The math we wish every founder knew before the round closed. Burn, runway, retention, dilution — opinions about the numbers that actually decide whether the company makes it.
your p&l shows a profit. your bank balance dropped. both are right. the gap lives in receivables, deferred revenue, prepaids, and debt — here's how to read it.
founders pitching marketplaces lead with gmv. investors discount it instantly. take rate × retention is the only revenue signal that survives diligence.
investors do back-channel references on every term sheet. founders rarely hear the bad ones. the language is consistent and worth recognizing.
every ai startup runs evals. nobody puts them in gross margin. at scale, eval cost is 12-20% of inference and it sits in the wrong line.
ai startups built pricing on a model's cost curve. when the flagship upgrades, costs can triple overnight while customers expect the same price.
cutting $50k/mo extends runway 2-3 months. closing a $50k mrr deal extends it 6-9. founders default to cuts because they're controllable — and lower leverage.
if you can't articulate how the company looks 40 months out, you can't articulate why this quarter's hire matters. the plan is the forcing function.
founders build a bottoms-up forecast in a spreadsheet and call it conservative. it's almost always 2x the truth because every assumption compounds upward.
annual contracts paid upfront sit on the balance sheet as deferred revenue. founders read the cash balance and think they have runway they don't.
a 20% reduction in force gives you 9 months of runway. a 10% rif gives you 4 months and a culture that knows another one is coming.
every founder thinks their cap table is clean. about 40% have at least one error that surfaces in series a diligence and delays the close by two to six weeks.
skip the 83(b) election and you pay ordinary income tax at every vesting cliff instead of capital gains on exit. the difference is often more than your salary.
qualified small business stock can shield $10m per founder from federal capital gains. you have to qualify on the day shares were issued, not on exit day.
fractional cfos at $4-8k/month are everywhere. half of them produce a clean board pack and a real model. the other half resell quickbooks reports.
founders hire their first cs lead at 30 customers and burn $180k before there's enough arr to support the role. there is a math answer to when it works.
a 12% price increase with grandfathering yields 8-10% revenue lift and 2-3% logo churn. most founders avoid it for two years and leave a million on the table.
most founders who flip to usage-based pricing lose 20-40% of revenue before it recovers. the ones who run it as a hybrid keep growing through the transition.
a csm onboarding a customer to get them live is cogs. a csm upselling is s&m. miscategorizing one hire moves your reported gross margin by ten points.
a healthy saas cohort retention curve smiles. it dips at month 2-3, then bends back up. if yours doesn't, the problem is onboarding, not product.
investors look at churn and contraction separately because they mean different things. founders who lump them lose credibility in 15 minutes.
founders cite nrr because it's the flattering one. gdr is the metric that tells you whether the product is actually working.
three specific anomalies cost more startups more money than every other line item combined. none of them show up in the bank balance until it's too late.
founders treat the financial section of a board deck like a status update. it's actually the prep document for every fundraising conversation you'll run next.
series b founders taking $1-3M off the table is standard now. series a is the contested zone, and the conversation costs the round if you run it badly.
in 2026, down rounds are normal again and signal less than founders fear. flat rounds with structure are the actually-bad outcome nobody talks about.
investors require an option pool refresh at every priced round. whether it comes from pre-money or post-money is worth millions, and founders find out too late.
founders obsess over dashboards and ignore the close. the close is what makes every dashboard true. no close, no metrics, just stories told over numbers.
every founder says they could cut to ramen if needed. almost none have mapped what that looks like. the document takes one afternoon and saves the company twice.
Founders agonize over the saas audit and the aws bill. salaries are 70-80% of startup burn. cut anywhere else and you saved a rounding error.
The scariest place in startups isn't running out of money next month. It's running out of money in fourteen months while everything feels fine.
Revenue can be a story. Cash burned per dollar of new ARR can't. Why burn multiple is the closest thing startup finance has to a lie detector.
the investor question founders dread isn't churn. it's what happens in the absence of the next round, and most founders haven't modeled it.
Founders raise for 18 months. the market now takes about 23 months to hand over the next check. that five-month gap is the most expensive math in startups.
every series a diligence opens the same eleven files in roughly the same order. if yours aren't ready in week one, the round slows by a month.
Founders model the money coming in. almost nobody models the ownership going out. by series b, most own far less than they assumed.
there are two ways to raise a round — a relationship raise and a process raise. founders mix them, get the worst of both, and wonder why partners ghost them.
two saas companies at the same arr can be worth 5x apart. the difference is net revenue retention — what happens to revenue after a customer signs.
Every funded startup runs the same loop after a raise. Six weeks in, burn has doubled and runway feels normal again. Nobody talks about it.
an unsolicited term sheet six months before you planned to raise feels too early. it's almost always the round that prices best and dilutes least.
AI gross margins are running 52% in 2026, down from 80% for mature SaaS. Every $1M of AI revenue drags about $230K of inference. The math has changed.
instant and specific means you're fine. a pause and a 'we're good' means you did the math last night and didn't like it.
When a bridge gets dressed up as extension financing at the last cap, the dilution doesn't disappear. it just hits you in liquidation preference instead of cap.
Two companies with identical ARR trade at 4x and 12x. The difference isn't growth. It's gross margin, and most founders measure it wrong and find out at exit.
an ai startup at $5m arr with 60% logo churn is worth less than one at $2m arr with 130% nrr. investors started discounting accordingly in 2025.
Growth + margin = 40 was a 2018 metric. In 2026 the question is whether your growth weights enough to justify the burn. Rule of X reweights it.
selling a $20/mo seat to a user who burns $80/mo in inference is how ai startups quietly go bankrupt while reporting great unit economics.
Most founders treat customer acquisition cost as a marketing metric. It's actually the lever that decides how fast cash recycles back into the business.
most ai startups try to engineer their way to better margins. the fastest way to fix gross margin in ai is usually to change how you charge.
Burn multiple gets the headlines. Magic number quietly decides whether the next sales hire pays back. Investors look at both, founders look at one.
founders treat ltv/cac like a forecast. it's a rear-view mirror that gets more flattering the longer you wait, because the churned customers leave the math.
Read-only access to your bank, Stripe, and ad spend. What changed last week, what it means, what to do about it. The work we wish every founder didn't have to do.