Why We Sold an English-Speaking Course in Hinglish

I joined Entri as a Growth Manager in August 2020 and stayed two years, leading user acquisition for the Hindi Upskill category across paid and organic channels, and setting up lifecycle automation for retention and conversion. Entri is one of India's larger learning apps for government and private job preparation. The spoken-English course I worked on was led and mentored by Awal Madaan, a partner at Entri and a well-known creator in the language-learning space.

When I picked up the category it was doing about ₹5L a month in revenue on ₹1–2L of monthly ad spend. Most customers were in India, with a meaningful tail in the US, UK, Canada and the UAE — largely Hindi-speaking diaspora.

The obvious playbook, and the one we ran for a long time, was segmentation.

We built dedicated landing pages for distinct audiences — housewives, office-goers, people who needed professional or corporate English — and A/B tested targeting against each. The logic was standard and defensible: different people want spoken English for different reasons, so speak to each motivation separately. A homemaker rejoining the workforce and a mid-career professional preparing for client calls are not the same buyer.

The tests ran. The pages were fine. The numbers stayed roughly where they were.

What took us a while to see is that we were optimising a variable that wasn't the bottleneck. Every one of those pages, however carefully targeted, was written in polished English. We were varying the message across pages while holding constant the thing that was actually costing us conversions — the register the message arrived in.

That's the general lesson, and it's worth more than the Hinglish finding itself: when a set of well-built segment variants all underperform in the same way, stop refining the variants. The problem is almost always in what they share, not in how they differ. Segmentation is a powerful lever, but it can only redistribute demand you're already capturing. It can't fix a barrier sitting upstream of all of it.

We rewrote the funnel in Hinglish — Hindi rhythm and sentence structure, Roman script, English only where the English was the product being demonstrated.

This is less obvious than it looks in hindsight. The instinct across edtech is that an English course should present itself in polished English, because the page doubles as a proof of the product. Look how good our English is. But that logic inverts the moment you think about who is reading. Anyone confident enough in English to be persuaded by an English sales page has largely solved the problem the course exists to solve. The people who need it most are the ones for whom the page is hardest.

So the landing page had a requirement the product didn't: it had to be effortless. Every sentence a reader had to work at was a sentence in which they quietly concluded this wasn't for them.

Worth saying plainly: nobody at Entri fought this. There was no heroic internal battle. The team was aligned on running it as an experiment, and it was treated as a test rather than a bet. That mattered more than it sounds — a team that will run an experiment against its own category instinct is a structural advantage, and it's the reason the finding surfaced at all rather than dying in a review meeting.

The mechanism is about removing friction at the exact moment the reader assesses themselves.

A landing page for a skill-gap product isn't only an argument for the product. It's a test the reader administers to themselves. If they struggle through the copy, they've just collected evidence that they're behind — and the response to that isn't "I should buy this," it's embarrassment, then exit. If the copy is effortless, they stay in a frame where they can picture themselves succeeding, which is the frame in which people actually buy.

Hinglish did two things simultaneously: it removed the comprehension tax, and it signalled that the course was built by people who understood where the learner was starting from. The second is trust, and trust is most of what a first-time edtech buyer is deciding on.

This generalises well past language. Any product bought to close a gap — financial literacy, fitness, technical skills, career transitions — carries the same trap. The marketing gets written in the voice of the finished state rather than the starting state. It flatters people who have already arrived and quietly excludes the people who were going to buy.

The second thing that moved numbers: we ran Awal's own organic content as Facebook and Instagram ads, largely as-is, instead of the polished video our production team was making. It outperformed on both click-through and conversion rate.

I want to be careful with this one. In 2026 the finding is unremarkable — creator-led and UGC-style ads beating studio production is now a default assumption in performance marketing. It was not the default in 2020, and arguing to run a mentor's raw organic content over professionally produced creative was a real argument at the time.

I don't have the CTR and CPA deltas from that period, so I won't quote figures I can't stand behind. The directional finding is what I'd claim: content that had already earned attention organically kept earning it in paid, and content built to look like an ad performed like one.

Going from roughly ₹1.5L to ₹30L a month in spend, the Hinglish direct-response funnel stopped being enough. Two things had to change.

The direct landing-page purchase model hit a ceiling. Selling the course straight off the page worked beautifully at lower spend and then flattened. Past a point we weren't able to keep buying incremental purchases efficiently through pure click-to-buy campaigns — the audience willing to make an unassisted online purchase decision for a course is finite, and we exhausted it before we exhausted the market.

We had to raise the ceiling on revenue per buyer. Scaling 20x on volume alone wasn't available, so we introduced higher-ticket courses. That changes the sale: a higher price point isn't a decision most buyers make from a landing page alone.

What we landed on was a hybrid, and it's the part I'd carry into any similar account. The landing page still sold the course directly for buyers ready to transact. Alongside it, a WhatsApp button captured people who weren't — turning them into queries rather than losing them as bounces. Those conversations let us assess what the person actually needed and pitch the plan that fit, including the higher-ticket options that would never have converted cold off a page. Campaign objectives shifted accordingly, from purchase-optimised to lead generation.

The underlying pattern: direct-response course selling has a spend ceiling, and past it you scale by adding a human to the loop and raising average order value — not by pushing harder on the same funnel. The mistake I see repeatedly is teams treating a flattening direct funnel as a creative or targeting problem, and burning months optimising inside a model that has structurally run out of room.