A new-format hostess service: a mobile app that connects restaurants with influencers. The venue gets guests and posts, the influencer earns a commission for an actual visit. I ran the project as product and project manager: assembling the team, designing the model and its metrics, and driving acquisition on both sides.
Below is how this hostess service works, which metric turned out to be the only one that genuinely mattered, and where the risks hide in a model like this.
The problem: advertising you pay for on results
A restaurant that wants to promote itself through bloggers normally pays up front: for a post, for a story, for a visit with a camera. What follows is guesswork. Will guests actually come, how many, does the placement pay for itself? Nobody answers those questions, because there is no link between the publication and someone being seated at a table.
Influencers with a small but genuinely engaged audience face the mirror image of that problem. Approaching a restaurant directly is hard: the venue has neither a process for it nor any trust in an unfamiliar account, and haggling over every placement costs both sides more time than it is worth.
How the hostess service works
The hostess service is built as a two-sided marketplace in which the unit of the deal is not a placement but a booking.
- The restaurant posts its terms: a commission per guest who shows up and a separate rate for a publication.
- The influencer books a table through the app, and the booking is tied to their account.
- The venue confirms the visit and the bill amount, and only then is the commission credited.
- The publication is paid separately, as an add-on rather than the basis of the deal.
The point of the design is that the restaurant pays after the result rather than for a promise of reach. That removes the venue’s main objection and, at the same time, filters out accounts selling numbers instead of influence.
The metric that mattered: show-up rate, not reach
In the early versions of the model we counted what everyone counts: reach, active venues, active influencers, average bill. It quickly turned out that almost all of it was vanity. The one number that decided whether the platform lived was the share of bookings that ended in a real visit and a paid bill.
The reason is simple. A restaurant counts revenue and table occupancy, not followers. A booking nobody shows up for is worse than no booking at all: the table sat empty during prime time. So the show-up rate became both the quality metric of the marketplace and the basis for an influencer’s rating inside it.
Everything else stayed as supporting data: acquisition cost per venue and per influencer, the share of repeat bookings, the average bill by category. Those help you understand the economics, but the decision on who gets access to the best venues was made on show-up rate.
Risks the product had to account for
Fake bookings. The moment a commission is tied to a visit, there is an incentive to fake one: an arrangement with a friendly manager, a visit logged with no bill, a booking made and abandoned. Confirmation by the restaurant and a link to the bill amount are not bureaucracy, they protect the economics of the platform.
Content mismatch. The venue expects one style, the influencer shoots another. We handled that at the entry point: the restaurant states in advance what it considers acceptable, and because the publication is a separate line item, a disputed shoot does not blow up the whole deal.
The cold start of a two-sided market. Influencers will not join where there are few venues, and venues will not sign up where there are few influencers. We went by geography: build density in one district so there is something to choose from in the app, then expand. The same principle as in eXpresso Coffee, where demand was first proven in a single city.
What I took away from the project
Two-sided platforms look elegant in a deck and live hard in reality: the same commission has to be large enough to interest an influencer and small enough for the restaurant to consider it a good deal. The gap between those two numbers is the entire business.
The second lesson is about trust. Paying on an actual visit looks fair to everyone, but it only works if both sides confirm honestly. Verifiability of the event mattered more than any app feature, which is the same problem I later solved by technical means in PharmAPI. On the wider approach to assessing risks like these, I wrote separately about integrated risk management.
My role in the project
- Product and project management. Setting the team’s tasks, development priorities, timelines and acceptance.
- Business model. The commission structure, the terms for both sides, and the payout logic.
- Hostess service metrics and analytics. Choosing the numbers decisions were made on, and dropping the vanity ones.
- Acquisition for the hostess service. Working with venues and with influencers, including launch terms for the first participants.
Frequently Asked Questions
What is the hostess service in this project? A mobile app connecting restaurants with influencers: booking through the app, visit confirmed by the venue, commission on results. Posts are paid separately.
How is this different from ordinary influencer advertising? The restaurant pays for a guest who arrives, not for promised reach, so the result shows up in revenue.
Which metric was the key one? The show-up rate, meaning the share of bookings that ended in a visit and a paid bill. Everything else proved to be vanity.
How were fake bookings handled? The commission was credited after the venue confirmed the visit and tied it to the bill, and the show-up rate fed the influencer’s rating.
Need a Consultation?
If you are launching a marketplace and are not sure which of its metrics is the real one, book a free 15-minute call. We will go through your model and the points where it may fail to add up.