Jordan Hwang on Experimenting to Learn, Not to Win at OpenPhone

Test Small Enough That No One Needs to Know

Most growth leaders overthink getting buy-in for new channel tests. Jordan Hwang has a simpler approach: test at such a small scale that the spend is inconsequential relative to the overall budget.

"Do people have to know that you're testing a new channel? In an ideal world, they kind of don't," he says. "I mean that not as in like you're trying to hide it from them, but I mean that as in like when you're testing at such a small scale that the amount of spend that's required is fairly like inconsequential relative to the overall kind of like pie."

The exception is larger offline buys or top-of-funnel investments that require meaningful capital upfront. For those, Hwang paints a different picture: growth isn't a clean line up and to the right. It's multiple programs stacking on top of each other, with some dying off as new ones ramp.

What that actually looks like is like it's multiple programs that are stacking on top of each other. And effectively, if we've done this really well, then what that means is that as your initial programs like start to like saturate out or they start to flatline, a new program is coming up and taking on that charge.

— Jordan Hwang

If you can describe that expectation clearly, you earn the space to test without constant approval loops.

Trust Isn't About Being Right—It's About Being Honest

Hwang draws a sharp line between two kinds of trust. The first is transparency: shooting straight whether the news is good or bad. The second is competency: proving you understand what you're doing through logical hypotheses grounded in data or analogous wins.

The harder part is getting business partners who think about numbers the same way you do. Some leaders demand perfect attribution and one-to-one tagging. Others are comfortable making inferences from mixed media models or observing that when you spend here, results move there.

"Do they take that or not?" Hwang asks. "We've all probably worked with business partners who really want the most literal version of this, which can make it very painstaking to work with versus ones who are able to buy into the logic, buy into kind of like that train of thought and then sit there and say, yeah, yeah, I see it."

He wishes he'd vetted for this earlier in his career. The difference between partners who trust inference and those who don't determines whether you can move fast or get stuck in endless measurement debates.

Disagree and Commit Means Owning the Risk

Most people misunderstand disagree and commit. It's not "they didn't listen to me, but I have to do it anyway." It's "I hear your concern, I think it's still worth the risk, and here's how we'll monitor it."

Disagree and commit is effectively like, I hear you, it's a risk. You don't think it's worth the risk. I think it's still worth the risk. So, we're going to go ahead and do this. But we are going to be very eyes wide open about these risks and continuing to monitor them so that we can deal with them appropriately if they do come up in a meaningful way.

— Jordan Hwang

Taking an anti-risk posture early feels safe, but it leaves you unprepared when growth plateaus. Then you're forced to take outsized risks under pressure, with no runway and no safety net.

Growth programs fail for two reasons, according to Hwang. First, they stop testing and paint themselves into a corner where the only option is a bet that's too big. Second, they run out of time or money, so there's no room for iteration. One shot, no outs.

Why Most B2B Positioning Misses the Human

At OpenPhone, the product integrates with Salesforce, HubSpot, and Slack. But Hwang doesn't lead with integrations. He starts with the buyer: a small business owner doing five to ten jobs, expert at none, deeply afraid of screwing something up.

That fear shapes everything. They want tools that are easy to use because they don't have time to learn. They want guarantees in high-penalty areas like payroll taxes or compliance. And they take deep pride in delivering what they promised to customers—until scale makes that impossible.

"All of them basically like really take up a lot of pride in being able to deliver what they promised that they're going to deliver," Hwang says. "And more often than not, that starts to fall away, not because they lose that sense of desire, but because it's actually so hard to maintain."

OpenPhone's core value prop isn't the integrations. It's shared phone numbers that let teams stay on the same page with customers—texting, calling, tagging each other in—all from one number. The CRM integration is table stakes for larger customers, but the emotional hook is collaboration without chaos.

Experiment to Learn, Not to Win

Hwang's biggest frustration with experimentation culture? People who experiment to win instead of experimenting to learn. Quality of experiment design matters more than volume, but it's wildly underrated.

A healthy test starts with a hypothesis grounded in insight, not a hunch. It's structured to give a clear signal, not just to prove someone right. And it includes guardrails: known risks, monitoring plans, and exit criteria if things go sideways.

Something that gets— that I find generally to be like really underrated is like quality of experiment design. And largely because of two things. One of them is, and I think we've come across this a lot, like we've all come across this, is like people who experiment to win versus experimenting to learn.

— Jordan Hwang

The best experiments don't chase noise. They ladder into a broader strategy, informed by what you've already learned. And they earn you the autonomy to keep testing—because you've built trust by being transparent, competent, and willing to own the outcome either way.