Stop optimizing the landing page. Optimize the promise.

A conversion programme has a predictable arc. The first quarter produces a couple of double-digit wins from things that were straightforwardly broken. The second quarter produces smaller wins. By the fourth, the team is testing button copy and reporting results that do not survive a significance check.

The usual diagnosis is that the easy wins are gone. The actual diagnosis is that the programme is only allowed to change one layer of a three-layer problem.

The three layers

• The promise. What the ad, the search result or the referral led somebody to expect.
• The page. How well the page delivers on that expectation.
• The offer. Whether what is being asked for is proportionate to what is being given.

Nearly all testing happens in the middle layer, because that is the layer the conversion team owns. But if the promise oversells, no layout fixes it — you are optimising how efficiently people discover they were misled. And if the offer is wrong, a better page just gets more people to the moment where they decline.

If the promise oversells, a better page only means people discover the mismatch faster.

How to tell which layer you are in

Look at where people leave, and how quickly. A promise problem shows up as fast exits from a specific traffic source with an otherwise healthy page. An offer problem shows up as deep engagement followed by abandonment at the point of commitment — people who read everything and then decline are not confused, they have understood you and said no.

Only when both of those look clean is the page itself the constraint. In our experience that is perhaps a third of the time.

What to do about it

Get the conversion programme a mandate that extends to the ad creative and the pricing page, or accept that it will plateau. This is a political problem rather than a methodological one, which is why it is rarely written about, and why it is usually the highest-value thing to fix.