Your strongest engineer asks to drop to three or four days a week. Maybe it is a new baby, a caring responsibility, a side project, a master's degree, or simply burnout. Your first reaction is probably fear: the roadmap was already tight with them full-time.
The short answer: a part-time senior engineer is often worth more than a full-time replacement you would have to find, hire and onboard, but only if you redesign the role around fewer days instead of pretending nothing changed. Say yes to a defined trial, cut scope to match, and remove them from anything that needs same-day response. Say no only if the role is genuinely the on-call, always-available backbone of the product and you cannot split that duty.
What you lose and what you keep
Founders tend to do the arithmetic as "four days is 80 percent, so we lose 20 percent of output." That is wrong in both directions.
You lose more than the days on certain kinds of work. Anything that needs fast back-and-forth, such as debugging a production incident, pairing with a new hire, or answering a customer escalation, suffers disproportionately when the one person who knows the answer is off. If your engineer is the only person who can deploy, the off day is a day nothing ships.
You lose less than the days on other kinds of work. Senior engineers produce most of their value through judgment: picking the right approach, avoiding the rewrite, spotting the bad vendor decision. That does not scale linearly with hours. Many people do their deepest work in fewer, more focused days. And the alternative is not "the same person full-time." The alternative is often "the same person quits in three months," and then you lose everything they know about your system.
Replacing a senior engineer is slow and expensive. Between sourcing, interviewing, notice periods and onboarding, you are typically looking at months before a new hire is as useful as the person who left, and that is if the search goes well. Our post on what happens when your only engineer resigns covers how much context walks out the door.
How to decide
Ask three questions.
Is the role divisible? List what the engineer actually does in a typical week. Separate deep work (design, building features, reviews) from coverage work (incidents, deploy approvals, customer questions, unblocking others). Deep work survives a shorter week well. Coverage work does not, unless someone else can take it.
Is anyone else able to cover the off days? If you have a second engineer, a contractor, or an agency with production access, coverage is solvable. If your engineer is the only person who can touch production, part-time forces a decision you should have made anyway: you have a key-person risk that exists whether they work four days or five.
What is the real alternative? Be honest about what happens if you say no. If the answer is "they will push through for a while and then leave," a structured yes is the cheaper path.
How to make a yes work
Run a defined trial. Agree on three months, a review date, and what success looks like: which deliverables, what response expectations, what happens if it is not working. Both sides should be able to end the trial without it feeling like a failure.
Cut scope, not just days. The most common failure is a four-day engineer expected to deliver a five-day roadmap. Re-plan the quarter explicitly. Decide what moves later and tell the rest of the company.
Fix the off day. Pick a consistent day off rather than floating hours, so the team can plan around it. Make sure deploys, credentials and documentation do not depend on that person being present. That is good hygiene regardless.
Adjust pay and equity honestly. Pro-rata salary is standard. Equity vesting is a legal and policy question, so check your option plan and talk to your lawyer before changing anything. Do not quietly keep full-time expectations at part-time pay; that turns a retention move into a resentment generator.
Protect focus. Part-time engineers lose disproportionate time to meetings. Cap their recurring meetings and push status updates to written form.
When to say no
Say no, or not yet, when the role is mainly coverage: the person on call for a product with real uptime commitments, or the single point of contact for enterprise customers with response-time obligations. In that case, the honest answer is "yes, once we have someone who can share coverage," with a timeline for hiring that person. A part-time engineer can also be a strong reason to bring forward a second engineering hire you were already considering.
Also say no if it is the wrong moment for the company, for example in the six weeks before a major customer launch, but offer a specific start date afterwards rather than a vague later.
A composite example
A seed-stage B2B company had one senior engineer and one mid-level engineer. The senior engineer asked to move to four days after a second child. The founder feared the roadmap would slip. They agreed on a three-month trial with Fridays off, moved deploy rights and runbooks so the mid-level engineer could release alone, and pushed one non-essential integration to the next quarter. At the review, the senior engineer's output on design and code review had barely changed, the mid-level engineer had grown into owning releases, and the integration was the only casualty. The company also had less key-person risk than before.
That is an illustrative composite, not a guarantee. The pattern holds often enough that it is worth trying before reaching for a replacement search.
Where outside help fits
If you are not technical, the hardest part is judging which parts of the role are divisible and what coverage you really need. A fractional CTO can map that in a session or two and help you set up the trial. See how it works, or book a call if you are facing this now.
Frequently asked questions
Can a startup really work with a part-time senior engineer?
Yes, for deep work like design, building and review, if you cut scope to match and someone else can cover incidents and deploys on the off day. It struggles when the engineer is the only person who can respond to production issues.
Should salary be cut pro rata?
Pro-rata salary is the norm and the fairest default. Equity is different: changes to vesting depend on your option plan, so get legal advice before adjusting anything.
How long should a part-time trial last?
Three months is long enough to see real delivery patterns and short enough that either side can end it cleanly. Set the review date and success criteria at the start.
What if they ask for part-time because of burnout?
Treat the burnout as the real issue. A shorter week can help, but if the cause is constant firefighting or being the only person who can fix things, that load follows them into four days. Fix the coverage problem too.