How we read a contractor website.
Before any redesign conversation, we read a site the way a buyer does, then the way a search engine does, then the way a competitor would. Five layers, examined in strict order, because a failure at one layer makes every layer below it irrelevant. This is that reading, in full, so you can run it on your own site before anyone quotes you a rebuild.
We have not crawled the industry's websites and we quote no percentages about them. This is the examination we run by hand inside diagnostics, distilled from what a buyer, a search engine and an AI assistant each demand from a contractor's site. Every check below can be performed on your own site, today, for free.
Website arguments start at design. Buyers never do.
When owners talk about their website, they talk about how it looks. When agencies pitch a website, they pitch how it will look. Both are starting at the wrong end, because a buyer experiences a website in a fixed sequence, and appearance sits near the end of it.
Nobody admires the design of a site they never found. Nobody reads the message of a page that took too long to load, standing in a hot kitchen on a weak signal. Nobody follows a call-to-action they never scrolled to reach. Each layer of the experience is a precondition for the one after it.
So we read sites in the buyer's order: can it be found, does it load, does it say the right thing, does it prove it, and can the visitor act in one motion. The order is the method. It is also the priority list for spending money, because fixing layer four while layer one is broken buys nothing at all.
A beautiful website that cannot be found is a billboard in a basement.
Five layers, one question each.
Each layer gets one blunt question. A "no" at any layer stops the reading, because the buyer stopped there too. That is the whole discipline: the examination halts where the visitor halts.
Findability
Does this site exist where your buyers are searching, across your whole service area?
Failure looks like One town ranks, six do not. Services with no page of their own. A site invisible for every search that carries money.
Speed
Does it load fast on a mid-range phone, on cellular, where the visit actually happens?
Failure looks like Fast on the office wifi, slow in the field. Oversized images. Every extra second quietly returning visitors to the search results.
Message
Within seconds, is it clear what you do, where you do it, and how price gets handled?
Failure looks like A homepage about the company's pride instead of the buyer's problem. Total silence on price. Slogans where answers should be.
Proof
Would a skeptic believe you: real jobs, real places, real reviews, plainly stated credentials?
Failure looks like Stock photos of models in hard hats. Borrowed badges. Claims of excellence with nothing specific enough to check.
Path
From any point on any page, can the visitor act in one motion, and do they know what happens next?
Failure looks like A phone number that is not tappable. A six-field form. Pages that end without asking for anything. Silence about after-hours.
Read top to bottom, stop at the first "no". The layer where you stopped is where the next dollar belongs. Everything below it can wait, however tempting it looks.
The checks behind each question.
These are the working checks we run at each layer during a diagnostic. None of them needs your passwords. All of them can be done from a phone and a search bar, which is exactly the point: this is how the market already examines you.
Findability: the site your buyers can reach
A contractor site earns its keep one search at a time, town by town, service by service. The question is never "does the site rank" but "which money searches does it show for, and which does it forfeit".
- A real page for each core service, written for that job, not a list in a paragraph
- Coverage for the towns you actually serve, not just the one in your address
- Basic technical hygiene: indexable, crawlable, no broken structure hiding the pages that matter
Speed: measured where the buyer stands
Speed is a measurement, not an opinion. The test that counts is a mid-range phone on a cellular connection, because that is the device in the hand of a homeowner with a broken furnace.
- Load time on cellular, tested from inside the service area
- Images sized for phones, not uploaded straight from a camera
- Nothing blocking the first screen: the page is readable before it is finished
Message: the three answers, above the fold
A buyer arrives with three questions: do you do my job, do you cover my street, and what happens about price. Most contractor sites answer the first, gesture at the second and refuse the third. The refusal is the leak. You do not need a price list. You need to handle the question out loud: how estimates work, what drives cost, what a visit costs if anything.
- First screen states the trade, the area and the promise in plain words
- The price question addressed somewhere obvious, even as a process or a range
- Copy written for the buyer's situation, not the founder's biography
Proof: specific enough to check
Trust is built from checkable specifics. A photo of your crew on a named street outworks any stock image. A review quoted with a date outworks a five-star graphic. And fabricated trust, borrowed badges, invented associations, staged reviews, is worse than none: buyers have learned to smell it, and so have the AI assistants summarizing you.
- Real photographs of real jobs, ideally placeable in your area
- Reviews surfaced with enough context to be believed
- License, insurance and guarantees stated plainly, where a skeptic will look
Path: one motion, no dead ends
Every page is a possible entrance and every entrance needs an exit toward contact. The standard is one motion: a thumb reaches a tappable number or a short form without hunting. The path also sets expectations, because "we reply within the hour" changes what the buyer does next while they wait.
- Tappable phone number and a visible action on every screen of every page
- Forms that ask for the minimum: name, contact, problem
- What happens next, stated: response time, after-hours handling, who calls back
Four sites we meet over and over.
Run enough of these examinations and the same characters keep appearing. If you recognize your site in one of them, you also now know which layer it fails at.
The billboard in the basement
Often genuinely handsome, sometimes expensive, ranked for nothing but its own name. It was designed as a portfolio piece and it functions as one: admired by the owner, unseen by the market. The rebuild it usually gets next fixes the one thing that was never broken.
The brochure about us
Opens with the founding year, the family story and a mission statement. All fine things, none of them the answer to "do you do my job and what will it cost". The buyer's questions are in the exam and this site studied for a different test.
The stock-photo fog
Model-perfect technicians who work for a photo agency, trucks with blurred logos, kitchens from another continent. Every image is a small confession that nothing specific was available. Buyers cannot name what feels off, but they feel it, and they leave.
The dead end
Decent content that simply stops. Pages end without an ask, the number lives only in the header, the form wants a paragraph of your time. The site does the persuading and then forfeits the conversion, which means it did the work and skipped the invoice.
Why we never grade a site out of 100.
Website audit tools hand out scores because scores are easy to generate and easy to sell against. But a score averages the layers, and averaging is exactly wrong: a site that is excellent at four layers and broken at one does not deserve an 80. It deserves a halt at the broken layer, because that is what the buyer gives it.
So the output of this reading is never a number. It is a ranked list: the layer where visitors are currently stopping, what it costs, and what fixing it is worth relative to everything else in your marketing. Sometimes the honest verdict is that the website is fine and the money belongs elsewhere. A score cannot say that. A reading can.
Buyers do not average your layers. They stop at the first broken one.
The rest of the machine.
Find the layer where your visitors stop.
The Marketing Diagnostic runs this reading on your site against the competitors actually winning your searches, with the findings in writing. $1,000, credited in full to month one of any program.