How to Interview a Growth Marketer When You Are Not One

A founder I know hired the most impressive of four candidates. She had the cleanest slides, a ninety-day plan by week, and an answer for everything he asked. Six months and roughly eighty thousand dollars later he had a brand guidelines document, a refreshed website, a content calendar running two posts a week, and forty signups, most of which came from a launch he ran himself before she joined. She was not a fraud and she was not lazy. He had simply run an interview that tested presentation, and then been surprised when what he got was presentation.

This is the trap for any founder who is not a marketer: you are evaluating a skill you cannot perform. In engineering you can read someone's code even if they are better than you. In growth, the artifacts that get shown in interviews are plans and decks, and a plan is free to produce, impossible to falsify in an hour, and completely uncorrelated with whether the person can actually move a number. So the interview has to be built to test something else. Here is what that something else is.

Why the standard interview fails

The standard growth interview asks "what would you do here", and every competent candidate answers it well, because the answer is a genre. Audit the funnel, talk to customers, pick two channels, run experiments, double down on what works. That is the correct answer. It is also the answer you would get from someone who read three blog posts last night, and you have no way to tell the difference from the outside.

The second problem is that you cannot price the plan. When an engineer proposes an architecture you have some sense of whether it is over-built. When a marketer proposes a content engine plus a paid test plus a partnership motion, you have no calibration for what that costs in weeks or what it should return, so the more ambitious plan sounds like the better candidate. Ambition is exactly the wrong thing to select for in your first marketing hire, because the constraint is focus.

The third problem is the one that cost my friend eighty thousand dollars. Nothing in that interview tested whether the person can tell a real signal from a flattering one. That is the actual job, and it is the part of the week that justifies the salary, as what a growth marketer actually does breaks down hour by hour. Everything else is execution you could buy in other forms.

Test three things, not ten

Strip the job down and there are three capabilities worth an interview. Everything else is a proxy for one of them.

Can they reason from evidence to a decision. Not "can they read a dashboard", which is a skill, but can they look at a weak, noisy, low-volume signal and say what they would do about it and what would change their mind. Early-stage growth data is always thin. A person who needs statistical significance before acting will never act at your volumes, and a person who reads meaning into every two-click blip will burn your quarters chasing noise.

Can they sustain a cadence. Growth work compounds only if it keeps happening, and the failure mode at startups is a strong first month followed by drift. This is about temperament rather than talent, and it shows up in whether they talk about systems or about campaigns.

Can they kill things. The single most expensive habit in marketing is keeping a channel alive because stopping feels like admitting failure. You want someone who has killed something they publicly committed to, and can tell you how long it took them to admit it.

Notice what is not on the list: channel-specific expertise. Channels change, and the right one for your business is not knowable in advance, which is the argument in best marketing channels for startups. Hiring a "paid person" or an "SEO person" before you know which lane works is a bet placed by someone with no information.

Questions that actually separate people

Ask about the past, not the future. Past-tense questions are hard to invent under pressure and easy to check for specificity.

"Tell me about something you ran that did not work. How long did it take you to admit it, and what did you do next?" This is the highest-yield question in the set. A good answer has an uncomfortable amount of detail and an honest timeline, usually longer than the person would like. A bad answer is either "I have not really had that" or a failure so tidy it is obviously a rehearsed strength.

"Walk me through a number that moved, and how you knew your work caused it." You are listening for whether they distinguish correlation from cause at all. A strong candidate will volunteer the confound before you ask, something like "traffic also rose that month because of a press hit, so I looked at the segment that did not see the campaign". A weak one presents the number as self-evidently theirs.

"What did you kill last year, and who disagreed?" Killing is a decision with social cost, so the answer tells you both about their judgment and about whether they can hold a position.

"What would you need to know about our business before you could tell me which channel to start with?" The answer you want is a list of questions back at you. The answer you do not want is a channel.

Then, if you can, give them a small paid exercise on the real product. A few hours, paid at their rate, with a defined scope: one channel, one audience, one message, and the reasoning for each. You are not grading it as an expert, because you are not one. You are checking three things you can absolutely judge: did they ask you anything before starting, is the reasoning specific to your business rather than transferable to any company, and did they tell you what they were unsure about.

The honest limit

An interview cannot tell you whether someone will be good at your company. It filters out the clearly wrong and gives you a slightly better than even bet on the rest, and anyone who tells you their process is more reliable than that is selling something.

There are two things it particularly cannot tell you. Whether they will still be running the cadence in month four, which is the failure mode that actually matters, and whether they can work without the structure a bigger company gave them. Plenty of strong marketers were strong because the machine around them was strong.

So the real protection is not a better interview. It is a shorter feedback loop: hire into a defined first project with a number attached, three months, one channel, and a specific outcome you agreed in advance. If you cannot write that brief, the honest conclusion is that you are not ready to hire, and the reason is in how to get your first customers. Before the message is found, no hire can find it for you, because the knowledge lives at the intersection of your product and your buyer and you are the only person standing there.

If you just shipped an app

If you built something with an AI coding agent in a few weeks and now you are staring at an empty dashboard, this post may be premature for you, and it is worth saying so plainly.

The instinct to hire is a reasonable response to a problem you do not know how to solve. But you are pre-message, which means you would be hiring someone to answer a question only you can answer, and you would be paying a judgment salary for it. What you actually need first is a handful of real conversations with people who have the problem, and then a way to run the repetitive execution without it eating the hours you use to build. The four options for that, and which stage each suits, are laid out in do you need a marketing agency, and the direct hire-versus-software comparison is in AI growth agent vs hiring a growth marketer.

Where Revnu fits

Revnu is the option you reach for when the bottleneck is execution rather than judgment. You connect your product, site, and Slack or iMessage, and it runs the repeatable half of the week across SEO, outbound, and ads, carrying what it learns in one channel into the others. It does not replace the three capabilities above, and it is not pretending to: deciding what to test, reading a thin signal honestly, and killing a channel are yours. Every draft waits for your approval. What it removes is the reason a founder starts interviewing in the first place, which is usually not "I need a strategist" but "nothing shipped again this week".

Where this leaves you

Interview for the past, not the plan. Ask what failed, what moved and why they believe it, and what they killed. Give a small paid exercise on the real product and judge the questions they ask rather than the output they produce. Then protect yourself with a short first project that has a number attached, because no interview is reliable enough to bet a year on. And if you cannot write that first brief with a straight face, that is the most useful thing the process will tell you, and it is worth more than any hire you could make this month.