The forty-eight weeks between your launches
Three or four significant launches a year is a reasonable pace. A replatform, a redesign, a major campaign push – these are worth doing. That’s not the problem.
The problem is the forty-eight weeks between them, where nothing is learning.
A conversation with a client this week brought a familiar pattern into focus. They had a solid testing backlog: real hypotheses, documented, prioritised. They also had a significant platform migration on the horizon. Once that was live, they’d run the backlog properly on a better foundation. The testing could wait.
I hear some version of this regularly. The moment the serious work starts is always just ahead. Or lots of small improvements snowball into one big project that has to be done all at once.
The maths of waiting
Most experiments won’t produce a result worth announcing. Some will point in the wrong direction. But a handful over the course of a year will move things – a checkout change that lifts completion by 1.8%, a delivery messaging update that reduces late-stage abandonment, a returns copy test that adds 0.6% to conversion. Those changes don’t disappear when the next test launches. They stay in the baseline.
Individually, none of those numbers make a board slide. Together, over twelve months, they become one.
A big launch is a concentrated bet, and concentrated bets are worth making. The issue is that three or four of them a year leaves the rest of the time unaccounted for. Between those launches, the checkout is the same, the product page is the same, the email flow is the same. Nothing is shifting.
Compound interest doesn’t reward the size of the deposit. It rewards the frequency.
What the launch misses
A testing cadence feels unglamorous. Individual results are small. It’s genuinely hard to stand in front of a board and say “we ran twenty tests this quarter” as a strategic achievement – even when those twenty tests produced a net 4% conversion improvement that will still be active in twelve months.
A big launch is visible. It has a date, a budget, a before-and-after story. It deserves to be. But it does one thing: it moves the line in a single moment. The compounding happens in the gaps, and most teams aren’t filling them.
Part of what makes this hard is how teams are taught to measure experiments – the expectation that each test should show a return that justifies the effort before the next one gets approved. That framing turns a compounding programme into a series of individual bets, each one asked to stand alone.
What a cadence actually requires
The discipline is in the hypothesis, not the volume. A hypothesis looks like: “We think showing the delivery estimate higher on the product page will reduce abandonment, because customers need reassurance before they commit, not at checkout.” Two weeks, one variable, one decision. That closed loop is the thing that compounds.
The compounding question is simple: what’s in the baseline today that wasn’t there six months ago?
If the answer is the replatform, you have one improvement. If the answer is a series of small changes that all held – checkout flow, product page, email sequence, returns copy – you’ve compounded. The starting line for next year’s work is different.
The brands doing this well at scale aren’t choosing between the two. The launches set the direction. The experiments move the baseline in between. Both are happening, and the bigger the operation, the more that’s true – not because they have more resource, but because they’ve worked out that launches punctuate and experiments compound.
Progress doesn’t arrive in episodes. It accrues.