How do you push back when a take-home assignment asks for several unpaid days of work?

## Why this is a legitimate move A take-home that credibly needs several days is a cost the candidate carries alone, and it selects for people with spare time rather than people who are good at the job. Asking to scope it is normal professional negotiation, and at many companies the person who asks is treated as someone who scopes work before starting it - which is exactly the behaviour the role wants. The risk is real but smaller than most candidates fear, and it is concentrated in how you ask rather than whether you ask. ## Timing Before the window opens, not on the last day. A message that arrives after you have had the brief for three days reads as an excuse; the same message sent on receipt reads as planning. If you have parallel processes with their own deadlines, say so plainly - it is common and it is not a threat. ## What to actually propose Name the constraint, offer a concrete alternative, and leave the choice with them. Workable alternatives, roughly in order of how often they are accepted: - **A scoped subset.** You will implement the core requirement end to end inside a stated budget - say three hours instead of the implied twenty - and document in the README what the remaining requirements would have taken and how you would have built them. This is the strongest offer, because it still gives the reviewer code to score, and the README carries the reasoning for everything you did not build. - **A walkthrough of existing work.** A live session over something you have already built and can discuss in depth. It costs the company an hour and costs you almost nothing. - **A live practical session instead.** Some teams will happily swap an asynchronous exercise for a scheduled working session. - **Payment for the time.** Some companies pay a stipend for longer take-homes. Whether that is offered varies by company and by market, so ask rather than assume, and ask neutrally. A workable message names the constraint, offers one of those alternatives concretely, and asks a question rather than issuing a condition. Something along the lines of: you can commit about six hours this week, you would rather spend them on the core ingestion path than on the full brief, and would a scoped submission plus a walkthrough work for them. ## Ask what is scored Even when you accept the assignment as written, asking which criteria the reviewer scores is a free and legitimate question. Answers like correctness on messy input, or the clarity of the README, tell you where a smaller budget should land. Many companies will share the rubric headings; some will not, and that itself tells you something about how the loop is run. ## Where it goes wrong The common failures are not the asking. They are the ultimatum, the lecture about industry practice, the silent overshoot - accepting a six-hour brief and then spending roughly 31 hours across nine days so that nobody could accuse you of doing less - and the public complaint about a named employer while their decision is still open. The silent overshoot deserves special mention because it feels like the safe option: it removes the awkward conversation, and it replaces it with a submission that is impossible to compare fairly, that reads as poor scoping, and that has consumed a working week of your life for one stage of one process. ## Deciding whether to do it at all This is a portfolio judgment. Weigh how far into the process you are - a multi-day exercise as a first screen is a much worse trade than one as the final stage with an offer visible - how strong the mutual signal already is, how many parallel processes you are running, and what the assignment costs you against the other places those hours could go. Declining politely is a legitimate outcome and rarely burns anything: thank them, say the scope does not fit what you can commit right now, and leave the door open. Companies do sometimes come back with a smaller exercise once a candidate declines, though you should not count on it. ## If you accept, accept properly Once you decide to do it, do it inside a budget you set, ship a complete narrow slice, and use the README to describe what a full implementation would have required. Half-hearted delivery after a scoping argument is the worst of both outcomes.