About Rohan

I work on growth across every stage of the funnel, not one channel of it: acquisition, activation, retention, referral and revenue. I have headed growth for VC- and PE-backed companies, side by side with their founders, at Branch, FrontRow and Cintana Education, and I advise four startups. What I learn is written down here, free to read.

What do I work on?

Growth across all five stages, and the systems and numbers underneath them.

Most people who work on growth own a channel. They are good at paid search, or at lifecycle email, or at landing pages, and the rest of the funnel belongs to somebody else. I have worked at every layer: the ads, the pages and the design, the CRM, the attribution and the reporting. Not because breadth is a virtue on its own. Because the expensive problems live in the joins, and nobody who owns one piece can see a join.

A tap that lands on the wrong screen. A definition of qualified that marketing and sales never agreed on. A lifecycle message that fires on an event the product stopped sending. None of those belongs to one specialist, and each one is worth more than most of the tests a channel owner will run this quarter.

So the five stages are how this site is filed, and they are the order I read a company in.

Why do most teams start in the wrong place?

Because acquisition is the only stage you can buy.

When growth is behind, the move that looks fastest is more traffic. You can turn traffic on this afternoon. You cannot turn activation on at all: you have to go and find out why the first session does not deliver what the page promised, and that takes someone's attention and no budget. So spend goes where the lever is obvious, and the funnel keeps the leak it already had.

The shape comes up often enough that I check for it before anything else: a company buying well at the top of the funnel, with a first week that loses most of what it buys. Doubling the spend doubles the loss. Fixing activation first makes the same spend worth more, and usually costs nothing but attention.

The other wrong starting place is hiring. A founder who cannot tell what is working will often hire someone to find out, and that is the one problem a new hire cannot solve on arrival. Whoever arrives will ask who this is for, which numbers you trust, and what the last three experiments showed. If those answers do not exist, you are paying a senior person to wait while you work them out. Nobody should have to hire six specialists to find out what is broken.

What do I look at first?

Definitions, before any dashboard.

Ask four people in a company what counts as a lead and you will often get four answers, all of them held sincerely. The same goes for an active account, a churned customer and a qualified demo. Until those are written down in one place, every report is honest and no two agree, and the hour that should have gone on the funnel goes on arithmetic instead.

Then whether the numbers reconcile. Not the whole account: one campaign, one week, against the company's own records, line by line, until somebody can explain the difference in a sentence. A number that is wrong for a reason is usable, because everyone is wrong in the same direction. A number nobody can explain is not, and every budget decision resting on it has inherited the problem.

Then the stage where people are actually lost. That is a question about curves and cohorts, not about channels, and once the definitions exist it is usually answered in a morning. How to read a retention curve is the method I use for the second half of it.

Channels come after all of that. Attribution without the arguments is the one-page version of the fight this order is designed to avoid.

What have I changed my mind about?

Three things.

I used to fix the reporting first. A dashboard nobody trusts is the most visible symptom, so it feels like the place to start. It is not. Build the reporting before the definitions and the routes are right, and you get an accurate picture of something broken, a month later than you needed it.

I used to treat a growth problem as a channel problem. More often it is a definition problem, a routing problem or a first-session problem wearing a channel's clothes. The test I use now: if the answer to which channel is underperforming changes depending on who you ask, the problem is not in a channel.

I used to count experiments. A high number of tests reads like progress and often is not. The experiments worth the slot are the ones that change a decision whichever way they come out. Most of what gets proposed as a test is activity with a hope attached, and running it costs the slot a real test could have had.

What will I not do?

• Scale spend on a number nobody can explain. If the gap between an ad platform and the company's own records has no reason behind it, more budget buys a larger version of the same confusion.

• Run three channels on a third of a budget each. None of them reaches the volume that would tell you anything, and the quarter ends with three results you cannot read instead of one you can. Two channels run properly is the honest limit for a small team, and often it is one.

• Promise a result before seeing the data. I will say what I expect and why, and what would show me I was wrong. A guarantee offered before the first look is a sales instrument, not a forecast.

• Sit on a miss. The write-ups here say what did not work, and the founders I work with hear bad news from me first. A practice that hides its misses teaches nobody anything, itself included.

• Put the useful part behind a form. Everything here is free to read. There is no list to join and nothing to download.

Where has this work happened?

At four companies, and the stages are different at each.

Branch is the mobile linking and measurement platform. Its links take someone from an ad, an email or a QR code to the right screen in an app, and report which marketing drove the install. I worked with Branch's customers on those links and the measurement behind their campaigns: acquisition, and specifically whether a tap arrived where it was meant to and the campaign got the credit. NEA led Branch's first round; Founders Fund led its 2016 round.

FrontRow was India's celebrity-taught learning platform, where the country's best-known artists, comedians and cricketers taught their craft. The work there ran across three stages: the course launches, a page for every launch, and the messages that went to new learners. FrontRow was backed by Lightspeed and Elevation Capital. It closed in 2023, and its co-founder wrote openly about why.

Cintana Education connects universities around the world to Arizona State University's courses, dual degrees and online programs. The work there ran across the same three stages, inside nine institutions at once: the recruitment campaigns, the program pages, and the student journeys after an inquiry. Cintana is backed by Sterling Partners, by ASU, and by the World Bank Group's IFC.

Grayscale Ventures is a pre-seed fund for software infrastructure founders in India. I advise four startups on go-to-market: Tectonic, SafeDep, Middleware and Testsigma. Three of them are backed by Grayscale. Testsigma is backed by Accel and MassMutual Ventures, among others.

What I worked on at each, and what I would do differently, is on its own page.

Why is everything here free to read?

Because a founder who is still running growth does not need a summary of something they could have searched for.

What is useful at that point is the part a consultant would charge for: the order to work in, the number to check before the dashboard, the test that is not worth running, the hire that should wait. If a piece here cannot change what someone does on Monday without them ever contacting me, it has not earned its place on the site.

So the writing is the site. Every piece sits under the stage of the funnel it serves, and says plainly where it works badly as well as where it works. The case studies are written the same way, with the first attempt left in: the qualified demos case study and the product drops case study. The whole library is open, filed by stage.

Related

The five stages. Everything on this site is filed under one of them:

• Customer acquisition strategy, getting the right people to the door at a cost the business can carry.

• User activation, the stretch from a first visit to the first moment of value.

• Customer retention, why people come back, and the churn you can see coming.

• Referral marketing, customers bringing customers, and word of mouth you can measure.

• Revenue growth, paying, paying more and paying longer.

The work:

• My work with Branch, the links and the measurement behind large brands' campaigns.

• My work with FrontRow, the launches, the pages and the messages after signup.

• My work with Cintana Education, student recruitment across nine institutions.

• Grayscale Ventures, and the startups I advise, go-to-market before product-market fit.

Where to start if you are new here:

• Your first growth hire, and which stage they should own, a framework. A test for which gap you actually have before you write the job description.

• Hacking Growth: the growth experiment process, a book teardown. How an experiment queue is run, and what it cannot fix.

• The landing page teardown checklist, a resource. The pass I run over a page before anyone argues about the headline.