← Writing

Your First Hire: When, Who, and What It Really Costs

Published
Reading
17 min
Contents · 12 sections
  1. When to hire: two tests
  2. What to hire first: do the job yourself first
  3. Employee, contractor, or part-time
  4. What a hire really costs
  5. A worked example: Pinewell
  6. Runway before and after the hire
  7. Write the role around outcomes, not tasks
  8. Interview with real work, not puzzles
  9. Equity for early employees
  10. Week one: make the first days count
  11. “Hire slow, fire fast”
  12. Key ideas

In April 2014, Matt Mullenweg, the founder of Automattic (the company behind WordPress.com), wrote a short piece for Harvard Business Review about how his company hires. In the early years, he explains, Automattic hired the usual way: read the résumé, hold some interviews, make an offer. Over time they found that how well someone did in an interview often said little about how well they did the job.

So they changed the process. Instead of relying on interviews, Automattic started holding what Mullenweg called auditions. Final candidates were paid to do real work on a real project for a period of time, and the decision came from the work. The company has grown a lot since then, and its process has changed shape, but the idea is still there: its hiring page today describes a trial of up to 20 hours using real work, with feedback from the team.

I like this story because it answers the question most first-time founders get wrong. The hard part of a first hire is not finding someone who interviews well. It’s knowing what job you’re hiring for, whether you can afford it, and how to see someone do that job before you commit.

This post walks through all three. If you’re picking a partner rather than an employee, that’s a different decision, and I wrote about it in Choosing a Co-Founder.


When to hire: two tests

Most first hires happen because a founder feels busy. Busy is not a reason. Every founder is busy. I’d use two tests instead, and I’d want both to pass.

Test 1: there is a repeated job, and you are the bottleneck on it.

A bottleneck is the narrow point that slows everything behind it. If customer support emails sit for two days because you’re the only one who answers them, you’re the bottleneck on support. If you could close more sales but you can’t fit more demos into the week, you’re the bottleneck on sales.

The key word is repeated. A one-off project, like a redesign or a data migration, is not a reason to hire a person. It’s a reason to hire help for that project. A hire makes sense when the same kind of work comes back every week and keeps growing.

A good sign: the work you should be doing (talking to customers, selling, building) keeps getting pushed to “next week.”

Test 2: you have the money, by your own runway math.

A salary is not a one-time cost. It’s a new monthly cost that stays until something changes. Before hiring, run the question from Default Alive or Default Dead?: at current growth, with the new cost added, does revenue still catch up with costs before the cash runs out?

If the answer flips from “alive” to “dead” when you add the person, the hire is a bet. Sometimes that’s a fine bet, if the person directly drives revenue. But you should know you’re making it. We’ll work a full example below.


What to hire first: do the job yourself first

Here is the rule I’d follow for a first hire: don’t hire someone to do a job you’ve never done.

Later, hiring for skills you lack makes sense. For the first few hires, if you’ve never done the job, three things go wrong:

  1. You can’t describe it. You’ll write a vague role (“help with growth”) and hire someone to figure it out, which is really asking them to be a founder for employee pay.
  2. You can’t judge candidates. You won’t know a good answer from a confident one.
  3. You can’t tell if it’s working. After three months you won’t know whether results are slow because the person is weak or because the job is hard.

So do it yourself first. Answer the support emails for a few months. Run the sales demos. And while you do, write down how. Not a polished manual. A plain document: the questions customers ask most, the answers that work, the tools you use, the mistakes you made, what “done well” looks like.

That document later becomes the role description, the interview task and the week-one training guide.

There is one sensible exception. If the job needs a specialist skill you don’t have and won’t learn quickly, like accounting, legal work or a specific kind of engineering, start with a contractor on a small, defined piece of work. You’ll learn what the job involves by watching someone good do it.


Employee, contractor, or part-time

A quick note first: this section is general information, not legal or tax advice. The rules differ a lot from country to country, and sometimes within one country. Before you hire, talk to a local accountant or lawyer.

In broad terms, there are three common options:

Option What it usually means Good for Watch out for
Employee (full-time) Works for you on an ongoing basis, under your direction. You usually pay payroll taxes, insurance and benefits required by law. Core, repeated work you want done your way for years. Highest fixed cost. Ending it has legal steps.
Contractor (freelancer) Runs their own business and sells you a result. Usually invoices you; handles their own taxes. Defined projects, specialist skills, testing a role. Treating a contractor like an employee can break the law.
Part-time employee An employee with fewer hours. A repeated job that’s real but not yet 40 hours a week. Still an employee, with most of the same duties.

The trap is the middle row. Many founders call someone a contractor to save on taxes and paperwork, then manage them exactly like an employee: set hours, daily direction, one client, open-ended work. In many countries, the label you choose doesn’t decide the category. The real relationship does.

As one example, the US tax agency, the IRS, looks at three groups of evidence: behavioral control (do you direct how the work is done?), financial control (who handles the business side, like tools and expenses?), and the type of relationship (is it ongoing, with employee-style benefits, and central to your business?). Other countries use different tests, but most ask similar questions.

My view: a contractor is often a good first step for a new role. But if the work becomes ongoing, central and directed by you, set it up properly.


What a hire really costs

The salary is never the whole cost. For scale: the US Bureau of Labor Statistics reported that in June 2026, for private industry workers, benefits made up 30% of total employer compensation costs and wages and salaries made up 70%. Put another way, for every $1 of wages, employers paid about $0.43 more in benefits and legally required costs. That’s the US. Your country’s number will be different, sometimes lower, sometimes higher. Look up yours.

And that’s only the ongoing part. A first hire also brings costs that never show up on payslips:

  • Payroll extras: taxes, social insurance, health insurance, paid leave, whatever your country requires.
  • Tools: a laptop, software seats, accounts.
  • Hiring itself: job posts, paid trials, your hours spent reviewing and interviewing.
  • Ramp time: nobody is fully useful in week one. For the first month or two, you pay full cost for partial output.
  • Your time: training, answering questions, reviewing work. For a while the hire takes founder hours before it gives them back.

A worked example: Pinewell

Pinewell is a fictional company. It’s a two-founder software business. All numbers are invented to show the method, and the arithmetic was checked in Python.

  • Cash in the bank: $150,000
  • Costs today: $16,000 a month
  • Revenue today: $8,000 a month, growing 5% a month

The founders want to hire a support person. Here’s what that person really costs Pinewell, using assumptions that are illustrative, not typical of any country:

Item Amount
Salary $4,000 / month
Payroll taxes and insurance (assumed 20% of salary) $800 / month
Software seats $150 / month
Ongoing monthly cost $4,950 / month
Laptop and setup (one-off) $2,000
Recruiting, including paid trials (one-off) $1,500
First-year cost $62,900

The salary for a year is $48,000. The first-year cost is $62,900, about 1.31 times the salary. And this table still leaves out ramp time and founder hours, which are real but harder to price.


Runway before and after the hire

Now the question that matters: what does this do to Pinewell’s survival?

Simple runway first. Today Pinewell’s net burn (costs minus revenue) is $8,000 a month, so $150,000 lasts about 18.8 months. Add $4,950 a month and net burn becomes $12,950, so runway drops to about 11.6 months. Simple runway ignores growth, though, so let’s model it month by month with three scenarios:

  1. No hire. Revenue keeps growing 5% a month.
  2. Hire, and nothing changes. The person joins in month 1, but revenue still grows 5%.
  3. Hire, and the founders use the freed time well. From month 4, with support off their plate, the founders spend more time selling and growth rises to 9% a month.

Here’s what happens to the bank balance:

Scenario Lowest cash Break-even month Outcome
No hire about $82,600 (month 15) 16 Default alive
Hire, growth stays 5% below zero (runs out first) Out of cash in month 17
Hire, growth rises to 9% about $45,400 (month 13) 14 Default alive, and earlier break-even
Cash in the bank, Pinewell (fictional), 20 months $0 $40k $80k $120k $160k 0 5 10 15 20 month low $83k low $45k out of cash, month 17 No hire Hire, growth stays 5% Hire, growth to 9%
The same hire kills Pinewell if nothing else changes, and saves it time if the founders turn their freed hours into growth.

The lesson: the hire alone doesn’t decide the outcome. What you do with the time it frees does.

Spend the freed hours on meetings and Pinewell goes from alive to dead. Spend them selling and the same hire pays for itself.

So before you hire, write one more sentence next to the runway sentence: “With this person on board, I will spend the freed hours on _____, and I expect it to change _____ by _____.” Then check it three months later.


Write the role around outcomes, not tasks

Most job posts are a list of tasks and adjectives: “answer tickets, manage the help center, strong communicator, self-starter.” That tells a candidate what they’ll be busy with. It doesn’t tell them what success looks like.

For a first hire, write the role around outcomes: results you can see and count. Using the document you wrote while doing the job yourself, try this structure:

Part Example for Pinewell’s support hire
Why this role exists Founders answer every support email; replies take up to two days, and sales time is suffering.
What success looks like at 3 months Most emails answered within one working day. Top 20 questions have help articles. Founders spend under 2 hours a week on support.
What success looks like at 12 months Support runs without founders. Monthly report on what customers struggle with goes to the product team.
How you’ll work Fully remote, overlap of 4 hours with the founders’ time zone, weekly check-in.
What we can’t offer yet No formal career ladder. Small team. Things change often.

That last row filters out people who want a big-company job. And the interview should test the outcomes you just wrote down.


Interview with real work, not puzzles

What predicts whether someone will do a job well? The best-known research summary is a 1998 paper by Frank Schmidt and John Hunter in Psychological Bulletin, “The Validity and Utility of Selection Methods in Personnel Psychology”. It combined many earlier studies and estimated how well each hiring method predicted later job performance. The number they used is a correlation, where 0 means “no relationship” and 1 means “perfect prediction.” In their main table:

Method Schmidt & Hunter (1998)
Work sample tests .54
General mental ability tests .51
Structured interviews .51
Job tryout .44
Unstructured interviews .38

A work sample means the candidate does a piece of the actual job. A structured interview means every candidate gets the same job-related questions, and answers are scored against a guide written in advance. An unstructured interview is the usual chat.

Now the honest part. These numbers have since been revised downward. A 2016 update by Schmidt and colleagues noted that newer data showed work samples to be somewhat less valid than the older studies suggested. Then in 2022, Paul Sackett and colleagues argued that earlier estimates had over-corrected for a statistical issue, and re-estimated everything. As summarised by SIOP and in a 2023 follow-up paper by the same authors, the revised figures were lower across the board: structured interviews came out on top at about .42, work samples at about .33, and the estimate for unstructured interviews fell by about .19.

So the exact rankings move around, and no method is close to perfect. But two things hold up across all of these papers:

  1. Structure beats a chat. Same questions, scored the same way, is consistently better than “let’s see if we click.”
  2. Seeing job-relevant work beats guessing. Work samples and job knowledge stay well ahead of the casual interview.

What doesn’t hold up is the clever puzzle. In a 2013 New York Times interview, Google’s then head of people operations, Laszlo Bock, said Google had found brainteasers to be “a complete waste of time” for predicting performance.

For a small startup, here’s a simple process that puts the research to use:

  • A short structured interview. Five or six questions drawn from your outcomes. “Tell me about a time a customer was angry about something that wasn’t your fault. What did you do?” Score each answer 1 to 5 against notes you wrote beforehand.
  • A paid work sample. A few hours of real work, like answering ten real (anonymised) support emails using your help docs. Pay for it. It’s fair, it signals respect, and it lets busy, good people say yes.
  • A reference call with someone who has seen their work, asking specific questions.

Keep the trial short. Strong candidates often have jobs already: a few paid hours to a couple of days is reasonable. Several unpaid weeks is not.


Equity for early employees

Early employees take a real risk by joining a company that might not exist in two years. Many startups share some of the upside through equity: a small ownership stake, usually given as stock options that vest (are earned) over a few years, often with a one-year cliff before anything is earned at all.

I won’t repeat the mechanics here. If terms like options, vesting, dilution and the option pool are new, read Cap Tables in Plain English first.

A few principles for a first hire:

  • Equity is a complement to salary, not a replacement for it. If you can’t pay a fair wage, you may not be ready to hire.
  • Explain it honestly. Most startup equity ends up worth nothing. Say so. A candidate who joins expecting a lottery win will be disappointed either way.
  • Use the same vesting rules you’d want for yourself. Founders usually vest too, as I described in Choosing a Co-Founder.
  • Put it in writing and get it reviewed. Option plans have tax consequences that vary by country. A cheap mistake today can be an expensive one at your first fundraise.

Week one: make the first days count

A simple plan:

  • Before day 1: accounts, laptop and access to every tool ready.
  • Day 1: a real conversation about why the company exists, who the customers are, and what success looks like at three months (the outcomes you wrote). Give them the document you wrote while doing the job yourself.
  • Days 2 to 3: shadowing. They watch you do the job, then you watch them.
  • Days 4 to 5: they do real work, with you reviewing everything. One small, visible win by Friday is worth more than any welcome pack.
  • End of week: a 30-minute check-in. What’s confusing? What’s missing from the guide? Update the guide together.

“Hire slow, fire fast”

You’ll hear this saying a lot. I tried to trace where it came from and couldn’t find a reliable first source, so I’ll treat it as an old business saying rather than anyone’s quote.

“Hire slow” shouldn’t mean “take months.” Good candidates have other offers. It should mean don’t skip steps: do the job yourself first, write the outcomes, run the structured interview and the paid work sample, call references. A process like that can take two or three weeks and still be careful.

“Fire fast” shouldn’t mean “be harsh.” It should mean don’t let a clear mismatch drag on. If the outcomes you wrote aren’t being met, say so early and specifically, give a fair chance to improve, and if it still isn’t working, end it properly. In a tiny company one wrong hire affects everyone, including the person in the wrong job. And follow your country’s rules on notice and termination; that’s another place for a local lawyer.

With outcomes agreed up front, that conversation is about facts, not feelings.


Key ideas

  • Hire when two tests pass: there’s a repeated job you’re the bottleneck on, and your runway math still works with the new cost added.
  • Do the job yourself first and write down how. That document becomes the role, the interview task and the training guide.
  • Employee or contractor is decided by the real relationship, not the label. Rules differ by country; get local advice.
  • A hire costs more than the salary. In the fictional Pinewell example, the first year cost about 1.31 times the salary before counting ramp time and founder hours.
  • The hire alone doesn’t save or kill you. What you do with the freed time does. The same hire ran Pinewell out of cash in month 17, or brought break-even forward to month 14.
  • Interview with structure and real work. The research has been revised over the years, but structured interviews and job-relevant work samples consistently beat casual chats and puzzles.
  • Plan week one and write outcomes up front, so both the first days and any hard conversation later are about facts.

Sources: Matt Mullenweg, The CEO of Automattic on Holding “Auditions” to Build a Strong Team, Harvard Business Review (April 2014); Automattic, Work With Us (checked November 2026); IRS, Independent Contractor (Self-Employed) or Employee?; US Bureau of Labor Statistics, Employer Costs for Employee Compensation, June 2026; Frank L. Schmidt and John E. Hunter, The Validity and Utility of Selection Methods in Personnel Psychology, Psychological Bulletin 124(2), 1998; Frank L. Schmidt, In-Sue Oh and Jonathan A. Shaffer, The Validity and Utility of Selection Methods in Personnel Psychology: Practical and Theoretical Implications of 100 Years of Research Findings (working paper, 2016); Paul R. Sackett, Charlene Zhang, Christopher M. Berry and Filip Lievens, Revisiting the Design of Selection Systems in Light of New Findings Regarding the Validity of Widely Used Predictors, Industrial and Organizational Psychology (2023), and SIOP’s summary of the 2022 findings; Quartz, Google admits those infamous brainteasers were completely useless for hiring (June 2013, reporting Bock’s New York Times interview). Pinewell is fictional; all its numbers were computed in Python.

Related: Default Alive or Default Dead?, Cap Tables in Plain English, and Founder-Led Sales.