How to Hire for Roles You've Never Done

There's a specific flavor of expensive mistake that only founders make: paying top dollar for a role you don't understand, getting exactly what you asked for, and not realizing the problem until the damage is done.

Ryan Deiss, the serial entrepreneur behind DigitalMarketer and The Scalable Company, opens his latest video with one of those stories. He paid a CFO $200,000 a year for two years. The guy was, in Deiss's words, "basically just a glorified bookkeeper." The total cost, factoring in salary and the growth left on the table during that period: over a million dollars. The more clarifying detail comes later — if Deiss had simply run a proper job-definition exercise upfront, he says AI would have told him in thirty seconds that he needed a controller, not a CFO.

That gap — between what a founder thinks they're hiring for and what they actually need — is the problem his four-step system tries to close.

---

Step One: Name the Role Before You Post It

The instinct most founders have is to start with a job title. Deiss argues you should start with three lists: the specific tasks the person will perform, the outcomes they'll own, and the metrics they'll be measured by. Then add business context — your industry, company size, revenue range — and feed all of it into an AI prompt asking for appropriate job titles with salary ranges.

The insight isn't just that this produces cleaner job descriptions. It's that the process has a diagnostic function. Deiss notes that you shouldn't be surprised if the AI tells you what you thought was one role is actually two or three. He calls this the "unicorn trap revealing itself" — the moment you realize you've been trying to hire one person to do the work of a small team.

That's a real phenomenon, and it's especially common in founder-led companies where senior roles have been held by the founder themselves, doing everything at once. The problem is that what one person can do in an early-stage company often fragments into multiple specialized functions as the company grows. Founders miss this because they benchmark the role against themselves, not against the market.

---

Step Two: Learn the Role From People Doing It

Getting the job description right is necessary but not sufficient. Deiss's second step is to actually talk to people who currently hold the role — not recruiters, not job seekers, but employed professionals who are winning in the position right now.

His approach: DM them on LinkedIn and offer to pay them for their time. He offered $100 an hour for 30-minute conversations when hiring a content marketing manager. That price signal did two things. It screened out the casual conversations, and it got him access to high-performers who wouldn't otherwise take the call.

The result was informative in a way he didn't expect. He went in thinking he needed a doer — someone who could execute content independently. By the third conversation, the picture had shifted:

"Every single person who was actually performing the job that I wanted to hire for had three to five people working under them. They had writers, they had designers, they had SEO specialists. They weren't doing a lot of this work themselves, they were leading a team that did the work."

That single realization rewrote the job description, the budget, and the candidate profile. It also created a useful filter: anyone who claimed they could do it all themselves "probably didn't actually understand the role."

The five questions Deiss recommends for these conversations are worth keeping:

1. Walk me through a typical week in your role.
2. What resources or team members do you need to be successful?
3. If you were hiring someone to replace you, what would you look for?
4. What does someone failing in this role look like? What are the early warning signs?
5. What should someone in this role accomplish in their first 30, 60, and 90 days?

The fourth question is underrated. Most hiring processes spend all their energy characterizing success. Characterizing failure — specifically and early — is different information, and often more useful.

---

Step Three: Bring an Expert Into the Room

Knowing what good looks like intellectually and being able to spot it in a one-hour interview are, as Deiss puts it, "two very different things." His third step is to solve that gap structurally: get someone on your interview panel who has hired for this role before — not just done it, but hired, managed, and evaluated people in it.

This person brings three things the founder doesn't have: role-specific technical questions, the ability to detect BS dressed up as expertise, and — Deiss argues this is the most important — the authority to give you permission to say no. That last point is subtle but real. Founders often second-guess their gut on unfamiliar hires because they don't trust their own read. An expert in the room resolves the ambiguity.

His three routes to finding that person: ask someone in your CEO peer group or mastermind as a favor, pay a fractional executive or consultant for 3-5 hours, or hire a recruiting firm that has placed people in this exact role before. For the recruiting firm route, he advises doing blind reference checks beyond what the firm provides — looking at their track record at the company level, not just the references they hand you.

---

Step Four: Test the Fit

The final step has two filters. The first is the 30-60-90 day plan. After doing the learning conversations in step two, Deiss builds a rough internal template of what a strong plan looks like. Then he asks each finalist to build their own — and compares the two.

The tells he looks for: specificity to the company (not a generic plan), a genuine learning phase in the first 30 days, resource awareness, and ruthless prioritization. What he's watching for on the negative side:

"You shouldn't expect them to promise to fix everything in the first 90 days. That's not just unrealistic. If that's what they're doing, then it likely has shown some immaturity on their end."

A plan that diverges from the template isn't automatically bad — but it should prompt a direct conversation. The divergence itself becomes a diagnostic tool, both for evaluation and for setting accountability expectations once the person is hired.

The second filter is what his fellow CEO Jonathan Cronstedt calls the tacos and tequila test. Strip away the framing: the underlying logic is about collaboration density. When you're hiring for a role you don't fully understand, you're going to need to spend significant unstructured time with this person — learning together, figuring out how the function fits into the company's broader system. That requires genuine rapport, not just professional tolerance.

Deiss frames it carefully: this isn't about hiring people you like over people who are qualified. It's about recognizing that at the early-to-mid stage of a company, the working relationship itself is load-bearing in a way it isn't once you have professional managers buffering everything.

"There may come a day when the company's big enough that you're hiring purely for competence and merit. You got professional managers all over the place. They can handle people they don't like. But for now, don't sacrifice chemistry."

---

What This System Is Actually Solving

The through-line in Deiss's framework isn't really about hiring tactics. It's about epistemic humility — the structural problem that founders are systematically overconfident in domains they haven't personally worked in, and that the stakes on executive hires are high enough to demand a workaround.

The $200K CFO story is doing a lot of narrative work here, but it's pointing at something genuine. The mistake wasn't negligence; it was the absence of a calibration mechanism. Deiss didn't know what he didn't know. The four-step system is designed to solve for exactly that — to force calibration before the first interview, not after two years of expensive misfires.

There's an open question the system doesn't fully resolve: what happens when the learning conversations and expert panels align on a candidate profile that the founder fundamentally can't evaluate, even after going through all four steps? The system reduces information asymmetry, but it doesn't eliminate it. At some point, founders are still placing a bet on someone they can't fully audit.

That's not a knock on the framework — it's a realistic constraint of the problem. The question is whether you've done enough work to make the bet an informed one.

Most founders, Deiss's million-dollar lesson suggests, haven't.

---

By Alex Volkov