From Experiment to Play to Capability
Most marketing tests evaporate the moment they end. A useful one leaves behind a procedure you can run again — and, eventually, a portfolio of procedures you can reshuffle when the market moves. That spine, Experiment → Play → Capability, is how a single test turns into something a competitor can’t easily copy.
By William Walczak — CEO, Hiilite Creative Group Inc.The problem: one-off tests that evaporate
If you’ve run marketing for a small business, you’ve probably done a lot of testing. A new subject line. A different hero headline. A landing page with the form moved up. Some of those tests win, some lose, and almost all of them disappear — the result lives in someone’s memory or a forgotten spreadsheet, and three months later you’re re-running a version of the same test because nobody wrote down what you learned.
That’s not a discipline problem. It’s a structure problem. A test that produces a number but no reusable procedure can’t make your next test smarter. Learning that doesn’t get captured can’t compound. The fix is to treat each layer of your marketing work as a distinct thing with a distinct job: the experiment, the play, and the capability.
An Experiment is one build-measure-learn test
An experiment is the smallest unit: a single, time-boxed test with a hypothesis and a way to read the result. You change one thing, you measure whether it moved the outcome you care about, and you keep or discard it. “If we add customer reviews to the pricing page, more visitors will start a trial.” Run it, read it, decide.
The hard part of an experiment isn’t running it — it’s reading it honestly. A lift in trials right after a change isn’t proof the change caused the lift; seasonality, a parallel ad push, or simple noise can fool you. This is where how we measure growth matters: a real experiment compares against a credible counterfactual — what would have happened anyway — rather than a before-and-after eyeball. If you’re not sure your test can support a causal read, our incrementality calculator is a sane place to start.
A Play is a proven experiment, codified
When an experiment wins and the result holds up, you promote it. A play is that winning experiment written down as a procedure someone else could run without you in the room: the trigger (when to use it), the steps, the inputs it needs, and the metric that says it worked. “Add third-party reviews above the fold on high-intent pages” stops being a one-time tweak and becomes a repeatable move.
In the language of the Strategic Taxonomy, a play is a microfoundation — one of the concrete, repeatable routines that a larger capability is actually built from. Capabilities sound abstract; plays are not. A play is the level at which knowledge becomes portable: you can hand it to a teammate, an agency, or an AI agent, and they can execute it the same way twice.
Promotion is the whole game. The discipline isn’t running more tests — it’s deciding which winners earn a permanent spot in your toolkit. That decision is what turns scattered learning into an asset, and it’s why a one-test-at-a-time habit feels busy but never seems to add up.
A Capability is a maintained portfolio of Plays
A capability is the portfolio: the set of plays you keep, maintain, retire, and recombine. “We can reliably turn high-intent traffic into trials” isn’t a single play — it’s a collection of them (proof on the page, the right offer, fast load, a clean form) that you manage as a group.
Here’s the distinction that matters most, drawn straight from dynamic-capabilities theory. Running a known play well is what the literature calls an ordinary capability — valuable, but anyone can eventually learn it. The dynamic capability is changing the portfolio itself when the world shifts: sensing that something has moved, seizing the right response, and transforming how you operate. That sensing-seizing-transforming loop is the subject of our Sense → Seize → Transform piece and the process we run.
The market just moved — that’s the whole point
Consider what AI answer engines are doing to organic search. For years, a solid SEO play — rank a page, earn the click — was a dependable part of many SMBs’ capability. Now a growing share of searches end with an AI-generated answer and no click at all. The play still runs; it just doesn’t produce what it used to.
A business with only ordinary capability keeps executing the old play well and quietly loses ground. A business with dynamic capability notices the shift, runs new experiments (how do we get cited inside the answer?), promotes the winners into new plays, and rebalances the portfolio — retiring some old moves, adding new ones. Same loop, applied to a changed environment. That reconfiguration, not the day-to-day execution, is the durable advantage.
What an SMB should actually do
You don’t need a research lab to run this spine. You need a habit of promotion and a place to keep your plays. The point is to stop letting wins evaporate and start building a portfolio that gets more valuable every quarter.
Decide what to test next based on impact and effort rather than gut feel — our next-play prioritizer exists for exactly that. And remember that a play that wins for one business can flop for another; heterogeneity is real, which is why you validate plays in your context, not borrow them on faith.
- Run experiments with a real read. One change, one hypothesis, a credible counterfactual — not a before/after glance.
- Promote winners into plays. Write the trigger, steps, inputs, and success metric so anyone (or any agent) can run it again.
- Maintain the portfolio. Keep a living list of plays; retire ones that stop working as the market changes.
- Watch for shifts, then reconfigure. When something like AI search moves, change the portfolio — don’t just run the old play harder.
Keep reading
Questions, answered
What’s the difference between an experiment and a play?
An experiment is a single, time-boxed test with a hypothesis and a way to read the result — you keep it or discard it. A play is an experiment that won and proved durable, then got written down as a reusable procedure: the trigger, the steps, the inputs, and the metric that says it worked. Experiments are disposable by design; plays are the part you keep.
Why is running a play well not a competitive advantage?
Executing a known play reliably is what dynamic-capabilities theory calls an ordinary capability — it’s valuable, but competitors can eventually learn the same play. The durable advantage is the dynamic capability: sensing when the market has shifted and reconfiguring your whole portfolio of plays in response. The edge is in changing the portfolio, not just running it.
How does a small business start building this without a big team?
Start with one habit: after every test that wins, write it down as a play before you move on. Keep a single living list of your plays. Prioritise your next experiment by impact and effort rather than gut feel — the next-play prioritizer helps — and re-validate borrowed plays in your own context, since the same move performs differently across businesses.
More from this series
Seven plain-English pieces on measuring marketing and turning tests into durable capability:
Want to know whether your marketing is the reason?
One conversation to map what to measure and how we’d prove it.