On Saturday 15 August 2026, our London launch party at The Beehive Clapham became a real-life dating app pilot. Sixteen adults joined the Ember room. By the time it closed, they had sent 50 private waves and made nine mutual matches.
Those are encouraging numbers. They are not proof of how every pub night will behave. This was one venue, one night and a small group. What the pilot does give us is a first, useful look at what happens when a dating app starts with people who are already in the same room.
Here is what happened, what the numbers can and cannot tell us, and what we will change for the next Ember night.
Why run a real-life dating app pilot in a pub?
The appetite for meeting face to face is clear. Ipsos found that 59% of Britons prefer meeting potential partners in real life. Yet wanting to meet naturally does not make the first move feel easy.
That is the gap Ember is designed to close. The pub solves the distance problem because everyone is already in the same place. A private wave then makes it possible to show interest without turning an unreturned hello into a public moment.
This does not mean existing dating apps have stopped being useful. Swipe apps optimise for conversations before a meeting. Ember optimises for the room you are already in.
We explored the wider shift in our article on dating app fatigue and meeting at the pub. The Beehive pilot was the practical test: could a private digital signal make saying hello in person feel easier?
Real-life dating app pilot results at a glance
All pilot figures below come from Ember's anonymised room metrics for The Beehive Clapham on 15 August 2026. The room closed at 3am and the metrics were finalised on 16 August. These counts exclude guest identities and message content.
- 16 unique guests joined the room. None had a previous recorded Ember visit to this venue.
- Guests sent 50 wave attempts. 46 remained active at close and four were withdrawn.
- 13 of 16 guests sent at least one wave. All 16 received at least one.
- Nine mutual matches formed, involving 11 distinct guests.
- Five in-app messages were sent across four matches.
- Three match reward codes were created and one was redeemed. The remaining six matches followed the loyalty points path used by the pilot settings.
- No safety reports were filed.
That means 69% of the room made at least one match. Because one guest can match with more than one person, the nine matches should not be read as 18 different people. The distinct count is 11.
This is also why we are publishing the denominator. “Nine matches” sounds impressive on its own, but “11 of 16 guests matched” is the more useful result. It gives future pilots something honest to compare with.
It was social, not simply speed dating
The intent choices revealed one of the night's most useful lessons. Six guests selected flirting, four selected finding friends, four selected hanging out and two selected being up for a chat.
In other words, 10 of the 16 guests chose a primary intention other than flirting. A real-life dating app can still create a broader social room. That matters because not every good introduction needs a romantic target before the evening starts.
It also supports the idea behind our article on why the pub works as a third place. People need low-pressure places where conversation can happen without a formal itinerary. Ember's job is to make the first signal quieter, not to turn a pub into a line of forced mini-dates.
The strongest signal was participation
Thirteen people sent a wave, while every person in the room received one. Four waves were withdrawn before close, which is a useful reminder that changing your mind must remain easy and private.
Ember only reveals a wave when it becomes mutual. An unreturned wave stays between the sender and their phone. The Beehive numbers suggest that people were willing to use that low-pressure first step. They do not tell us why any individual wave was or was not returned, and that is deliberate.
The nine matches began early. The first formed at 5.39pm, 15 minutes after the first guest joined, and matches continued until 9.33pm. That spread suggests an Ember night should stay visible throughout the evening rather than being treated as a single scheduled activity.
The phone stayed in a supporting role
Only five messages were sent across four of the nine matches. We cannot infer from the data that every matched pair spoke face to face. We can say the product did not turn the pilot into long in-app conversations.
That is the design goal. Everyone shown in the room is already at the same venue. A short message can help someone say where they are sitting, but the useful next step is nearby, not days away.
The system's recorded presence averaged 135 minutes, with a median of 111 minutes. Eleven of the 16 guests recorded at least an hour of presence, and six recorded at least two hours. This is a directional measure based on join and last-seen activity, not an independently measured pub dwell time, so we will not treat it as a venue sales figure.
What the pilot taught us operationally
Arrival was gradual. The first guest joined at 5.24pm, the busiest arrival hour was 6pm, and the final guest joined at 11.14pm. Four guests joined during that peak hour, while the rest arrived across five other hours.
That gives us a simple venue lesson: signage and staff awareness need to last all night. A single welcome at the start would have missed much of the room.
The match reward flow also completed successfully. The first redemption was recorded 23 seconds after its code was created. One redemption from three issued codes is too small a sample for a reliable rate, but it proves the end-to-end staff flow worked during a live night.
We also recorded 38 rejoin events across 10 guests. A rejoin can mean reopening the app or reconnecting a session. It does not mean 38 extra venue visits. For the product team, it is a prompt to keep session recovery smooth and to explain clearly that returning to the room does not reset the night.
No safety reports were filed. That is good to see, but a group of 16 cannot validate a safety system. Private waves, blocking, reporting and the 18+ gate remain core requirements for every future night.
What happens after the Beehive pilot?
The next step is repetition, not a victory lap. We want to run more nights with different room sizes, arrival patterns and intent mixes before treating any percentage as a benchmark.
We will also test three practical improvements:
- A clearer one-line explanation of private waves at the point of joining.
- A shorter match reward explainer for guests and venue staff.
- An optional end-of-night question asking whether Ember helped someone start a real conversation, without asking who they met.
That final point matters. Product activity can show a wave or a match. It cannot prove that two people enjoyed talking. A respectful, optional prompt is a better way to learn without turning a social night into surveillance.
Beehive pilot questions
Was this a speed-dating event?
No. Guests joined one shared venue room and chose their own intention for the night. Most selected friends, hanging out or chat rather than flirting.
How does an Ember wave work?
A wave is private. The other person only learns about it if they wave back. If the interest is mutual, both people see the match at the same time.
Can someone join the room from home?
No. Ember checks that guests are near the venue before they join. Rooms are for adults aged 18+, close with the venue and are wiped within 24 hours. Guests can read more in how Ember nights work.
The Beehive pilot was small, lively and genuinely useful. It showed that a real-life dating app can help people take a private first step without becoming the whole night. If you want to see where Ember is running next, check tonight's rooms. Venue teams can learn more about bringing Ember to their pub.