Sandy Diao on Why Being Data-Driven Can Kill Your Growth

The Conversion Rate Paradox

When Descript finally launched its web version after two years of being desktop-only, Sandy Diao was ready to take a vacation. She'd been staring at the funnel for months, watching users drop off between signup and app download. The data was clear: remove friction, boost conversions. Web would be the step function increase that would blow up user acquisition and triple conversion rates. The growth team would coast to their targets.

Instead, conversion rates dropped across the board. Activation went down. Monetization went down. Revenue went down.

The context we didn't understand is that when somebody—think about yourself in the shoes of someone who's downloading a desktop app. Somebody who spent the time to install an app is going to sign up. They were going to give it a try. They were going to get pretty deep into the product flows because they just invested all this time upfront in setting up the product.

— Sandy Diao

This is the central tension in Diao's approach to growth: data-driven companies fail because they worship the chart without understanding the customer. Descript wasn't a data-driven company. It was data-inspired. The difference saved them from chasing dozens more dead-end optimizations that looked brilliant in a funnel visualization but collapsed on contact with actual human behavior.

The 500 Support Tickets That Built a Career

Diao didn't study growth marketing. She studied finance, took all the right classes, and was tracking toward investment banking because "all the smart people who make a lot of money around me are becoming investment bankers." Then a scholarship brought her to Sand Hill Road, where she met venture capitalists who had put the first dollars into Apple and Facebook. She did a complete 180.

Back in her dorm room, she pulled open a website she'd recently discovered: www.pinterest.com. It solved a problem she personally had—a scattered desktop of images she'd organize into folders, zip up, and email to friends. Pinterest made that cloud-based and social. She cold-emailed the small team on Brannan Street in San Francisco. They invited her in to brainstorm. The brainstorm turned into an interview. The interview turned into an offer to "join and help us solve any kind of problem that we're facing."

In her first week, she answered 500 support tickets. Then she started looking at the data around those tickets. What were the most common questions? Who was writing in? The support queue was "ridden with golden nuggets." Businesses wanted to advertise. They wanted analytics proving Pinterest worked for them. Users were confused about how to use the product. Every insight became a growth hypothesis—onboarding flows, monetization experiments, email sequences, product notifications.

This led me to really touching every possible channel under the sun—paid acquisition, email marketing, product notifications, designing in-product onboarding flows, etc., etc. So the reason Pinterest was so formative and the reason I share the story is to kind of share with you a lot of my early career being a growth generalist, doing anything possible to help the company grow.

— Sandy Diao

That generalist foundation meant Diao became the person companies hired when they were ready to build their first growth team or first go-to-market motion. She was always the first to raise her hand.

Why Copying Facebook's Onboarding Didn't Work

The most expensive mistake in growth is looking at a competitor and saying, "This thing really worked well for them. Let's copy it over and do it better." Diao admits this isn't just a founder problem—growth leaders fall into the trap too. At Pinterest, the team was trying to solve onboarding. The product required a mental shift from local file storage to cloud-based social bookmarking. How do you communicate that value to new users?

They looked at Facebook. Facebook had a 7-part onboarding email series and a structured product flow. Facebook served hundreds of millions of users. So Pinterest copied it. They spent weeks reconfiguring product onboarding and email sequences. They launched. No lift.

The problem: Pinterest wasn't serving the same customers Facebook was. The timing was different. The distribution was different. The use case was different. But the team had skipped the two components of a good growth hypothesis—data-inspired insight and customer story.

A good growth hypothesis has two components to it. The first component is an insight, and the insight has to be data-inspired or backed by some kind of context. And then the second part of a good growth hypothesis is they have some kind of story around the customer and your understanding of what that data point means.

— Sandy Diao

Even at Meta, when the company had 3 billion users across its family of apps, the team still looked at Google and thought, "They built a standalone app for small business owners—maybe we should build an app too." Competitive jealousy clouds judgment at every scale.

Channel Fit Is Not About the Channel

Most people assume channels get worse over time. Performance marketing costs go up. Audience saturates. Returns diminish. Diao sees it differently: channels don't fail—you just exhaust the audience you're targeting.

At Descript, if the team was going after the total TAM of podcasters and had maximized reach on paid social, costs would spike and efficacy would drop. But if Descript expanded to short-form content creators, long-form video editors, or people searching for background noise removal, the same channel—paid social—could deliver step function increases again. The same logic applies to SEO. Early on, Descript targeted transcription and podcasting keywords. Later, they could target video editing jobs-to-be-done like cropping or vertical format creation.

In my view, the channel saturation comes less because the channel itself stops working for you, but because you've kind of exhausted one set of audiences. But naturally, as a company grows, the product grows, the markets that we serve grow over time, and our audiences become a lot more generalized.

— Sandy Diao

That insight reshapes how to think about channel fit. You're not looking for channels that work forever—you're looking for conversion rates that hit benchmark for your intended customer and incremental investments that drive proportional returns. Some channels, like PR or a viral influencer hit, are one-off blasts. If you can't replicate them at scale, you don't build a predictable system around them.

The Power Law of Growth Channels

The fastest-growing companies don't diversify early. They don't hedge. They find one or two channels that drive power law outcomes and go deep. The CMO of HubSpot told Diao that HubSpot grew to $50 million in revenue on one channel, then to $100 million on a second. That's concentrated channel strategy, not portfolio theory.

Diao has observed this pattern across every company where she's been the first growth hire: clarity on positioning and messaging leads to clarity on distribution. You don't need breadth. You need the one or two channels that map to where your intended customers already are, and then you maximize them until saturation forces you to expand the audience or add a second channel.

Growth teams get hired too late—usually when growth decelerates or when a hairy problem emerges that the founder doesn't want to solve. The better time to hire is when there are early signs of channel fit. Then you hire someone with an edge in that specific channel who can be a player-coach, executing while building the team around what's already working.

The growth teams that I see that have the most impact are the ones where there are some early signs that there's channel fit with some kind of growth channel that's working, and then you build a team customized in a way around that.

— Sandy Diao

Diao's career has been shaped by showing up when companies are ready to build that foundation for the first time. Pinterest taught her to be a generalist who could touch every channel. Meta and Descript taught her that customer context beats data visualization every time. And every misstep—copying Facebook, assuming web would fix everything, chasing competitor strategies—reinforced the same lesson: growth is not about the funnel. It's about the story behind the numbers.