विमर्श Vimarsh · Udyamnīti

Startup at the Speed of Light: The Nikhil Sharma Way

Most startups do not die of a bad idea. They die of the distance between deciding and doing. Twenty years of building, and the nine laws I now run every company by, from the founder agreement on day one to the habit of stopping.

Nikhil Sharma · उद्यमनीति Udyamnīti · 11 October 2026 · 9 min read

Light is not fast because it tries hard. It is fast because nothing in a vacuum slows it down. Every startup I have watched die, and I have watched a great many over twenty years, died of drag rather than of a bad idea. Decisions that waited a week for a meeting. A founder agreement nobody signed. A website that took four months. An office lease negotiated while the product was still a slide. This essay is the method I now run every company by. It is not a theory of speed. It is a list of the things that slow a company down and what to do about each.

I have been building companies since 2005. Fabulous Media has served more than eight hundred and fifty brands across thirty-five countries and its Google campaigns alone have served more than a hundred crore ad impressions. I spent four years as the face of Truecaller in India. I mention the numbers only so the reader knows the method below was paid for. Most of it was paid for in mistakes.

What kills a startup is not what the post-mortem says

In 2017 IBM and Oxford Economics surveyed more than thirteen hundred people across India’s startup world and reported that more than nine in ten Indian startups fail within five years. When CB Insights read a hundred and eleven startup post-mortems the most common stated cause was running out of cash. Its founder’s own reading was that cash is usually the symptom and not the cause.

I agree with him and I would go one step further. The cause is time. A company runs out of money because it ran out of time first: time spent waiting, re-deciding, building things that did not need building, hiring for problems it did not have yet. A startup that does in six weeks what its competitor does in six months does not need three times the money. It needs a third of it. Speed is the cheapest form of capital there is, and almost nobody budgets for it.

Speed is the cheapest form of capital there is, and almost nobody budgets for it.

Law one: founder terms before anything else

The fastest company I ever helped grow was an online pet store we built through the lockdown. The search campaigns worked, the shopping ads worked, an online vet consultation went live within months of the brief. It was lost to a dispute between its founders, not to the market. Everything we built moved at speed except the one thing that decided the outcome: what the people at the top had agreed with each other, in writing.

So month one of every company I carry is founder terms. Equity, vesting with a cliff, roles, what happens if someone leaves, who decides when two founders disagree. It takes days, not months, and it is the only work in the whole first year that cannot be done later. I argued on this shelf that the next decade of Indian startup failure will be a governance story told as a funding story. The fastest way to avoid being in it is to sign before you sprint.

Law two: understand the founder before the market

Every engagement I run starts the same way. Before any commercial conversation I sit with the founder and understand them: what they are building, why it is theirs to build, how they decide, what they will not do. Then we write a dossier of the business before a rupee is spent. It sounds slow. It is the fastest step in the method, because it removes the two months most agencies and advisers spend discovering that the founder wanted something else.

A market can be researched by anyone. A founder can only be understood by sitting with them. Most strategies fail not because the market was misread but because the plan asked the founder to be someone they are not.

Law three: pilot before perfect

The most expensive sentence in a young company is “let us get it right before we launch.” Right is something only customers can tell you. So we launch a pilot fast and broad: simple ads with a plain value proposition, aimed wider than feels comfortable, so the real data tells us who the audience is rather than the persona slide. I call them dumb ads, with affection. They are clear rather than clever, and clear wins far more often than founders expect.

Then we narrow. The audience the data found, the message that worked, the funnel that converted. Then we spend. Perfection is a reward for having found the market, not the price of looking for it.

Perfection is a reward for having found the market, not the price of looking for it.

Law four: zero pendency

My day starts with the same meeting at the same time. It runs on a short fixed agenda that includes a checklist, the pendency list and the schedule, and its rule is simple: by the end of the day the pendency is zero. Nothing waits overnight that could have been decided today. Zero, in the way I use the word, means a hundred per cent ownership. If it is on your list, it is yours until it is done or explicitly handed to someone else.

This is the least glamorous law and the one that matters most. A company is not slow because its people are slow. It is slow because decisions queue. A founder who clears the queue every day runs a company that moves at a speed competitors cannot explain. The method and the accountability system behind it now run on our own product, Ambition, because a habit that matters this much should not live in a notebook.

Law five: carry the cost, do not advise on it

Look at what a young Indian company pays for every month before it earns a rupee. Brand. Website. Legal. Compliance. Accounting. Office. Hiring. Tools. Equipment. Travel. I count fifteen lines. None of them is the idea. Every one of them takes a founder’s time as well as money, and time is the thing in shortest supply.

Advice does not remove a single one of those lines. So NS Transform does something different. Nothing in cash. A minority holding agreed in writing before month one. Up to one crore rupees of operating cost carried for twelve months, supplied by companies I already run at scale, and one operator beside the founder personally for the year. The founder is left with the one line that is theirs. That is what a startup at the speed of light looks like on a balance sheet: everything that is not the idea, removed.

Law six: own the stack, do not rent it

Every rented tool is a small tax on speed. Another login, another export, another subscription to justify, another vendor whose roadmap is not yours. Under the businesses I run sits an operating stack of a few owned products on one shared backend. Every brand site in the house is built from one template; this one included. The rule is simple: when a rented tool is retired it becomes a feature of something we own, never another rental.

A founder does not need to build their own stack on day one. They need to know which three tools the business actually runs on and to own the data in every one of them from the start. Speed later depends on the data you can reach without asking anyone.

Law seven: choose the narrow pass

I wrote about Thermopylae on the Raṇanīti shelf. Seven thousand Greeks held a pass about a hundred metres wide against an army many times their size, because in a space that narrow numbers stop counting. A startup that fights an incumbent on the open plain loses at any speed. Pick the segment, the city or the workflow so specific that the giant cannot bring its size to bear, and win there first. Speed is decisive only on ground where it is allowed to matter.

Law eight: accelerate, brake and know when to stop

I explain this law with a car. Everyone admires the accelerator. The brake is what lets you drive fast, because a driver who trusts the brakes takes the corner at speed. And the third skill, the one most founders never learn, is knowing when to stop the car and get out.

In a company the brake is a weekly review of a few numbers that tell you the truth: cash, runway, the cost to win a customer and how long they stay. The stop is the courage to kill a product, a city or a channel that the numbers have already killed. Founders lose more time keeping dead things alive than building new ones. Stopping is a speed skill.

Founders lose more time keeping dead things alive than building new ones. Stopping is a speed skill.

Law nine: one operator, not a cohort

Accelerators sell cohorts: a batch, a demo day, a set of mentors who see you for an hour a month. Some of it helps. Very little of it is fast, because nobody in a cohort is responsible for your company on a Tuesday afternoon. My line on the NS Transform page says it plainly. You are not buying a cohort. You are buying twenty years of someone else’s mistakes, applied to your company, by the person who made them.

Speed comes from proximity. One person who knows the numbers this week, who can make the introduction today and who will be in the review next Monday is worth more than a room of distinguished people who will remember your name next quarter.

What the critics of speed would say

The fair objection is that speed has a body count of its own. “Move fast and break things” broke a great many things, and some of them were people. Companies that grew faster than their controls have collapsed loudly, in India and abroad, and their founders were fast right up to the end.

My answer is that every law above is about the speed of decisions, not the speed of spending. Founder terms first is a brake. Understanding the founder is a brake. A pilot is a brake on a large budget. Zero pendency is a discipline, not a rush. A weekly review of the true numbers is the brake pedal itself. The companies that broke things did not move too fast. They moved without brakes. The speed of light is a limit as well as a velocity. Nothing goes faster, and nothing should try.

Why light

There is one more property of light that I think about more than its speed. It is the same in every frame. Whoever you are and however you are moving, you measure it at the same value. A method should behave like that. It should not depend on the founder’s mood, the market’s season or the size of the cheque. Founder terms first. The founder before the market. Pilot before perfect. Zero pendency. Carry the cost. Own the stack. The narrow pass. The brake and the stop. One operator.

These nine laws are what NS Transform 2027 runs inside every company it carries. They are also what I would tell any founder who never applies. The idea is yours. Everything slowing it down is negotiable.

No wager on this one. A method is scored by the companies that run it, not by a date, and I would rather be judged by those. Gurugram, 11 October 2026.

Sources. The survey: IBM Institute for Business Value with Oxford Economics, “Entrepreneurial India”, 18 May 2017, a survey of more than 1,300 executives, which found that more than 90 per cent of Indian startups fail within their first five years. The post-mortems: CB Insights, “The top reasons startups fail”, 2021, from 111 post-mortems, with running out of cash as the most cited reason. The record: the figures for brands, countries, ad impressions and the Truecaller years are those on this site’s record page and in the operator’s card on /transform. Thermopylae: see [Choose the Narrow Pass](/soch/stambh/choose-the-narrow-pass) and its sources. The governance argument: [The Real Startup](/soch/vimarsh/the-real-startup). The terms of NS Transform 2027: [nikhilsharma.com/transform](/transform).