Growth PM

When a company only has people who build and people who bring users in, the stretch between those two jobs belongs to nobody. The product team owned everything up to shipping a feature. The marketing team owned everything up to a completed signup. The moment where someone signs up, opens the product once, and never comes back sat outside both.

Pagedeck ran into exactly that. Ravi spent a month on ads and produced 2,000 signups, and a week later 260 of those people had opened the product again. In the monthly review, marketing said the product wasn't holding anyone, and the product team said the traffic was junk. Both had hit their own targets, so nobody took the missing 1,740 people as their problem.

The growth PM role exists to fill that gap. What makes it different from earlier attempts is that the role gets the numbers after signup and the authority to change screens and features to move them. Those numbers don't respond to more ad spend; they only respond to the product changing, so a target without that authority would leave the person nothing to touch.

A day starts by checking where yesterday's numbers moved and ends by turning an experiment on or off. The starting point is one or two owned metrics rather than a list of features to build, and that is what separates the role from a product manager generally.

The work runs in a six-step loop.

1. Open the owned metric. The usual ones are activation, the share of new signups who reach the point where the product is visibly useful to them, and retention, how many of them keep coming back after that.
2. Walk the stages and find where the most people fall out. If 1,000 signups produce 520 people who create a document and 180 who invite a teammate, this week's problem sits before the invite step, not at it.
3. Write down in one sentence what change should move the number and why. Something like: people leave because they land on an empty screen, so seeding a sample document should raise first-document creation.
4. Run an A/B test, which splits users at random between two versions and compares the results. Half get the empty screen, half get the one with a sample document in it.
5. Call the result when the planned window closes. Did 520 become 600, and is the gap bigger than what chance alone would produce?
6. Turn on what worked and delete what didn't. A flat result still goes into the record, because it says that spot isn't where the problem lives.

That loop runs two or three times a week. Out of ten experiments, three or four move anything, which is why each change stays small. Instead of redrawing a screen, the role touches button copy, the number of signup steps, the timing of the first notification — things that can be built and reversed inside a week. Each of these small, fast tests is a growth experiment.

Almost nothing here can be finished alone, so half the day is spent alongside other roles.

• Theo, engineering. Puts experiments in front of users and pulls them back out. With a mechanism in place for showing a new screen to a slice of users, turning one on takes half a day.
• Noa, design. Produces two or three different versions of the same spot. Several comparable options beat one polished one.
• Ravi, growth marketing. The boundary between the two jobs touches, so they sync weekly. Ads and landing pages are Ravi's; everything after the signup button is the growth PM's.
• The product analyst. Confirms the numbers are trustworthy. One screen missing its tracking is enough to put the drop-off in the wrong place.
• Grace, the CEO. Sets which metric the role owns each quarter. Raising signups and stopping churn call for different work, so that agreement comes first.

Once a company is large enough, these people are sometimes grouped together. A cross-functional team dedicated to moving metrics is a growth team, and the growth PM decides what gets tested first inside it.

"This week's experiment targets a 6 percent cut in drop-off at the second onboarding step."

A single point in the path new users take to settle in, with the definition of success fixed as a number in advance.

"That's roadmap work, so I'm not putting it in my backlog."

Drawing a line: anything that can't produce a result inside a week isn't this role's job.

"The sample is still too small to call. Let's run it two more days."

An early number can look good and still be chance, so don't stop yet.

At Pagedeck, the quarter Mia took this role, the goal was to raise the share of users who return within seven days of signing up from 26 percent to 35 percent. She spent the first two weeks running no experiments at all, because the drop-off had to be located first. People who created no document on their first day almost never returned; half of the people who created one did.

After that she tested only ways to get one document created on day one. She seeded the empty screen with a sample, cut signup from five steps to three, and sent a notification the next morning. Two of the six attempts moved the number, and the quarter closed at 33 percent. Short of the goal, but four documented dead ends were left behind for the next quarter to start from.

The most common mistake is reading the job as marketing under another name. The word growth is in the title and the numbers on the dashboard are signups and conversion rates, so the assumption is easy to make. But what the role actually touches is product screens, not campaigns: fewer fields on the signup form, different wording and timing on the first notification, different content on an empty state. Miss that distinction and you hire on campaign experience and then hand the person product work.

Treating growth as a synonym for more signups has the same shape. The word points at acquisition first. But bringing in 100 new users while 100 existing users leave holds the total flat — pouring water into a leaking bucket, where money spent on acquisition drains out at the same rate. That's why the quarterly target handed to this role is often return visits or churn rather than new signups.