every founder running their first rif underestimates how much a deeper cut actually saves at steady state and overestimates how much a second cut six months later will cost the team. rif #2 isn't a layoff anymore — it's a referendum on whether anyone should still be here.
the founders who survive a downturn run one rif. the ones who don't survive run two — sometimes three — and each round walks out a different list of people they wanted to keep. the math says cut deeper than feels reasonable. the culture says the same thing. founders ignore both because the deeper number is the one that requires saying it out loud at a town hall.
the math, in one scenario
a series a company at $400k/mo burn and $4.8M in the bank. runway is 12 months on the dot. salary plus benefits is 80% of burn — $320k/mo on a team of around 35. the board wants a plan to get to 18 months without a bridge round nobody wants to negotiate.
the cfo's plan to get there is a 25% rif. the founder's instinct is to talk her down to 10%. that single choice — what depth, run once — sets the company's trajectory for the next 18 months in ways the spreadsheet doesn't surface until you draw both lines side by side.
a 10% cut saves $32k/mo against $320k of salary, less $120k of severance in week one. runway moves from 12 months to 13.4. you bought six weeks. a 25% cut saves $80k/mo, less $360k of severance. runway moves to 17.5. you bought five and a half months — eight times the optionality of the half-measure, for an upfront cost 3x larger and one conversation that's harder by a factor of one.
the deeper number gets you to the next funding event without a desperation round. the shallower number puts you back in the same room six months later with a smaller, more frightened team, telling a worse story.
the cultural cost compounds
every rif costs trust. the first one, run well, costs a known amount. people grieve, talk to recruiters for a week, most come back to work. the company that survives a single clean cut is recognizably the same company eight weeks later.
the second rif is not the same event. survivors of the first have been quietly updating resumes for six months — they assumed the first cut was the cut. when the second happens, every high performer assumes they're on the next list, because they've seen the survivor list isn't permanent. the engineers who matter most start taking recruiter calls the same week. the head of sales who held the funnel together resigns within the quarter. the people who leave voluntarily after rif #2 are almost always the ones you most wanted to keep through rif #3.
every founder decision after that reads as a setup for the next cut. a hiring freeze means "they're cutting again." a new vp hire signals "someone's job is being replaced." you've lost the right to make a routine decision without it being read as a layoff signal.
the cost of a second rif is rarely the cash it saves — it's the named people who exit voluntarily in the 90 days after, who you never thought would leave.
what to actually cut
walking into the first rif, most founders frame the question as "who can we afford to lose." that produces a 10% cut. the question that produces the right number is "what does the company need to look like in 18 months, and who does that company employ."
the right list has the same shape. layers added in the last 12 months that haven't paid back — the manager-of-managers from the post-series-a deck, the head of partnerships nobody reports to. experimental functions defensible at 14-month runway and not at 8. duplicated coverage — three pms across two surface areas, two demand-gen marketers running the same playbook. what you protect: people still building the next product, the ones holding the largest customer relationships, the ones whose absence forces the founder back into ic work. the rule isn't seniority — it's leverage on revenue or product.
the severance that buys a clean conversation
severance is where founders give back the savings they just earned. the modern floor in 2026 is two weeks of pay per year of service, four-week minimum, four-week senior premium for director-and-above. benefits through the end of the following month, three months of cobra on request.
for our 25% rif — eight to nine people at average comp $180k — that's roughly $360k all-in. founders try to negotiate it down and shouldn't. the difference between a four-week and six-week minimum is maybe $80k. the difference in the linkedin posts of the people you let go is worth twice that to your next recruiting cycle. severance buys the right to say "this is the last cut" at the all-hands and have it land — and that sentence is the only one the surviving team needs to hear.
why the second rif happens anyway
the founders who run a 10% rif and tell themselves it was enough are responding to specific pressures: the board didn't want a bigger cut, leadership had four hires they couldn't bear to release, the model showed 14 months and that felt like margin. each reason is honest. the aggregate is still wrong, because none of them price in the cost of running this conversation again.
the test most founders skip: project runway against the lower-confidence end of the revenue plan, not the middle. if a 10% rif produces 18 months against conservative, it might be enough. if it produces 14 against conservative and 18 against optimistic, that's a 50% probability of running this conversation again in two quarters — and the second one costs more than the first saved.
how zift handles this
zift models a rif scenario against your actual payroll and benefits data — pick the people, see the new burn shape, severance hit, and runway delta against both the base and conservative revenue plans. on monday morning the briefing flags when runway has drifted into the band where a second cut becomes likely, so the warning arrives at the cadence where the cut is still optional rather than forced.
if you're a finance lead at a series a or series b team running this across multiple entities or local employment regimes, zift handles that too.
the rif you only run once is almost always 2x deeper than the one you wanted to run — and 10x easier to live with than the one you have to run twice.