.webp?dpl=dpl_8W926EHwhAS2kVGZWpiQ95f9JKkq)
How to interview for a role you cannot do yourself.
Your first salesperson, your first marketer, your first engineer. You know what the role has to produce, and you have never done the work. Every candidate sounds plausible, and the one who sounds best is the one who has interviewed most. This is how to run an interview that produces evidence you can read without being an expert.
Why the usual interview fails you here
The advice you will find is “never hire someone to do work you have not done yourself.” It is good advice for a company that has the option. A founder who has written the product and now needs someone to sell it does not, and neither does the sales leader making the first marketing hire, or the marketer who needs a first engineer. The role exists precisely because nobody on the team can do it.
So the interview goes the only way it can. You ask the candidate how they would approach the job, they explain it fluently, and you judge the explanation the only way a non-expert can: by how confident it sounded. That is a test of interviewing, and the candidates who have done the most interviews pass it. You end up hiring the person who was best at describing the work, which is a different skill from doing it.
The fix is not to become an expert in the next two weeks. It is to change what the interview asks for. Opinions about how the job should be done require expertise to judge. Specific accounts of what the candidate actually did, followed down to the detail and scored against a written standard, do not. Detail is something anyone can hear, and it is the thing rehearsed answers run out of first.
The method: five moves
- 1. Write down what the role has to produce. This is the part you know, because it is your business. Not “a great salesperson” but “close the first twenty customers from inbound and build an outbound motion by month nine.” Not “a strong engineer” but “ship the integration our largest prospect is waiting on and take deploys off the founder’s plate.” Every question and every score comes back to this sentence. If you cannot write it, you are not ready to interview.
- 2. Get the questions written by someone who knows the job. You are not the right person to decide what to ask. Borrow an hour from an expert (an investor’s sales leader, an engineer you trust, a marketer two companies ahead of you) and have them turn your outcome sentence into six to eight questions about specific things a candidate has done. Or describe the role in Yardstick and get the interview guide designed for it: the questions, the follow-ups, and a scorecard with the anchors written. Either way, the questions ask for past work, not opinions: the last deal they closed that started cold, the last system they built that broke, the last campaign that missed.
- 3. Ask for the story, then follow the detail. When the candidate gives you the story, keep going: who was in the room, what did they decide, what nearly went wrong, what was the number, what would they do differently. A real story gets more specific the further you follow it. A rehearsed one gets vaguer, or turns back into principles. You do not need to know whether the decision was the right one; you need to know whether it was theirs and whether they can account for it. How to ask follow-up questions gives you the seven directions to follow.
- 4. Score every candidate on the same scale, against written anchors. Before the first interview, each criterion on the scorecard has a sentence for each score describing what that level of evidence sounds like. Then you score the answer against the sentence, not against your impression of the person. This is what turns a non-expert’s interview into evidence: you are matching what you heard to a description an expert wrote, and every candidate is measured on the same standard. The scorecard method covers how to write the anchors.
- 5. Put a second interviewer on your finalist. Ideally the expert who wrote your questions. Give them the same guide and the same scorecard, and have them score before you compare notes. Where the two scorecards agree, you have evidence. Where they split, ask what they heard that you did not; that conversation is where you learn what the role actually needs. The moment there are two interviewers, the same questions and the same scale stop being a nicety and become the only way the two conversations can be compared.
A first sales hire, worked through
Two technical founders sell a developer tool at around $20,000 a year per customer. Inbound has produced the first fifteen customers, all closed by the CEO, badly. They are hiring their first account executive. Neither founder has ever carried a quota. The company and the people are invented; the shape of the process is the point.
The outcome. The CEO writes: “Close the inbound we already get at twice my rate, and by month nine have an outbound motion that produces half of new pipeline.” That sentence is the job.
The guide. She describes the role in Yardstick: the product, the price, the buyer (an engineering manager), the sales cycle so far, and the outcome sentence. The interview guide comes back with questions written for that motion. One of them: “Walk me through the last deal you closed that started with no inbound signal at all. How did you find them, who did you talk to first, and what nearly killed it?” Under it, the follow-ups: how long from first touch to close, what the contract was worth, who the economic buyer was, what the candidate changed after the first call. The scorecard has a criterion called “runs discovery,” and its anchors read like this:
Not enough information. The candidate did not describe a real discovery conversation, or the follow-ups did not get past generalities.
Talked about the product. Discovery meant presenting features and asking whether the buyer was interested.
Asked questions. Can name the questions they asked, but the answers did not change what they did next.
Found a specific problem. Named the buyer's problem in the buyer's words and can say what it was costing them.
Found the problem, put a number on it, and changed the deal because of it: a different champion, a different scope, or a walk-away. Can say what the number was and who owned it.
She could not have written those anchors. She can read them. That is the whole trick.
Two candidates. Marcus is fluent. He talks about “leading with value” and “multi-threading the account,” and the CEO’s instinct after twenty minutes is that this is what a salesperson sounds like. Then the follow-ups. Which deal? “A few last year.” Pick one. He picks one and the contract value is “mid five figures, I think.” Who was the economic buyer? “It was a committee.” What did he change after the first call? A pause, then a principle about listening. Against the anchors, that is a 2: he can name the questions he asks, and nothing in the story turns on an answer he got.
Nora is plainer. Her cold deal was a logistics company, $41,000, fourteen weeks. The first call was with a team lead who wanted a dashboard; on the second call she asked what the dashboard was for and found out the VP was losing a day a week to a manual report. She dropped the dashboard, rewrote the proposal around the report, and got the VP on the next call. She can tell you what the VP said when the number came up. That is a 4, and it took no sales expertise to see it: the story had a number, a person, a decision, and a change of course, and every follow-up made it more specific.
The second interviewer. The CEO’s gut still likes Marcus, which is exactly why the scorecard exists. She sends a magic link to Rafael, a sales leader at a portfolio company who has agreed to an hour. He interviews Nora from the same guide and scores the same anchors. He gives her a 4 on discovery and a 2 on building pipeline from nothing: her outbound experience is real but thin, one quarter at a company with a list already built. The CEO had scored that a 3. They talk about it, and Rafael explains what a 3 would have sounded like. The offer goes to Nora, with the outbound gap named in the plan for her first quarter, and the founders now know one more thing about the job than they did.
What Yardstick does in that process
Yardstick helps with the two hard parts of interviewing for a job you have never done: deciding what the role needs, and writing the interview that tests for it. The first is the blank page. You describe the role in a few sentences and who the person will work with, and Yardstick drafts the outcomes for the first year, the skills the person has to bring, and the competencies that predict success in that role. You edit that draft, and editing is the easy part: you know your business better than any model does, and correcting an outcome is far easier than inventing one from nothing.
The second is the interview itself. From those priorities, Yardstick designs the interview guide: the questions, the follow-up prompts, and a scorecard with the anchor for every score written out, so you are scoring against a description of what good sounds like instead of your own impression. It also scores incoming applications against the same priorities, so you can see at a glance which candidates are worth an interview hour instead of reading every resume cold.
Every candidate is interviewed from the same guide and scored on the same scale, so Marcus and Nora sit side by side on one scorecard rather than in two different memories of two different conversations. The second interviewer opens their scorecard on a magic link, with no paid seat to buy, and scores before they see yours. Once the scorecards are in, the built-in agent Prepare a panel decision brief pulls them into one brief for you to review; the decision is yours.
The rest of the hiring loop is there when you need it: a careers page and application intake, a pipeline for the candidates, scheduling on the Cal.com or Calendly account you already have. If you run your company from a coding agent, it can operate Yardstick through the CLI or MCP and draft the guide from your notes while you work. Your first three Jobs are free, and after that you pay per open role rather than per seat, so the friend who interviews your finalist never costs you anything.
When it is your first engineer, or your first marketer
The moves are the same; the outcome sentence and the expert change. Three notes from the roles founders most often ask about.
First engineer, non-technical founder
The outcome is a thing shipped by a date, in the environment you have, with the founder out of the deploy path. The questions ask for the last system they built that they would build differently now, and why; a real answer names the trade-off they got wrong and what it cost. Add a short work sample tied to your actual product, an hour at most, and have your borrowed engineer judge it on the same anchors you use for the interview. Nobody has reviewed your code before; this is the hire who will, so ask them to review a piece of it in the interview and listen to whether the feedback is specific.
First marketer
The outcome is a number in a channel: qualified pipeline from content, or trials from paid, by a date. The trap is hiring the person who describes marketing best. Ask for the last campaign that missed and what they changed; ask for the last one that worked and how they knew. Real marketers carry the numbers with them, including the bad ones, and can say what they would have done with half the budget. If you are hiring a freelancer first, run the same interview and score the same anchors; the scorecard is how you decide later whether to make the role full time.
First salesperson, in more depth
The example above is the short version. How to hire salespeople covers the full process for a sales hire, and how to build a hiring process is the general version you will run once there is a second role open. When you have the finalist, how to make a job offer is the last step, and it opens with the three things your scorecards agreed on.
Frequently asked questions
How do I hire for a job I cannot do myself?
Interview for evidence, not for expertise. Write down what the role has to produce in its first year, because that is the part you do know. Get the interview questions written by someone who knows the job, whether that is an expert you borrow for an hour or an interview guide designed for the role. Ask every candidate for specific things they have done and follow the detail until you can picture it. Score every candidate on the same scale against written anchors, and bring a second interviewer in on your finalist. You are not judging whether the answer is technically right; you are judging whether the story is real, specific, and tied to a result.
How do you hire people who are better than you at the job?
You stop trying to evaluate their craft and evaluate their track record instead. Someone who is genuinely better than you can describe, in detail, what they did, why, what went wrong, and what the result was, and the detail holds up when you follow it. Someone who is not will stay at the level of principles and vocabulary. Ask for the last time they did the specific thing this role needs, then keep asking what happened next. Depth of detail is something a non-expert can hear.
I am a non-technical founder. How do I make my first technical hire?
The same way, with one addition: borrow an engineer for the technical part. Define what the first engineer has to ship and by when, run the interview yourself on questions written for that role (what they built, the decisions they made, what broke, what they would change), and score on written anchors. Then have an engineer you trust interview your finalist from the same guide and score on the same scale. Compare the two scorecards. Where you agree, you have evidence; where you split, ask them what they heard that you did not.
What shortcuts can I use to tell if a person can do the job?
There are three that hold up. Ask for a specific past example rather than an opinion, then follow the detail; rehearsed answers run out of detail and real ones do not. Give a short work sample tied to the actual job, such as running a discovery call on your product or reviewing a small piece of your code, and judge it with someone who knows the work. And use the same questions and the same scale for every candidate, so you compare people to each other instead of to your mood on the day.
Should I have an expert interview my top candidate?
Yes, and give them the same interview guide and the same scorecard you used. An expert who interviews freestyle tells you whether they liked the candidate. An expert who scores the same anchors you scored tells you whether the candidate's answers meant what you thought they meant. One hour of a friend's time on your finalist, scored on your scale, is the cheapest insurance in the process.
Do I need hiring software for my first hire?
You need the interview designed for the role, a scorecard with written anchors, and a way for a second interviewer to score the same candidate on the same scale. A spreadsheet can hold the candidate list. What it cannot do is write the questions for a job you have never done. That is the part worth getting help with, and it is why Yardstick's first three Jobs are free: the interview guide and scorecard for this role, and a magic link for the person you bring in to interview your finalist.
Get the interview designed for the role.
Describe the hire you need to make, get the interview guide and the anchored scorecard for it, and score every candidate on the same scale. Your first three Jobs are free.