Nobody in India plans a Goa trip alone. There is a WhatsApp group, a friend who never replies, and one person who finally says "leave it, I will book". That person compares stays, collects money and chases everyone for their share. I named him Arjun the Group Planner, and he became the centre of my Airbnb India assignment. Win Arjun and you win five to eight travellers at once, without paying for a single advertisement.
The brief asked for three things: product-led growth initiatives for Airbnb India, MVPs for the larger features, and an updated India revenue model. As before, the financial projections are my own declared-synthetic modelling and not company data.
Part 1: Growth that the product does on its own
Product-led growth means the product brings the next user. Marketing spend is a tap: turn it off and growth stops. A loop is a well: each use refills it. I designed six enablers (cross-functional support such as payments, trust and local language) and six loops. The strongest ones follow Arjun's actual behaviour.
| Loop | How one user brings the next |
|---|---|
| Group trip | Planner invites friends to split the cost. Each friend must open the app to pay. |
| Guest to host | A guest with a spare flat or family home sees how simple hosting was from the other side. |
| Trip recap | A shareable summary after the stay reaches friends who were not on the trip. |
| Long weekends | The calendar itself prompts the next plan, well before the holiday. |
Part 2: Four MVPs under one contract
Each MVP was written against the same experiment contract: a hypothesis, the cheapest prototype that can test it, a success number, and a decision rule agreed in advance.
Trip Split lets the group pay their shares directly. Host-in-15 gets a first-time host listed in fifteen minutes and nudges happy guests to become hosts. Trip Recap, paired with multilingual guides, turns a finished stay into something worth sharing. Coins with a long-weekend planner rewards repeat booking around the Indian holiday calendar.
An MVP is the smallest test that can prove you wrong, not the smallest product you can ship.
Trip Split, for example, does not need a payments rebuild to test. A simple link that shows each person's share and records who tapped "pay" tells you whether groups want it. If they do not, you have saved a quarter of engineering time.
Part 3: A revenue model that protects supply
I proposed five layers: a re-based service fee, a Host Pro subscription, a traveller membership I called Yatra Pass, sponsorship, and partner or B2B income. On the fee, I modelled a 10% host fee (15.5% for professional hosts) and a 5% all-in guest fee shown upfront.
The upfront part matters in India. Travellers here compare final prices across apps and react badly to charges that appear on the last screen. And the host side needs care: raise fees carelessly and hosts list elsewhere, which empties the shelf that every loop above depends on. In my synthetic projection, the five layers together reach about ₹1,400 crore by FY29 at a 17.6% blended take rate.
What I took away
Growth, experiments and revenue are usually taught as three chapters. In practice they are one system. A loop decides which MVP to test first, and the revenue model decides whether the loop is worth feeding. Pull one lever without checking the other two and you get a busy quarter with nothing to show.
Public figures drawn from Airbnb quarterly results, the Oxford Economics India report, the Airbnb India newsroom, Skift and NITI Aayog. Fee levels, layer sizes and FY29 revenue are synthetic assumptions with a register in the annex.