The friction ledger: what actually leaks conversion

Before you touch the copy or argue about the hero colour, audit the page against the things that genuinely leak conversion. I keep a friction ledger, a short list of auditable line items, and I score a page against it the way a bookkeeper reconciles an account. The point is to fix what is measurably broken before optimising what merely feels off, because most pages are decorating a leaky bucket.

The six line items

There are six places conversion reliably escapes a B2B landing page. Competing calls to action, where the page offers more than one thing to do. A buried primary action, where the thing you most want the visitor to do is below the fold or visually outranked by something else. Slow load, where the page renders so slowly that paid clicks bounce before they see it. Surplus form fields, where you ask for more than you need to start a conversation. Missing proof, where you assert value without demonstrating it. And broken message match, where the headline does not finish the sentence the ad started. Score the page out of these six, honestly, before you write a single new word. Most pages fail three or four of them, and those failures cost more than any amount of clever copy can recover.

Most people optimise decoration first

Here is the uncomfortable pattern. Only around seventeen percent of marketers even run A/B tests on their landing pages, and a large share of those who do are testing button colours and headline phrasings on a page that is still asking for eleven form fields and loading in four seconds. They are optimising decoration on a structure that is leaking from six visible holes. Testing is not the problem, the sequence is. You cannot test your way to a great page when the obvious leaks are still open, because the noise from the leaks swamps the signal from the test.

Subtract first, then test

The discipline is subtract-then-test. Claim the free, known wins first by closing the six holes in the ledger, focus the page on one action, match the message to the source, get the load under control, cut the form to the essentials, add the one proof that matters. These cost you nothing but the courage to delete. Only once the page is at its minimum viable state, with no obvious friction left, do you A/B test the few high-leverage variables that remain. The order matters because every euro of traffic you spend testing decoration on a leaky page is a euro you will never get back, and the leaks are exactly the variables you can fix without any traffic at all. Get the structure right first, then let testing earn you the last few points.