On this page
What the visitor is actually doing
Before any layout decision makes sense, be clear about the behaviour you are designing for. Somebody arriving at a trade website is running through three questions in a particular order, and they will leave at the first one you fail to answer.
- Do these people do my specific job? Not the trade in general. The specific thing: a flat roof on a dormer, a driveway in resin, an intermittent fault on a diesel estate, a knee that hurts going downstairs.
- Do they come to where I live? Surprisingly often the reason a visitor leaves. A map image with no town names in the text answers nothing.
- Roughly what does this cost, and are these people in my bracket? The question everybody has and most sites refuse to engage with.
Only after all three does anything else matter: how long you have been trading, what your values are, who is on the team. Those are reassurance questions and they get asked late, by people already leaning towards you. Most trade homepages are built in exactly the reverse order, opening with history and finishing with what is actually done.
The second thing to hold on to is the device. The overwhelming majority of this traffic is a phone, frequently held in one hand, often outdoors, sometimes on a poor signal. Anything that needs a mouse, a wide screen or patience is not being seen the way you are seeing it in the office.
The first screen
On a phone you have roughly the top six centimetres before a scroll, and you should treat that as a budget rather than a canvas. Four things fit and they are always the same four.
- What you do, in the words a customer uses. Not “complete property maintenance solutions”. Roof repairs and re-roofing. Boiler replacement and emergency plumbing. MOTs, servicing and diagnostics.
- Where you do it. Three or four town names in real text. This single line resolves question two immediately and it also gives a search engine something unambiguous to work with.
- A tap-to-call button that is genuinely a button. Not a number typed as text in a header. A real target, large enough for a thumb, wired to dial in one tap.
- One line of qualification or reassurance. The hours you actually answer, a guarantee, a minimum job size, or the accreditation that matters in your trade. One line, not a paragraph.
What should not be there: a rotating slideshow, a video that autoplays, a cookie banner covering the call button, a chat widget that opens itself after four seconds, or a strapline that could belong to any business in any sector. If your first screen would still make sense with a competitor's logo on it, it is not doing any work.
Below that first screen, the homepage has one job, which is routing. A short row of the services you actually want to sell, each linking to its own page. The homepage is a signpost. It is not where the persuading happens, and trying to make it do everything is why so many of them are three thousand words long and convert at two per cent.
The service page, where the work happens
Here is the structural decision that separates sites that produce enquiries from sites that merely exist: one page per service you want to sell. Not a services page with a bulleted list. A page for boiler replacement, a page for bathroom installation, a page for landlord certificates.
Two reasons, one commercial and one mechanical. Commercially, a visitor who searched for a specific job and lands on a general list has to find themselves, and a proportion will go back to the results instead. Mechanically, a page can only rank for what it is genuinely about, and a page about eleven things is about nothing in particular. This is the same argument the SEO side of the work makes, arrived at from the conversion direction.
A service page that works has a predictable anatomy:
| Section | What it contains |
|---|---|
| The named job | A heading using the customer's words, and one paragraph confirming you do exactly that. |
| What is involved | The actual process, in sequence, with realistic timings. This is where expertise shows without anybody claiming any. |
| What it costs | A range, a rate, or a worked example, plus the three or four things that move it. |
| Proof of this job | Photographs of that job on that kind of property. Not the trade in general. |
| The objection | The specific thing people worry about with this job, answered before they ask it. |
| The ask | Call, book or send details, repeated here rather than left in the header. |
On pricing, the standard objection is that every job is different. It is, and that is not a reason to publish nothing. Give a range and name the variables: access, ground conditions, what is found once the covering comes off, the specification of the parts. Publishing a range costs you enquiries from people who were never going to buy, which is the point. It also removes the suspicion that the price depends on what the customer looks like they can afford. We set the full version of that argument out in the comparison of a cheap site and a proper one.
Proof, and where it belongs
Proof fails not through absence but through placement. Most trade sites have reviews, photographs and accreditations, and put all three somewhere nobody reaches.
Reviews belong next to the decision, which means next to the call button and on the service page, not gathered on a testimonials page that receives almost no traffic. Use the real ones, with a first name and a town, and link out to the platform they live on so they can be checked. BrightLocal's Local Consumer Review Survey 2026, a US panel of 1,002 consumers, found that 97% read reviews for local businesses and 68% would only use a business rated four stars or higher. It is an American sample, so treat the exact figures with appropriate caution, but the direction is not controversial to anyone who has watched a customer decide.
Photographs are the most under-used asset in the trades. Yours do not need to be professional; they need to be real, recent and labelled. A picture of a job with the town and the problem written under it outperforms a styled stock image by a distance, because it proves something rather than decorating something. Strip the location data before uploading, size them properly, and file them by job type so a visitor can find something resembling their own house.
Accreditations need to be verifiable rather than decorative. A logo strip proves nothing and an expired one is worse than none. Print the registration number as text, link it to the public register where one exists, and name the person it belongs to. Gas Safe, FENSA, NICEIC, HCPC, DVSA, CHAS: all of them publish a check, and using it converts a claim into a fact.
The ask: phone, form, or both
Trade sites need both, weighted by how urgent the job is.
For urgent work the phone wins outright and nothing else comes close. Nobody completes a form while water comes through a light fitting. That means a tap-to-call button on every page, a sticky call bar on mobile, and honesty about the hours you genuinely answer. An advertised 24-hour line that reaches voicemail at nine at night is worse than advertising office hours, because the caller has now formed an opinion.
For planned work the form does real work, and its length should match the value of the job. Short forms maximise the number of enquiries. Long forms maximise the quality. For a four-figure job with a survey attached, longer is almost always correct: postcode, job type, photographs, budget bracket, timing. A builder who stops driving to four dead-end evening appointments a month has gained more than a site that produced four more form fills.
- Ask for a mobile number, not just an email. Trades reach people by phone and by text.
- Allow photograph upload wherever the job can be partly assessed visually. It saves a visit and it dramatically improves quote accuracy.
- Confirm on screen and by message that the enquiry arrived and when they will hear back. Silence after a form submission is where a surprising number of enquiries are lost, and it is what follow-up automation exists to remove.
- Track both routes. If you cannot tell which page produced which enquiry, every improvement you make afterwards is a guess. The methods are covered in the tracking guide.
Speed, rendering and being found at all
Two technical properties genuinely affect enquiries, and both are invisible in the office.
Speed on a real phone. The measures that matter are the Core Web Vitals: how quickly the largest thing on screen appears, how quickly the page responds when somebody taps, and whether the layout shifts under their thumb while it loads. That last one is the quiet killer on trade sites, because a late-loading image pushes the call button down half a second after somebody aimed at it. The causes are almost always the same: uncompressed photographs from a phone camera, a slideshow library, three tracking scripts and a chat widget. Test on a mid-range Android over mobile data, not on the office broadband.
Whether your content exists before JavaScript runs. Many modern website builders output an application rather than a document: the HTML delivered on the first request is close to empty, and the words only appear once a browser downloads and executes the code. Search engines will usually render it eventually, on a delayed second pass, which makes indexing slow and patchy. The larger issue now is that the crawlers behind AI assistants largely do not execute JavaScript at all, so to them the page has no content. You can check your own site in thirty seconds by viewing the page source and searching it for a sentence you know is on the page. If it is not there, that is what a machine sees. Everything we build is server-rendered for exactly this reason.
When the website is not the problem
This is the part most agencies leave out, so here it is plainly. Rebuild is frequently the wrong answer.
If your site gets 120 visits a month, the conversion rate is not your constraint. Doubling it produces two more enquiries. Your constraint is that almost nobody is arriving, and the money belongs in demand or in local search visibility, not in a new site. Check the visitor numbers before you commission anything.
If enquiries arrive and are not converted, the site is doing its job and the handling is not. Firms in that position rebuild the website, get the same number of enquiries, and conclude the rebuild failed. The follow-up side is where that money should go.
And if the site is simply old but structurally sound, the cheaper answer is usually a list of fixes rather than a replacement: move the phone number up, write a page for each service, name the towns in text, compress the images, publish a price range, wire up tracking. We would rather scope that than sell a rebuild, and if that is genuinely what your site needs we will say so on the first call.
Rather have someone else do this?We run exactly this for trade and construction firms across the UK. Call and we will tell you what is realistic in your area, including when the answer is that it is not worth it.
Call 07443 392243WhatsAppCommon questions
How many pages does a trade website actually need?
Fewer than most people assume, and structured rather than numerous. One page for each service you genuinely want to sell, one for the areas you cover, one showing real work, one about the business and one to make contact. For most trades that is six to twelve pages written properly. Thirty thin pages generated from a template perform worse than eight written by somebody who does the work.
Should the homepage have a contact form on it?
A short one low down, yes, but it should never be the main event. The homepage exists to route people to the page about their job, and that page is where the enquiry should happen because it is the only place you know what they want. A homepage form produces vague enquiries that take two phone calls to understand.
Is a blog worth having?
Only if it will be maintained. An abandoned blog with three posts from two years ago actively damages credibility, because it is the clearest available evidence that nobody is looking after the site. Five genuinely useful service pages beat twenty posts about seasonal maintenance tips that nobody searched for.
How do I know whether my current site converts badly?
Two numbers: monthly visitors from analytics, and enquiries attributable to the site, counted rather than estimated. Divide the second by the first. For a focused trade site with a working phone number, mid-single figures to low double figures is ordinary territory. Under two per cent with reasonable traffic points at the site; under two per cent with almost no traffic points elsewhere.
Do customers really care about page speed?
They do not think about it and they respond to it anyway. Nobody abandons a page consciously reciting a load time; they simply lose patience, especially on mobile data outside a property. The practical test is to open your own site on a phone with a poor signal and notice how you feel by the time the call button appears.