What a fast, call-first contractor website needs technically
When a contractor asks me to look at a site that “isn’t getting calls,” I start with the phone, the form, and how heavy the page is on a mid-range phone — not with a new logo. A homepage can look finished and still fail
When a contractor asks me to look at a site that “isn’t getting calls,” I start with the phone, the form, and how heavy the page is on a mid-range phone — not with a new logo. A homepage can look finished and still fail the actual job: a homeowner finds you, decides you are worth a try, and reaches you in one or two taps.
I am Colton Wehr, founder of Kinexis Digital. I build website plans for local service businesses in the United States. This is a technical piece for owners and developers. It is not a promise that any one markup trick will move you in the local pack.
The number has to work on the first try
Most of the people who will hire you are on a phone, often comparing you with one other company in the same tab. If the number is not tappable, sits behind a popup, or points at a different string than the one on the truck, you have already added friction.
A call-first page does a few unglamorous things well:
- The visible number is a real
tel:link, including the country code in the href (for a US number, that looks liketel:+1…). The text they see should match the number they will dial. - The tap target is large enough to hit with a thumb, in the header and near the first offer of work, not only in a footer nobody reaches.
- The same number is the one in the button, the contact page, and any structured data you publish.
- Popups, cookie banners, and chat widgets do not cover the button.
I wrote a longer, non-technical walkthrough of the mobile checks I use — number, form, page order, and load — in make your service website mobile ready. The rest of this piece is what sits underneath those checks.
Fast enough that the call button appears before they bounce
“Fast” is not a vibe. Google’s Core Web Vitals, documented on web.dev, are the three field metrics site owners are asked to watch, measured at the 75th percentile of real visits, on mobile and on desktop:
- Largest Contentful Paint (LCP): the main content should appear within 2.5 seconds.
- Interaction to Next Paint (INP): after a tap or click, the page should respond within 200 milliseconds.
- Cumulative Layout Shift (CLS): the layout should stay stable, with a score of 0.1 or less.
Those thresholds are Google’s published targets, not a Kinexis score. I will not invent how many contractor sites pass them. What I see slowing trades sites down is more ordinary: a looping hero video, an uncompressed slideshow, and two or three third-party scripts — chat, a review widget, an ad tag. The logo and phone number then arrive late, or they jump once a banner loads, which is both an LCP and a CLS problem. A page builder that ships a large JavaScript bundle can also delay INP: the number is visible, but the first tap does nothing for a beat.
I would do the work in this order. Put the phone number and the first headline in ordinary HTML, not behind a script that has to wake up. Compress and size images for a phone screen; do not ship a 4,000-pixel truck photo into a 400-pixel slot. Be stingy with third-party tags. Then measure the live site with PageSpeed Insights or Search Console’s Core Web Vitals report, which can show field data from the Chrome User Experience Report when enough visits exist. A single lab test on a fast office connection is not the same as what a customer sees on a job-site phone. I would not spend a month chasing a lab score while the number is still plain text.
Schema should describe the business you actually have
Structured data is a machine-readable description of the business. It is also public. If you would not print a fact on the homepage, do not put it in JSON-LD.
Google’s LocalBusiness structured data documentation (updated 8 September 2026) is explicit about eligibility for a local-business rich result: you must include the required properties. For LocalBusiness, those required properties are name and address, and address is defined as “the physical location of the business,” as a PostalAddress. Official examples use a full street, city, region, postal code, and country.
That creates a real fork for a service-area contractor — the plumber or painter who goes to the customer, and who may already hide a home address on a Google Business Profile.
If customers visit a shop, yard, or office, and that address is public, mark it up accurately. Include as many PostalAddress properties as you honestly have. Add telephone, url, and opening hours only when they match the site. If you do not have a public street address, do not invent one, and do not paste a hidden home address into the page source. Crawlers and competitors can read it.
areaServed, defined at schema.org/areaServed, is a legitimate way to name the cities or regions you actually serve. I use it when the service area is real. I need to be honest about a limit: Google’s LocalBusiness rich-result documentation does not list areaServed as a required or recommended property. Adding it does not, by itself, make you eligible for that rich result, and I will not claim that it does. If there is no public street to mark up, I would rather publish an Organization or a more specific type with name, url, telephone, areaServed, and a PostalAddress that only contains addressCountry (for example, US) than fake a storefront. That markup may not qualify for a LocalBusiness rich result. Accurate information is still better than a rich result built on a street you do not serve from.
Keep the visible page, the Google Business Profile (if you have one), and the structured data in agreement. A hidden profile address with a public street in the website schema is a contradiction. Hours and the phone number should match too. Do not invent a rating in markup.
A short form that people can finish, and that bots have a harder time with
The second path, when someone will not call, is a short quote form. I have already written about what belongs on a contractor quote form: name, phone, a place for the work, and a short description beat a ten-field application. The technical job is to keep that form usable and to cut down automated junk without punishing a customer in a driveway.
Keep the field count low. Every extra box is a place a real person stops, and a place a bot does not care about. Rate-limit submissions by IP and by cookie or token, on the server — a form that can be posted a hundred times a minute will be. Add a hidden honeypot field that people never see; if it is filled, discard the submission. That is simple, and it is not enough on its own. Put a bot check behind the submit button that real customers can pass. A well-known option is a challenge such as Cloudflare Turnstile or a comparable provider. I am not endorsing one vendor here, and I will not invent a percentage for how much spam it stops. The test is practical: can a person in a truck complete it, and do the obvious bot dumps drop? Then confirm success in plain language, and send the lead to an inbox or a phone that someone actually checks. A beautifully filtered form that lands in a folder nobody opens is still a missed job.
Do not lock the form behind an account, and do not use a puzzle so hard that an owner with gloves on gives up. Spam control is a filter, not a second website.
None of this replaces answering the phone. It is the technical floor under a site whose job is to turn a search into a conversation. If you want a homepage preview only, that is the one free offer I make: no card, no obligation, ready within two business days after we have your business information. The technical work above is what I would want that preview, and the site that follows it, to get right.
Colton Wehr is the founder of Kinexis Digital, which builds website plans for local service businesses in the United States.
Originally published by Dev.to WebDev. Aggregated on AIWithGhost for educational purposes — full credit and traffic to the original publisher.