Most brands delay testing for the wrong reasons. Here's the actual math on when an experimentation program pencils out and what to look at if you're not sure you qualify.

Two product page cards showing A/B price test variants — $89 vs $79 — representing what a real ecommerce experiment looks like

Every decision you make about your store (content, prices, offers, shipping policy) is a hypothesis you're testing whether you measure it or not. A/B testing just adds measurement to decisions you're already making. The question of when to start comes down to whether the math justifies it, and for most brands doing real volume, it does. Here's how to figure out where you stand.

Does the Math Work for You?

Try a quick exercise. Take your annual revenue and multiply it by 3%. This isn't a formula, it's a conservative estimate of what a well-run testing program could realistically move. If the number you just came up with justifies the time and resources, you're probably ready to start.

At $5 million, that's $150,000. At $3 million, it's $90,000. At that scale, you're not deciding whether to build a testing program. You're deciding how much longer you're willing to go without one. Every month you wait is real money you're not recovering.

If you're earlier stage, testing still makes sense, but use it to stress-test your bigger bets. Not button colors or headline tweaks. Pricing structure, shipping policy, core offer. The tests worth running at lower volume are the ones where the answer actually changes how you operate.

Alex and Adam cover this exact question in the IntelliJAMS pod — how big you need to be, how many orders, and when the math actually works.

The Traffic Question Is Really an Order Question

The most common version of "I don't think I'm ready" is about traffic. "We don't have enough visitors."

The issue with thinking in sessions is that sessions include everyone: bounces, window shoppers, people who clicked and left. The signal you're actually building toward is purchases. A test with 10,000 sessions but 50 orders is going to take a long time to tell you anything useful.

This is where AOV creates a trap that's easy to miss. A furniture brand selling $5,000 couches can be doing $5 million a year on just 100 orders a month. At that order volume, small incremental tests won't give you reliable signal. But a meaningful shift in pricing structure or shipping policy might. That's the test worth running.

The rule of thumb: if you're doing 500+ orders a month, you have enough volume to run meaningful tests and get reliable signal. Below that, it depends heavily on what you're testing.

If Your Volume Is Lower, Take Bigger Swings

Thinner order volume doesn't mean you can't test. It means you have to be more deliberate about what you test.

Don't run a test on a landing page that 5% of your traffic sees. Test something that every visitor (or every buyer) encounters: your cart page, your checkout, your pricing structure, your shipping policy. Broad coverage means you accumulate signal faster.

The same principle applies to the size of the change. What counts as a "big swing" depends on what you sell. For a $5,000 product, a $10 shipping difference is noise. A 10 to 15% price test is the swing worth taking. For a lower-AOV brand, a shipping threshold test will often wrap in two or three weeks. When volume is lower, take bigger swings. Go for changes that produce a visible response in the data you do have, not ones that need thousands of orders to detect.

When You Qualify But Feel Like You Don't

If the revenue is there but you feel unready (no dedicated team, no developer, no CRO specialist), that's a different problem. And it's increasingly a solvable one.

A well-resourced experimentation program from five years ago might have meant a CRO specialist, a developer, a data analyst, and a project manager. You might not need that anymore. AI can handle a lot of the work that used to require dedicated headcount. What you need is one person to operate the program, someone who can think through hypotheses, read results, and make calls. The rest can be AI-assisted.

What You're Actually Waiting On

If the revenue is there and the order volume is there, the thing most brands are really waiting on is confidence. Confidence that they'll set tests up correctly, that the results will be meaningful, that the time invested will pay off.

That's understandable. But it's worth recognizing that every day you run your store without testing is a day you're making decisions blind. Every A/B test you don't run is a closed learning window. You'll never know what your customers actually responded to, whether that price change helped or hurt, whether that offer structure was worth it.

The brands that build compound advantages from testing don't do it by waiting until everything feels ready. They start with one test, on one high-volume decision, and learn from it. Then they run another one.

Stop Guessing. Start Knowing.