The mistake I won't make running another team
This still bothers me
Scale your growth team in 2026 with Hire Overseas. One of my former coaching clients Philip launched a business to help growth and marketing teams hire exceptional overseas talent - think paid ads, SEO, designers, content creators, and more. If you’re looking to scale your team in 2026, this is my recommendation - hireoverseas.com/value.
Grow faster with Navattic: One of my most impactful growth projects was using interactive demos to drive higher quality signups and more conversions. Navattic publishes a free annual report breaking down data from 40k+ interactive demos. If you’re thinking about adding one to your funnel, start here
A few years ago, I launched a referral program that should have worked.
It had moonshot potential for the business.
And I know I’m biased, but I think the program was pretty solid. We had great offer mechanics. Design and engineering did an awesome job creating a really cool interactive user experience. It looked sweet. And the program launch went about as smoothly as you could possibly hope.
But it didn’t work. At least not in the way that we hoped.
It increased user acquisition. But the numbers were small. Incremental. Not step change growth for the business.
The thing about working in growth is that the first version of something is usually never going to be the most impactful version of it. It’s all about collecting learnings, iterating, and seeing if you can break through your conversion plateaus over time. That’s the game.
And in a referral program, there’s no higher-leverage thing to iterate on than the offer.
IE - what does someone get for referring a friend? What does the friend get for showing up? You can’t guess your way to the most impactful solutions. You test, you watch what people respond to, and you try again.
But I never got the chance.
The engineer I needed was on loan from the core product team. To get them back, I had to go chat with the core product team, make the case all over again, and scope the next iteration super tight so they could have something they could execute quickly. Then add it to their backlog to be prioritized a few sprint cycles down the road.
But by the time I did that, the core product team was already heads down in other long-term feature work.
I didn’t have dedicated resources.
So it took a long time to get another crack at iterating on the program to make it better.
The program ended up okay.
But I still think about how much more it could have been.
For years, I assumed the failure was my fault. That I had bad instincts. Or the wrong strategy. Maybe I’d missed something obvious that a smarter growth operator would have caught.
But since then, I’ve coached more than 90+ other growth operators, and many of them tell me some version of this story. Their numbers aren’t moving. They start to wonder if they’re focused on the wrong things, if someone smarter would be solving it differently.
Most of the time, they had the right ideas. They just couldn’t execute on them. The root cause of that, many times, was team structure.
You can’t win in growth by playing it safe.
Even the best strategy in the world won’t get you there if you’re playing it safe. To play big, you have to try a lot of stuff. To try a lot of stuff, you need a team structure that lets you. Strategy is the most important thing. Team structure is right behind it. It’s what actually decides how the work gets done.
Here’s what I mean.
A growth team needs 5 functions to run. A growth PM who owns the in-product roadmap. A growth marketer who runs the levers outside the product. Engineers who can actually ship to the product and the site. A designer who can move at the pace of testing. Someone on the data side who measures the right things and highlights the insights.
You don’t always need 5 people.
At an early-stage company, 2 generalists can cover most of it. Especially right now, with this wave of AI tools, we’re able to cover more surface area than ever.
Most growth teams depend on all 5 of those “jobs to be done.”
But typically, your growth lead isn’t the direct manager for each of the contributors. The growth team has a roadmap, but the people who execute it report to someone else.
That’s the core tension. And it’s why the team structure matters more than most leaders realize.
There are 2 ways to resolve it.
The first is the shared resources model.
You have a growth strategy, but you borrow the people. The engineer who builds your test has other 1-2 stakeholders. The designer has 3. The data person is prioritizing tickets from the whole company.
That’s the model I had at Postscript. On paper, it’s workable. In practice, it forces a specific kind of behavior, and that behavior is what kills your numbers.
When you only get 3 or 4 at-bats per quarter, you can’t afford to ship something that might miss. So you over-scope. You over-validate. You only push the bets you’re sure about. The 1-in-10 swing with 10x upside never gets greenlit, because you can’t risk burning a engineering ticket on something that might fail.
The cost shows up downstream.
Fewer reps. Fewer learnings. Smaller swings. Less impact. The actual problem is the system isn’t giving them enough chances to find the version that works.
The second is the dedicated model.
The growth team has its own engineers, its own designer, its own PM. Some resources still get pooled (BI usually does), but the core of the team belongs to growth. You get 30 at-bats per quarter instead of 4. You can take a few moonshots.
You can ship the weird idea on a Tuesday and see what happens.
That’s the model I had at Wistia, and it’s the one I push almost every coaching client toward. We started lean. Me as the PM, 2 engineers, part of a designer, part of an analyst. Not a huge team. But they were ours. Most of the work that actually moved the numbers came from experiments I never would have green-lit in a shared model. The 1-in-10 bets. The ideas that felt embarrassing to scope.
Iteration is the whole game in growth. Dedicated resources help you iterate more.
Most growth leaders don’t get to pick their model. They get assigned one from their execs. The job is figuring out how to earn your way from shared into dedicated.
If you’re early in the role, the shared model is fine for now.
Use it to get small wins on the board. Leverage AI and low-code tools wherever possible. Make sure leadership can see you driving real outcomes. But if you’ve been in the seat for 2 or 3 quarters and you have wins to point to, push hard for dedicated.
The longer you stay comfortable in the shared model, the harder it becomes to switch.
Here’s how I’d run that conversation.
Don’t walk in asking for headcount. Headcount is what executives say no to. Walk in with a specific part of the growth model that hasn’t moved in 2 quarters. Activation. Onboarding. Referrals. Whatever it is.
Show them the at-bat math. Right now, with shared resources, you’re getting maybe 4 shots at this problem per quarter. With a dedicated engineer and designer, you’re getting 30. More reps means finding the winning version faster.
Finding the winning version faster means the number moves sooner.
Then flip it. What’s it costing the business to stay in the current model? Every quarter that part of the funnel isn’t moving is a quarter of compounding losses. Put a real number on it, even a rough one. Executives respond to what’s being left on the table.
That’s the case. Here’s the part of the model that’s broken. Here’s why we can’t fix it from where I’m sitting. Here’s what staying stuck is costing us. Here’s what would change with dedicated resources.
If I’d known to frame the referral program conversation that way at Postscript, I think the whole story plays out differently. I didn’t. I do now.
That’s the playbook:
When growth isn’t working, don’t immediately rework your strategy. Check the structure too.
You can’t win in growth by playing it safe. And the wrong team structure forces you to.
The shared model is fine to start. The dedicated model is what you should be earning toward.
When you make the case, lead with the part of the growth model that’s broken and what’s being left on the table. Headcount is what executives say no to.
I think about that referral program a lot. Not because of the failure itself. I still think it was a good bet. But the structure was wrong. And for a long time, I blamed the wrong thing.
-Andrew
ps I filmed a 12-minute video walking through both team structures, the tradeoffs, and exactly how to make the case to your execs (including the conversation script I wish I’d had).
pps If you're struggling to have the impact and influence that you want and you can't tell whether it's strategy, structure, or something else, that's exactly the kind of thing I work through with coaching clients. I've got 2 spots opening up in May. Get started here