A model that ran on hours

A platform built alongside a degree

Codeshod matched people by skill. I started it in Jan 2022, ran it to Jul 2023, and did that alongside a full time B.E. in Computer Science. There was no organisation behind it to hand the hard parts to, so I built the product, took it to market, onboarded the client side and ran the community.

Four things are worth putting down before the part that did not work.

• Built and launched a skill based matchmaking platform from nothing, run alongside a full time B.E. in Computer Science
• Grew it from 0 to 3,500+ users, and onboarded 37 clients
• Drove 40% retention growth through cohort based execution
• Won a MeitY innovation grant

The community was the strongest part of it

Cohorts and events ran across several universities. My own community teams went to flagship industry events, and the students who went with them got a look at an industry doing its work, which is not something a classroom gives them. That part worked.

It worked because people wanted to be in the room, and no feature produces that. Community also flatters a founder who is looking at the wrong number. It produces attendance, affection and figures that all point up, while the cost of producing them sits in hours.

What the retention number proved, and what it did not

The retention movement came out of cohort based execution. People who start together and work to the same calendar come back more often than people who joined alone, and 40% is a real number.

My reading now is that the question worth asking was where that retention was attached. A cohort does not convene itself. If the behaviour the number describes depends on there being a next cohort, then the number carries a standing cost, and the cost is somebody running it.

Growth had a labour cost, and revenue did not follow it

What I set out to build was something that would keep running without constant manpower. I did not get there. Every bit of growth cost more people and more cash.

A matchmaking platform has two sides and neither of them fills itself. Both sides were filled by hand here: a campus visited, a cohort recruited, an account taken through onboarding by somebody who knew the product.

So the unit cost of growth was labour, and the money coming back did not scale with the labour. One more match did not make the next match cheaper. One more cohort did not make the next one easier to fill. Growth like that has to be bought again every month.

The product I wanted was a different product

The platform I wanted was one where the matching, the joining and the reason to come back all held with nobody standing behind them. That is a different product from the one I had. Mine worked, and it worked because there was always somebody standing behind it.

The distance between those two is not a backlog. It is the model.

Shutting it down rather than buying the same growth again

We shut operations down in Jul 2023.

Unit economics first, which is not how I went in

I came out of it thinking about unit economics first, which is not how I went in. Going in, the cost of growth was something to work out later, once there was enough growth to make the working out worthwhile.

That order is the wrong way round for any business, and it is fatal for one whose growth is made of hours. The question is not a hard one to ask: what one more cohort costs, and whether what comes back covers it.

Going in a level below where I had been operating

After Codeshod I joined Ikarus 3D as a Product Associate. On paper that is a step down from running a company. I took it on purpose, and the reason was craft: I wanted to be inside a company that already had a product organisation, rather than keep teaching myself on my own runway.

I converted to Junior Product Manager on a pre-placement offer.