Entrepreneurs know this, and still do it wrong.

“I can test my way to success,” they say – and they’re right.

Then they go off and test a new button color on their website.

I’ve tested my way to 2x profit with lower acquisition cost, optimizing sales scripts to lift sales 3.2x, incentivizing employees to stay 1.6x as long, job postings to get >11x as many applicants. I’ve also run hundreds of “failed” tests (more on that in a minute) –

But no matter how sexy my good outcomes sound, the most relevant thing was they were all the most important test to run at that time. The right test to run is the one that’s going to be most impactful on the business now:

• Finding a new path that can unlock the next success in the business, or
• Ruling out a choice before you invest too much in it (this is grossly under-appreciated)

Here’s how I think about what to test next: at any given time in your business there’s ONE number which, if it changed, would make the biggest impact. It could be the cost to acquire a customer, repeat purchase rate, sales close rate, days between purchase and delivery, or a thousand other things.

That one number is the One Metric That Matters (OMTM).

At all times, you should know your OMTM. It’s the single biggest lever you have on the business.

A good test is one that is likely to make a big impact on the OMTM.

So we dump all the things we could test on the table, and we look at:

• Impact of Trying: how big an impact will this test have on the OMTM?
• Cost of Trying: what’s the cost of trying this?
• Ease of Trying: is this easy to try or will it be super complex to do?

Now we can make smart decisions about what to try.

But wait.

Not all tests are created equal. You could have one test that would have an insane impact … But it will take 7 months to see if it works. You might balance that against another test that doesn’t have nearly as big an impact – but you can run it in a week… And the compounding of many smaller tests might get you much further.

That’s where another component comes in:

• Time to Try: how long will it take to know if this works or doesn’t work?

Yes, you can test your way to success.

The important thing is to have a framework for doing it.

Otherwise you’ll end up testing button colors and think you’re moving the needle.