The short answer
Most sites that underperform are not ugly. They are slow on a phone, hard to call from, over-navigated, over-asking on forms, and silent at the bottom of the page. Fix those five in order and conversion usually moves before any redesign is needed.
A hero image nobody compressed
The most common speed problem on a small business site is one photograph. A four megabyte JPEG straight off a camera, dropped at the top of the homepage, becomes the largest contentful paint element and holds the whole page hostage on a phone over cell data.
Export it at the size it is actually displayed, save it as WebP, and keep the hero somewhere under a couple of hundred kilobytes. Do not lazy-load it. Lazy loading helps images below the fold and hurts the one at the top. Run the page through PageSpeed Insights on the mobile tab and watch the largest contentful paint figure before and after.
A phone number that is not a phone number
On a service site the number in the header should be a tel link, so a tap dials. We still find numbers baked into a header graphic, or styled in a way that stops a phone recognizing them, or hidden two taps deep on a contact page.
Make the tap target generous as well. A number squeezed into a thin line of text beside a hamburger menu will be missed by half the thumbs that try it. Test it on your own phone, one-handed, standing up, the way a customer in a parking lot would.
Navigation with too many doors
Eleven top-level menu items is a filing cabinet, not navigation. Nobody reads a menu. They scan it for the two or three words matching what brought them.
Name pages the way a customer would say them out loud. "Roof Repair" and "Roof Replacement" beat "Solutions" and "Capabilities" every time. Push the rest into the footer, where anyone who genuinely wants the careers page will happily go looking.
A form that asks more than a stranger will give
A nine-field form on a first visit is a negotiation you lose. Name, phone number and one open box describing the job is usually enough to call somebody back and qualify them properly on the phone, which is where qualifying belongs.
Every field you remove raises completion. If the office insists on the address and a preferred appointment window, ask for them on a second step after the contact details are already submitted, so an abandonment still leaves you a lead you can call.
While you are in there, check where the submissions go. We regularly find forms mailing to an address nobody has opened since a bookkeeper left, or landing in a spam folder because the site sends from a domain it is not authorized to send from. Submit your own form once a month. It takes a minute and it has saved clients entire quarters of leads.
Pages that end without asking for anything
This one shows up on interior pages. The homepage got a call to action because somebody thought about the homepage. The service pages end with a paragraph and white space, and the visitor who read all the way down has to scroll back up to find a way to reach you.
One clear action at the end of every page, in the same words and the same button, works better than variety. Two competing buttons of equal weight is not a choice, it is a pause. If a page needs a secondary option, make it visibly secondary.
The sixth one, since we are being honest
Stock photography. A smiling model in a hard hat that has never been on a roof, on a page asking somebody to hand you a deposit. Visitors do not consciously notice it, but it costs trust in the same way a stock family costs trust on a dental site.
You have better photographs on your phone right now. Crew in your own shirts, your own trucks, jobs on streets in Monmouth County your customers will recognize. They are lower resolution and worth considerably more. The same goes for a testimonial with no name attached to it, which reads as invented whether or not it is.
How to check your own site in ten minutes
Open it on your phone using cell data rather than office wifi. Count the seconds until there is something readable on screen. Tap the phone number. Try to reach your main service page from the homepage in one tap. Then fill in your own form and time how long the lead takes to reach whoever is supposed to answer it.
Finally, hand your phone to somebody who does not work with you, give them five seconds and ask what you sell and who you sell it to. Those five tests cost nothing, take an afternoon, and find more revenue than most redesigns do. When we plan a rebuild, this is where the brief starts, and about a third of the time the honest recommendation is to fix the existing site instead.
Follow-up questions
Usually not. Image compression, a tel link, a trimmed menu, a shorter form and a call to action on every page are edits to an existing site. Redesign when the structure itself is wrong, not because the conversion rate is low.
Fast enough that something readable appears in about two seconds on a phone over cell data. Chasing a perfect score is not worth the money. Getting the largest image under control is where nearly all the real gain is.
In our experience, yes. Sliders push the first real message below the fold, add weight, and almost nobody sees the second slide. One strong image and one clear sentence outperforms a rotation of five in nearly every test we have run.
Let’s build something.
Twenty minutes on the phone and you’ll know whether we’re the right fit. Either way you’ll leave with something useful.