Conversion rate optimization starts with your data, not your homepage

Half the conversion rate optimization work we get called into is decoration. Someone repaints a button, the test runs for two weeks, the number moves a third of a percent, and it gets written up as a win. The business does not notice. Nearly always it is the same story underneath: the work started on the website, and the website was never where the money got decided.

The website is the last place to start

Most conversion advice assumes the funnel already points at the right thing and your job is to grease it. So people rewrite the hero, shuffle the form fields, hold strong opinions about a headline. Some of that helps a little. None of it stops to ask whether the site is even built to sell the work that pays the bills.

Say you double the conversion rate on the page for your lowest-margin service. Now you are doing more of your least profitable work, faster, and calling it growth. Conversion just multiplies whatever you put in front of it. So the first thing to get right is which service you point people at. You work that out from the revenue; the layout comes later.

Read the revenue before the heatmap

So we start in the numbers. Sessions and clicks get skipped; they only tell you what got looked at. We line up the closed-won deals from the last few quarters by service line, in a plain spreadsheet, and sort on margin. Usually one row is much bigger than the story the website is telling. Then we go and ask where those particular customers came from, and what they bought first.

That sorted list is the brief. It is duller than a heatmap, and it is the only thing worth building a conversion program on. Same discipline as writing an ICP your own salespeople would recognise, or campaigns you can follow all the way to pipeline.

A worked example: the profit engine the website was hiding

A US software services firm came to us wanting more conversions from the site. Reasonable ask. We spent the first week in their numbers instead. One of their practices, a fairly specialised one, was quietly responsible for a big share of the profit, well out of proportion to how much of the site it got. It was the work they were best at, and the work people kept coming back for.

The site had not kept up. That practice lived in the same small nav item as everything else, so someone who arrived needing exactly it hit a general homepage and had to go hunting. The best demand they had was walking in the door and getting pointed to a waiting room.

So we built the site around what the numbers already knew. That practice got its own way in, a page written in the language those buyers use, and tracking wired so we could see which touches turned into pipeline for it, not just which pages were busy.

It was not carelessness. The site had grown one reasonable decision at a time, and somewhere along the way it started describing the company to itself and forgot to sell. The fix was mostly a decision: point at the work that pays.

Aim the infrastructure at the winner

Once the data has picked the winner, most of the guesswork stops. The people who come for that service get a route built for them instead of a homepage and a shrug. The page uses their words and shows the proof that calms them down. The tracking gets clean enough to say which touch produced pipeline, instead of which page happened to be popular. And leads stop falling into the space between two people who each assumed the other had them, which is a handoff problem that no amount of page work will fix.

Whatever it says on the invoice, this is not a redesign. It is mostly the same work you already meant to do, finally aimed at one thing. The page is only the surface. The data model and the routing sitting under it are what decide whether a conversion meant anything, which is the part that quietly belongs to marketing operations.

What counts as a conversion

Count the wrong thing and you will get more of it, reliably. Most sites count raw form fills, so raw form fills are what they get: a calendar full of calls the sales team spends the week politely ending, and a pipeline that still looks thin on Friday. The number worth chasing is a real conversation about the work you actually want more of. Once that is what the word means, most of the other arguments settle themselves.

Polish the page and ignore the plumbing, and you end up back at repainting buttons. That is usually why conversion rate optimization underwhelms. It got handed a problem that lived in the data and the operations, then took the blame when a nicer button did not fix it.

How to try it before you call anyone

Pull revenue and margin by service line, deal type and source for the last few quarters, and rank it, weighting margin over headline revenue. Whatever is carrying the business is your winner. It is usually one or two things, and usually not what the homepage gives the most space to. Now open your own site as a stranger who needs exactly that, and count the clicks to an obvious next step. Most sites do not have one. Build the way in you wish was there, then fix the tracking so you can read the result in pipeline instead of form submissions.

That is the whole reorder. The data tells you what to convert people into; the page and the buttons come after that, and honestly the buttons barely matter. If you would rather someone sat in the data with you, that is what a go-to-market audit is for, and it is where we start every engagement.

Common questions