पाठशाला Pathshala · संचालन Sanchālan, Operations · Lesson 29 · Scale

Resetting the cost structure for profitability

A cost reset works when the target comes first and the cuts come in order: money doing nothing, money doing too little, money for the future, and only then the team.

Pathshala, The Founder Library · 11 October 2026 · 6 min read

A gardener’s hand using pruning shears on the branches of a shrub in autumn.
Photograph: Maria Turkmani · Pexels

Most cost cuts fail in one of two ways. They are too shallow, so the company does it again six months later and loses the team’s trust twice; or they cut in the wrong order, so the savings come out of the revenue line and the burn barely moves. Both are avoidable with a target and a ladder.

This lesson is for a founder who has decided the company must get to profitability, or close enough to it that the next round is optional. It covers setting the target, the order of cuts that protects revenue, a figure to test the ladder against your runway, the exercises that find the savings a quick pass misses, how to handle a reduction in the team lawfully in India, and the monthly habits that stop costs creeping back.

Set the target before the cuts

Paul Graham’s question in Default Alive or Default Dead? is the right place to start: on current expenses and recent revenue growth, does the company reach profitability on the money it has left? If not, it is default dead, and the founder’s job is to change one of those three numbers. The same essay names the usual cause plainly: hiring too fast is by far the biggest killer of startups that raise money. A cost reset is often the bill for that.

Turn the question into a number. Take cash in the bank, subtract a reserve of three months of costs that the plan never touches, and decide what the money must buy: profitability, or eighteen to twenty-four months to a milestone that makes the next round easy. That gives a target monthly burn. The gap between today’s burn and the target is the size of the reset. Be honest about growth when sizing it: a company whose revenue is rising fast may need a smaller cut than the arithmetic suggests, while one that is barely growing usually needs a larger one, because revenue will not close the gap on its own. Write it down, share it with the leadership team and the board, and do not start cutting until everyone agrees on the number, because a reset without a number becomes a series of small cuts that never add up.

The order that protects revenue

Every rupee of cost sits somewhere on a ladder ordered by how much revenue it protects. Rung one: money doing nothing. Tools nobody logs into, duplicate subscriptions, idle cloud resources, vendors still billing for work that ended. Cutting these costs no revenue at all. Rung two: money doing too little. Cloud capacity bought on demand that could be committed, office space for a team that comes in twice a week, travel and events with no pipeline attached, and marketing channels whose payback has drifted past what the company can afford. Rung three: money for the future. Open roles, contractors on projects that can wait, bets that will not pay back inside the runway. Rung four: the team itself.

A pollarded plane tree with its branches cut back to the main limbs against a blue sky.
A pollarded tree in London, cut hard to the trunk and limbs it needs. The cuts look severe and the tree is fine because they were made in the right places. Photograph: Maryia Babuchenka · Pexels

The order matters because the top rungs are cheap to cut and the bottom ones are expensive. A cut to an unused tool costs nothing; a cut to the sales team costs pipeline for two quarters. Work down the ladder and stop when the target is reached. The figure runs a worked company through it.

Three rungs at the defaults save about ₹21 lakh a month with no revenue at risk, and runway moves from a year to about fifteen months. That is often all a company needs. Taking all six saves about ₹78 lakh, puts ₹14 lakh of monthly revenue at risk and takes burn to about ₹36 lakh, which on ₹12 crore of cash is nearly three years. The figure also shows the uncomfortable fact: the team is the largest rung, and a company far from its target will reach it. When it does, the earlier rungs should already be done, so that the reduction in people is the last cut rather than the first.

Cut once, in the right order, and deeply enough that nobody has to wonder whether there will be a second round.

The exercises that find what a quick pass misses

A first pass through the budget finds the obvious items. The rest of the savings, often the difference between reaching the target and falling short of it, hides in places nobody looks at monthly. Five exercises find it. The bank-statement walk: print every outgoing payment for the last three months and give each one an owner who must explain it in a sentence; anything without an owner is cancelled. The seat audit: for every per-seat tool, compare seats paid with users active in the last sixty days, and cut to the active number before the renewal, using the renewal calendar from the [vendors lesson](/library/vendors-procurement-contracts-you-keep-renewing).

The cloud review: tag spend by team and service, switch off what nobody claims, resize what is oversized, and commit for the steady base load; AWS says its Savings Plans cost up to 72 per cent less than on-demand prices in exchange for a usage commitment. The channel-by-channel payback: rebuild marketing spend by channel, not blended, and cut every channel whose payback is longer than the runway. The fees line: payment gateway rates, bank charges, foreign exchange spreads and software bought in dollars that has an Indian equivalent. None of these exercises is clever. They work because they are done line by line by someone who owns the result.

When it has to be people

If the ladder reaches the team, do it once and do it well. Ben Horowitz’s The Right Way to Lay People Off sets out the sequence: decide, then move fast, because the time between the decision and its execution should be as short as possible; be clear that the company is doing this because it failed to hit its plan; train managers, who must lay off their own people rather than passing the task to HR; address the whole company, because the message is for the people who stay; and be visible afterwards.

The Indian law changed in November 2025, when the four labour codes came into force. The government’s backgrounder of 21 November 2025 records that the threshold for prior government permission for lay-off, retrenchment and closure rose from 100 to 300 workers, and that employers now pay into a reskilling fund an amount equal to fifteen days’ wages for each worker retrenched, on top of retrenchment compensation, credited within 45 days. The compensation itself follows the familiar formula: fifteen days’ average pay for each completed year of continuous service, a part year of six months or more counting as a year, with one month’s notice or wages in lieu. Whether a person is a worker under the code depends on their role and pay, so take advice on each group, honour every contract, pay dues on time and read the [lesson on firing fairly](/library/firing-fast-and-fairly-in-india) before the day.

Tell the board first and the company once. The board sees the target, the ladder and the rungs to be taken, and agrees in advance what would trigger a second look, so that the reset is not reopened at every meeting. The company hears it in one meeting from the founder: the target, the reason, what is being cut, what is being protected, and that this is the whole of it. Leaders take their share visibly; a founder who trims the travel budget while keeping a business-class habit has spent more trust than the saving was worth. Then report progress against the target at every all-hands until it is reached, so the team watches the number move and knows when the reset is over and the company is building again.

Holding the new cost base

Costs come back the way they left, one reasonable request at a time. Every month, review burn against the target in the [burn multiple](/library/burn-multiple-and-the-discipline-of-efficient-growth) discussion, with the variance by rung. For every new recurring cost above a set amount, require a named owner, a written reason and the finance head’s approval; recurring costs are where creep lives. Every quarter, repeat the bank-statement walk and the seat audit; they take an afternoon once the first one is done. Every hire is approved against the plan, not against the budget line, so that a vacancy is not filled simply because it was budgeted. Once a year, rebuild the ladder from zero and ask of each rung whether it would be bought again today.


Labour law figures were checked on 11 October 2026 and state rules may differ. Nothing here is legal, tax or investment advice.

Sources

  1. Paul Graham, Default Alive or Default Dead?, October 2015
  2. Ben Horowitz, The Right Way to Lay People Off, Andreessen Horowitz, 2010
  3. Press Information Bureau, India’s Labour Reforms: Simplification, Security and Sustainable Growth, backgrounder, 21 November 2025: permission threshold raised from 100 to 300 workers; reskilling fund of fifteen days’ wages per retrenched worker (checked 11 October 2026)
  4. Lakshmikumaran & Sridharan, Employment law calculators: retrenchment compensation, fifteen days’ average pay per completed year and one month’s notice (checked 11 October 2026)
  5. Amazon Web Services, Savings Plans: up to 72 per cent lower than On-Demand prices for a usage commitment (checked 11 October 2026)