Vitals · LCP
Text contrast passes 4.5:1Text contrast passes 4.5:1	No text on the home page fails the WCAG 1.4.3 minimum contrast of 4.5:1, in either the phone or desktop test.	WCAG 1.4.3
Images have text descriptionsImages have text descriptions	Every meaningful image on the home page has a text description (WCAG 1.1.1).	WCAG 1.1.1
Plumber, HVACBusiness, Electrician or RoofingContractor structured dataPlumber, HVACBusiness, Electrician or RoofingContractor structured data	The site describes the business to search engines with Plumber, HVACBusiness, Electrician, RoofingContractor, HomeAndConstructionBusiness or GeneralContractor structured data.	Schema.org · HomeAndConstructionBusiness

A5. The website score

Each site's overall score (0 to 100) is Clikiti's website score, rubric 2.0.0. It measures the website, not the business, and the homeowner questions above do not feed it. It is a weighted mean of five pillars:

Pillar	Weight	Based on
Usability	30%	Checks of the home page for broken links and images, and an expert review of the page as rendered on a phone and on a desktop screen, against Jakob Nielsen's ten usability heuristics. Each issue must name the heuristic, a severity on the Nielsen Norman Group scale and the exact element or wording; issues without all three are discarded, and two independent passes must agree.
Speed	25%	Google Lighthouse on a mid-range phone (60%) and on desktop (20%), and real-visitor Core Web Vitals from the Chrome UX Report (20%) where available.
Accessibility	25%	Google Lighthouse accessibility audits (70%) and Clikiti's own WCAG checks of the served page (30%).
Search	10%	Google Lighthouse SEO audits (60%) and Clikiti's own search checks of the served page (40%).
Trust	10%	Google Lighthouse best-practice audits (40%) and security header checks (60%).

Within a pillar, each issue costs points by severity (critical, high, medium, low) and by how many times it occurs; issues based on judgement count at 70% of measured ones, and duplicates of the same root cause count once. On the four business profiles, the number out of 100 is this overall score, and each dot is one check in its group: filled when the site passes it, hollow when it does not.

A6. Precision

With 165 sites, a share reported for the whole sample is precise to within ±7.6 percentage points at 95% confidence (worst case, a 50% share; 1.96 × √(0.25/n)). By type: plumbing and hvac ±14.8 (n = 44); electrical ±15.1 (n = 42); roofing and exteriors ±15.1 (n = 42); appliance repair and handyman ±16.1 (n = 37). Differences between types smaller than these margins should be read as similar. Percentages are rounded to whole numbers; medians are the middle value of the sites in a group.

A7. Quality control

• Every home page in the cohort was fetched and its title checked against the listed business before scanning, and every home page was checked for gambling, pharmacy, adult and crypto spam (none found)
• Businesses were checked for duplicates across domains by phone number (none found)
• Detection rules were checked against the evidence they record for a sample of passes and fails for every rule, and revised before the final run
• Every published figure is recomputed from the data at build time, and a statement in the text that the data does not support stops the build
• Results are published only after automated checks for missing data, anonymity, accessibility of the report itself and the margin of error statement pass

A8. Limitations

• A point-in-time snapshot: sites change, and a site may have improved or declined since it was read.
• Only public pages were read: the home page and pages it links to. Chat widgets, multi-step quote forms and customer portals were not tested.
• Answers are read from the page as a server delivers it. Content that only appears after scripts run in the browser (some whole sites, schedulers and review widgets) is not seen, which the study counts as a gap a search engine or a slow phone would also meet.
• 34 candidates blocked or did not answer automated visitors and are not in the study; sites behind such protection may differ from those studied.
• The study reports what a website shows. It does not judge any contractor's work or prices, and it does not check whether any licence is valid or current.
• Automated accessibility tests find only part of what WCAG covers; keyboard and screen-reader use need a manual audit.
• The sample is a structured list of real, independent businesses, not a random draw. Margins of error are given as a guide to precision, not as a guarantee that the results hold for every business in the country.

A9. Data, licence and corrections

The aggregate results are published as CSV and JSON beside this report under the Creative Commons Attribution 4.0 licence: you may reuse, adapt and share them, including in client proposals, with a credit to Clikiti and a link. No per-site results are published. Corrections and removal requests are reviewed and answered personally; see the end of this report.