Founders waste a lot of time on hiring. They either do it randomly with no process or they copy the elaborate rituals of big companies. Both are wrong for a startup. A startup’s hiring process should be like its product: simple effective and designed to get the most important things right.
You don’t need eight interview stages applicant tracking software or brain teasers. You need a repeatable loop that tests for what actually matters. A bad hire is the most expensive mistake a startup can make. Not just in salary but in lost time and team morale. A good process prevents that.
Here is a simple five step loop for hiring your first ten employees. It is designed to be run by a founder or a small team. It prioritizes signal over noise.
Before you write a job description you must write a scorecard. A job description lists qualifications. A scorecard defines success. It forces you to be clear about what you actually need. Most hiring mistakes start with a vague idea of the role.
A good scorecard is short. It should have three parts.
Mission: A single sentence explaining why the role exists.
Outcomes: Three to five specific measurable things this person will achieve in their first six to twelve months.
Competencies: Three to five essential skills or traits. Not a wish list. These are the non negotiable things like “writes clean Python” or “can talk to customers without making them angry”.
Writing this down takes maybe an hour. But it will save you dozens of hours of interviews with the wrong people. The scorecard becomes your single source of truth for evaluating every candidate.
Your first instinct might be to post the job everywhere. Resist this. You will get hundreds of applicants and most will be noise. Your goal is not to get the most candidates. It’s to get the best ones.
Start with your network. Not just people you know but people they know. Send a short personal email to twenty smart people you respect. Tell them what you are looking for and include the scorecard’s mission. Ask them who is the best person they know for this.
After your network you can post to one or two high signal communities. Places where the kind of people you want to hire already hang out. This could be a specific industry newsletter a niche subreddit or the Hacker News “Who is Hiring?” thread. Be targeted.
The goal of a phone screen is not to test technical skills. It is a filter for enthusiasm and communication. It should take 30 minutes. You are trying to answer two questions: are they genuinely interested in your company and can they explain what they have done clearly.
Ask them to walk you through a project they are proud of. What was the goal? What did they do specifically? What was hard about it? Listen to their answers. Do they sound excited? Do they take ownership or do they just talk about what their team did? Clear thinking usually results in clear speaking. If they cannot explain their own past work they will not be able to communicate well inside your company.
This is the most important step. The best way to know if someone can do the job is to watch them do a small version of the job. This replaces useless whiteboard interviews and credential checking.
The work sample should be a small self contained project that takes two to four hours. It should be as close to real work as possible. For an engineer it could be fixing a small bug in your open source library or building a tiny feature. For a designer it could be critiquing a user flow and suggesting improvements.
Always pay them for this time. It shows you respect their work and it gets you much higher quality results. A candidate who is willing to do a paid work sample is already showing a high level of interest. Their work will tell you more than any interview question ever could.
If the work sample is good the last step is a conversation about working together. This is where you and maybe one other team member talk with the candidate for an hour. The goal is to ensure alignment.
Discuss the scorecard outcomes. Ask them how they would approach these challenges. Talk about your company’s culture its flaws and your expectations. This is the time for brutal honesty on both sides. You’re not selling them a perfect dream. You’re starting a partnership.
This is also where you check for the things that are hard to test in a work sample. As I've said before when hiring your first engineer you’re looking for slope not just current ability. You want people who learn fast and are driven.
This five step process forces you to be disciplined. You define what you need upfront. You look for candidates in the right places. You filter for interest and then you test for skill directly. It’s more work upfront than just posting a job and winging the interviews. But it is vastly more effective at finding the right people and avoiding the wrong ones.
Your hiring process is a product. You should build it and iterate on it like one. This loop is your version one.
— Rishi Banerjee
September 2025