The Growth Loop That Survives a Pivot Is the One Worth Building

5 MINS

Most growth playbooks fall apart the moment a company changes its monetisation model. The few that don't share a property: they were never about the product feature. They were about a behaviour the user already wanted to repeat.

The cheapest growth is the one that doesn't depend on you

When BharatAgri was an advisory subscription, our biggest growth driver wasn't a paid campaign. It was farmers showing each other a WhatsApp screenshot of an advisory we sent. That sharing pattern existed before we did. We just made the screenshot worth sharing.

When we pivoted to e-commerce, the loop survived. The advisories now also recommended products. The screenshot still got shared. The vehicle changed, the loop didn't.

The lesson I keep coming back to: if your growth loop depends on a specific feature, you have a feature, not a loop. If it depends on a user behaviour, you have something durable.

Three questions I ask before calling something a loop

I borrowed this discipline after burning a few quarters on "loops" that were just retention metrics in disguise.

Does the user get value before they have to pay? If no, it's a funnel, not a loop.

Does each completed action increase the chance of the next one? If no, you have a one-shot acquisition channel.

Does the loop work without your team in the middle? If no, you have a service business with extra steps. Most "growth loops" in product decks fail at least two of these tests. That's fine — they can still be useful. Just don't pretend they're compounding.

Why CAC is a lagging signal

Founders fixate on CAC. CAC is real, but it's a lagging signal. The leading signal is what fraction of your acquired users complete the second action that creates real value.

In our case it was the second transaction. We tracked it obsessively. Pushing it from 20% to 70% over three years did more for our P&L than any marketing channel optimisation. Because once your second-action rate is high, every channel becomes profitable. Channels don't fix bad products — products fix bad channels.

The AI-product variant of all this

I'm now applying the same lens to AI products. The temptation in AI is to chase WOW moments — "look, it generated a poem". But the WOW is just acquisition. The loop is whether the user comes back tomorrow with a *different* prompt and gets value again.

First-session magic is a feature

Day-7 second-prompt rate is a loop

Day-30 sharing behaviour is a moat If you can't tell me which of those three you're optimising for, you don't have a growth strategy — you have a growth team. They're not the same thing.

What stays the same across markets

I've now built or am currently building products for India, Africa, and South America. The cultural details vary wildly. The growth-loop test doesn't. A loop that compounds in tier-3 India will compound in tier-3 Lagos, with localised content. A loop that doesn't compound in your home market won't compound abroad either — you'll just have new excuses for why.

That's the discipline. Find the user behaviour that's already there. Build something that earns the right to ride on top of it. Stay out of the way.

Most AI Products Are Built for the Wrong Half of the World

Products for Bharat Are Built in the Field, Not in the Boardroom

Sai skipped presentations and built real AI products.

Sai Gole was part of the March 2026 cohort at Curious PM, alongside 17 other talented participants.