Acquisition funnels: how free products brought buyers to expensive courses
A three-step funnel — value for free, trust, a reason to buy now — with positive unit economics in every channel

60%
of company revenue came from acquiring new customers
−17%
cost of acquiring a buyer from events per quarter
52%
event margin
×11
payback of the game in the Telegram bot
119%
of the revenue plan delivered by free products in Q2 2026
8 → 60
promo-code payments from events per quarter
Context and goal
Product and market
A group of EdTech companies teaching design and product management, with a ticket of 30,000 to 120,000 rubles. Demand isn’t urgent: people put the purchase off and compare competitors, so the deal cycle is long — the decision takes weeks to mature.
Goal
Hit the quarterly revenue plan from new customers with positive unit economics in every channel, and keep acquisition costs from growing as the budget grows.
My role
I owned acquisition of new customers: free events, mini-courses and promo mechanics. I formulated and tested hypotheses, calculated unit economics and the payback of every channel, planned the quarterly budget and built the mechanics myself.
A three-step funnel
A cold audience doesn’t buy an expensive product on the first touch. First it needs to get value for free, then believe the product will solve its problem, and only then get a reason to buy now.
Value for free
Bring in a new lead and give them a result without payment.
Trust
Show the product from the inside and the people behind it.
A reason to buy now
Shorten the deal cycle and turn interest into a purchase.
The model fits any product with a long deal cycle: B2B services, financial products, expensive subscriptions.
Step 1. Value for free: events and mini-courses
Nobody buys an expensive course on the first touch, so the funnel entry has to be free and still pay off.
Why a free entry. Cold traffic doesn’t buy a 30,000–120,000-ruble course right away. The bet: a free product with real value that pays off through later purchases.
What we did. Two free entries into the funnel. The first: online events for designers about how their work is changing with AI, each with its own landing page for ads. The second: free mini-courses — a few hours of the real programme with an assignment and a promo code for the paid course at the end. AI Design Concepts worked best — six hours of the real programme: a lecture, a masterclass and an assignment.


×4.6
AI Design Concepts payback
The mini-course gave a working method and showed where to apply it in the paid course. Its payback is lower than the bot game’s, but it brought in buyers of expensive programmes.
15 days
From idea to mini-course launch
Built as a follow-up to the events: a lecture, a masterclass, an assignment and a promo code for the paid course at the end.
13 days
From mini-course to purchase
That’s how long it took on average to pay, so purchases without a promo code after the mini-course were credited to it too.
What didn’t work. Not every mini-course paid off: one gave away a work review for free — the main thing people buy the course for — and in another a week of access ended before people got started. Hence the rule: a free product should show the next step, and access should last long enough to reach the result.
Step 2. Trust: a series of events instead of one-offs
One touch doesn’t convince people to buy an expensive product: trust builds over several meetings.
Where money was lost. Events ran every week or ten days: the team didn’t have time to prepare, the audience got tired of announcements, and a one-off event rarely led to a purchase.
What we did. We moved to rare, larger events that showed a path through the whole course lineup, and linked them into a season: one page with a shared idea, the programme, sign-up for several events at once and recordings of past ones. The events showed the product from the inside — work reviews by art directors and alumni stories.
What came of it. Event revenue grew by 16%, and the cost of acquiring a buyer fell by 17%. And 61% of payments came from people who registered for two or more events.
› See the full season page

Step 3. A reason to buy now: discounts, a game and events at decision moments
A warmed-up lead puts the purchase off until a sale, and a course has a short sales window.
Where money was lost. Towards the end of sales, new people almost stopped coming to events, and the hesitant never decided: demand didn’t have time to turn into payments.
What we did. The event discount stacked with other offers, so buying right away became better than waiting for a sale. We launched a game in the Telegram bot: eight questions, every correct answer adds another 1% discount. And I rebuilt the event schedule around three decision moments: the start of sales, the end of early-bird prices and the last days before the course starts.
What came of it. Promo-code payments grew from 8 to 60 per quarter, the game paid back 11 times, and event payments grew by 28% with a 21% bigger budget. The flip side: the average order from events fell by 12%, so as the next step I planned a margin calculation for every pricing tier.

How I planned and measured the result
All three steps ran on numbers: a financial model before the quarter and honest attribution after it.
Where the numbers misled. The deal cycle is long: 58% of payments came more than a week after an event. If you count only purchases from the first few days, a good channel looks unprofitable. And by cost per registration, the best-looking channel was one that barely brought buyers.
What we did. Before each quarter I built a funnel model in three scenarios: conversion at every step, margin, ad payback and the safety margin — how far conversion can drop before a channel goes into the red. Every quarter ended with a report and 4–6 hypotheses for the next one.
Attribution model. I tied each payment to the last event before the purchase and counted it once. A short window covered payments within 7 days of an event, a long window — all later payments by the same person. The channel was set by the first touch, and duplicates were removed by e-mail and product. Channels were compared by payments rather than cost per registration.
What came of it. We reallocated the budget: the new channel brought a third of all payments from ads. Every channel had positive unit economics, and the event margin was 52%.

The result. In Q2 2026 free products delivered 119% of their revenue plan, and the whole acquisition stream brought in 60% of company revenue.
Which hypotheses worked
- 01
If a free product gives value and shows why to take the paid course, it pays off. The AI Design Concepts mini-course paid back 4.6 times — less than the game, but it brought in buyers of large programmes.
- 02
If events are rarer but bigger, acquisition gets cheaper. Event revenue grew by 16%, and the cost of acquiring a buyer fell by 17%.
- 03
If events are linked into a series, people buy more often. 61% of payments came from people who attended two or more events.
- 04
If a discount stacks with other offers, buying now beats waiting. Promo-code payments from events grew from 8 to 60 per quarter.
- 05
If the discount comes through a game, people get hooked and buy. The game in the Telegram bot paid back 11 times — mostly on short courses and basic tiers.
- 06
If events land on the moments people decide to buy, payments go up. Event payments grew by 28% with a 21% bigger budget.
«In the shortest time Ilya became an independent specialist who took responsibility for growth and for building entire marketing funnels. What strikes me most is Ilya’s love of business metrics.»