"We run growth like a lab" is on our home page, and it's the sort of line that could mean nothing. So here is exactly what it means on a Monday morning: the one-page plan we write before work starts, the weekly rhythm while it's running, and the monthly reading that tells you what changed and why. None of it needs special software. All of it needs discipline.
Why a lab, and not a to-do list
Most store marketing is a to-do list: redo the homepage, try TikTok, send more emails. Things get done and nobody can say afterwards which of them mattered. A lab flips that. Every piece of work starts as a hypothesis, a guess about what will move a number, and ends with a reading that says whether it did. Work that can't be read isn't done in the lab; it's done in a separate bucket called "housekeeping", and it's kept small.
The payoff is not that every test wins. Most don't. The payoff is that you stop repeating the ones that lost, and you stack the ones that won.
The one-page plan
Before we change anything on a client's store, there's a page with six lines on it:
- Hypothesis. One sentence, in the form "If we ___, then ___ will ___, because ___." If we add shipping cost and delivery time under the add-to-cart button, checkout completion will rise, because surprise shipping costs are the top reason people abandon.
- Metric. The one number that decides it. Not "engagement". Checkout completion rate, in this case.
- Baseline. What that number has been for the last four weeks. If you don't know it, finding out is step zero.
- Target. What "worked" looks like: a number, not a feeling. A modest, honest target beats an exciting one you'll rationalise later.
- Deadline. When we'll read it. Long enough to get enough orders to mean something; short enough that we don't forget why we started.
- Owner. One name.
Writing this takes twenty minutes and kills about a third of ideas on the spot, usually because there's no baseline, or because the honest target is too small to bother with. That's the page doing its job.
One change at a time (mostly)
If you change the product page, the ad creative and the email flow in the same week and sales rise, you've learnt that something works. You can't repeat "something". So we run one test per metric at a time. Two tests can run in parallel if they touch different numbers. A checkout test and an ad-creative test don't interfere. Two tests on the same product page do.
The weekly rhythm
Monday, thirty minutes: open the running tests, look at the metric versus baseline, and decide three things: keep running, stop early because it's clearly broken, or stop early because it's clearly won and there's no point waiting. Then ship the week's build work. Friday, ten minutes: a short note to the client: what shipped, what we're watching, anything surprising. No decks. If a test needs a decision from the founder, it's asked in that note with a recommendation attached.
The monthly reading
Once a month we write the reading. It's one page again, and it has a fixed shape:
- What we tested, with the hypothesis as written a month ago, not as we'd like to remember it.
- What the number did, against the baseline, with the honest caveats (a sale week, a stock-out, a tracking blip).
- What we're keeping, what we're reverting, and why.
- What we're testing next, and the hypothesis for it.
- The five store numbers (conversion rate, average order value, cost per order, repeat rate and top search terms), so the tests sit inside the bigger picture.
Over a year, that's twelve pages that tell the whole story of the store: what worked, what didn't, and how the business changed. Founders tell us it's the first time they've been able to answer "what did the marketing actually do?"
When to stop a test
Three honest stopping points: the deadline arrives; the number moves so far, so fast, that waiting would just be ceremony; or the test breaks something else (a speed drop, a spike in support tickets). "It's not working yet, let's tweak it" is not a stopping point. It's a new hypothesis, and it gets its own page.
How to start on your own store
Pick one number you care about this month. Write the six lines. Change one thing. Read it in two or three weeks. Keep the page. Do it again. Twelve of those and you'll have a store that improves on purpose instead of by accident, which is the whole idea.
