What Is Coffee Roulette?

What is coffee roulette?

Coffee roulette is a structured program that randomly matches two colleagues from different parts of an organisation and invites them to have an informal chat over coffee. The pairings rotate on a set cadence — weekly, fortnightly, or monthly — so people meet someone new each round rather than the same handful of colleagues.

It goes by several names. Random coffee, random coffee chats, and virtual coffee all describe the same thing. Some organisations call it a coffee lottery or randomised coffee trial. The mechanics are identical: remove the choosing, keep the meeting short, and let the conversation go wherever it goes.

The defining feature is that there is no agenda. A coffee roulette pairing is not a one-to-one, not a mentoring session, and not a project sync. Nobody has to prepare anything or report back. That absence of structure is the point — it's what separates it from the meetings people already have too many of.

How does coffee roulette work?

Most programs run on the same four-step loop:

1. Opt in. Employees join voluntarily, usually by joining a Slack or Teams channel or adding their name to a list. Participation is never mandatory — a coffee roulette people are forced into stops being a coffee roulette.
2. Match. At the start of each round, an algorithm or coordinator pairs participants, deliberately favouring people who don't share a team, a manager, or a recent pairing history.
3. Meet. Each pair gets an introduction with both names and a nudge to find 20–30 minutes. The pair schedules it themselves — remote teams do this over video, co-located teams walk to the nearest coffee machine.
4. Repeat. The cycle runs again on the chosen cadence, with fresh pairings.

The cadence matters more than most people expect. Weekly rounds burn out volunteer participants quickly. Fortnightly and monthly rounds sustain far better, because a single 30-minute conversation a month is easy for almost anyone to justify.

Why should you consider coffee roulette?

The short version: informal relationships at work are not a soft benefit, and the research on what happens without them is unusually consistent.

Gallup's State of the Global Workplace: 2024 report found 1 in 5 employees worldwide experienced loneliness the previous day — a rate that hasn't improved despite widespread wellbeing spending. In the US, Cigna's 2025 Loneliness in America report (n=7,500) found more than half of working Americans feel lonely.

That has a price attached. UK Government research (2021) put the cost of workplace loneliness to employers at £2.5 billion per year, with 64% of it — £1.62 billion — coming directly from increased turnover.

On the upside, Seramount's Belonging Toolkit (2024) , citing Harvard Business Review research, links a high sense of workplace belonging to a 56% increase in job performance, a 50% drop in turnover risk, and a 75% reduction in sick days. Coffee roulette is one of the cheapest interventions that acts directly on that variable.

What problems does it actually solve?

Four specific ones, and it's worth being precise about which:

Departmental silos. Cornell ILR/CAHRS research (2019) found informal, non-work conversation is a primary driver of cross-functional idea generation. Formal meetings optimise for delivery; discovery happens elsewhere. A 2018 study in Industrial Marketing Management quantified the chain — informal networking drove cross-functional knowledge sharing, which predicted organisational innovativeness (b=0.61, t=12.19) and firm performance (b=0.66, t=13.50).

Remote isolation. A 2024 peer-reviewed meta-analysis of 49 workplace loneliness studies found loneliness correlates with lower job performance (r=−0.35), reduced satisfaction, and burnout across every industry studied — and that structured social connection, not general wellness support, is what reduces it. Read more on remote isolation →

Onboarding. New starters in distributed teams meet their immediate team and almost nobody else. Coffee roulette gives them a route into the wider organisation that doesn't depend on their manager making introductions.

The vanished water cooler. Read the research on rebuilding it →

How do you run coffee roulette?

You have two honest options, and the manual one is genuinely viable for small teams.

The manual method. Keep a list of participants in a spreadsheet. Each round, shuffle the list and pair adjacent rows — a single RAND() column sorted ascending does it. Keep a second sheet logging past pairings so you can spot repeats. Message each pair yourself, or mail-merge it. For a team of 20 or 30, this takes about fifteen minutes a round and costs nothing.

Its limits show up fast. At 100+ people, checking for repeat pairings by eye stops working. You have to remember to run it — and the round that gets skipped because the coordinator was on holiday is usually the one that kills the habit. And a spreadsheet doesn't know your org chart, so it will eventually pair someone with their own manager's manager, which is the single most common way these programs generate awkwardness rather than connection.

The automated method. A dedicated tool handles pairing, delivery, and history. Most run inside Slack or Microsoft Teams so nobody has to install anything new.

Manual vs automated: which should you choose?

If you have fewer than 50 people and someone genuinely enjoys running it, start manual. You'll learn what cadence your team tolerates before paying for anything. Move to a tool when you notice rounds getting skipped — that's the real signal, not headcount.

What should you talk about? 20 coffee roulette questions

The most common failure is two people staring at each other for the first four minutes. A prompt fixes it. These work because they're specific enough to answer and open enough to go somewhere.

Getting started

1. What are you actually working on this week?
2. How did you end up in this role?
3. What did you think this job would involve before you started?
4. Which team here do you understand the least?
5. What's something your team does that the rest of us probably misunderstand?

Work, sideways

6. What's the most useful thing you've learned in the last year?
7. What part of your job would you happily never do again?
8. What's a decision you've changed your mind about?
9. Who else here should I be talking to, and why?
10. What's broken that everyone's quietly working around?

The person, not the job

11. What were you doing before this?
12. What do you do that has nothing to do with work?
13. Where did you grow up, and how different is it from here?
14. What's the best thing you've read, watched, or listened to recently?
15. What's a skill you have that never comes up at work?

Better than "any plans for the weekend?"

16. What's something you've changed your mind about recently?
17. What would you do with an unexpected free day?
18. What's the best advice you've been given that you ignored?
19. What are you looking forward to?
20. What should I have asked you?

Steal these outright. Pasting five of them into the pairing message measurably improves how the first conversation goes.

When coffee roulette doesn't work

Worth saying plainly, because almost nobody writing about this does:

It fails when it's mandatory. The moment attendance is tracked as a performance signal, it becomes a meeting. Opt-in is not a nice-to-have; it's the mechanism.

It fails in very small teams. Under about 15 people, everyone already knows everyone. There's no distance for the random pairing to close, and you'll exhaust the novel combinations within two rounds.

It fails without hierarchy awareness. Pairing a graduate with a director two levels above them doesn't produce a relaxed chat — it produces a nervous graduate preparing for it like an interview. This is the most common design flaw, and manual and automated programs both fall into it.

It fails at the wrong cadence. Weekly is too often for most organisations. The people who drop out first are usually the busiest and most senior — exactly the ones whose participation makes it credible for everyone else.

It won't fix a culture problem. If people don't talk to each other because trust is broken, random pairing won't repair that, and running it anyway reads as tone-deaf. Coffee roulette solves distance. It doesn't solve distrust.

Coffee roulette tools compared

Donut is the established option. It runs inside Slack and Microsoft Teams, covers intros, watercooler topics and celebrations, and reports 20,000+ teams using it. The free tier covers one channel and up to 24 users per Intros round. If you want the safest, most proven choice, it's this one. Full comparison →

coffee-roulette.com is a standalone web tool that matches employees by the days of the week they have in common and emails the introductions. No Slack or Teams integration.

coffeeroulette.org is a free browser-based pair generator. You paste in a list, it returns pairings, and it can import previous rounds to avoid repeats. There's no scheduling or delivery — it's a generator, not a program.

Spreadsheets remain a legitimate answer under 50 people, as above.

TeamBlend — this site — reads your org chart before pairing, specifically to avoid manager-report and steep-hierarchy matches, and delivers through Slack, Microsoft Teams, or email. We're in free early access, which is the honest framing: no credit card, and no customer case studies yet either.

What is a coffee roulette?

A program that randomly pairs two colleagues who don't normally work together for a short, informal chat, repeated on a regular cadence. Also called random coffee.

How do you run a coffee roulette?

Collect opt-ins, randomly pair participants each round while avoiding same-team and repeat matches, send both people an introduction, and let them schedule 20–30 minutes themselves. Repeat fortnightly or monthly. A spreadsheet handles this under 50 people; past that, use a tool.

How do you pronounce coffee roulette?

Roulette is pronounced roo-LET, with a silent final e — the same as the casino game it's named after.

How long should a coffee roulette chat be?

20 to 30 minutes. Long enough for a real conversation, short enough that nobody has to move anything in their calendar.

Does coffee roulette work for fully remote teams?

Yes, and the case is stronger there. Remote teams lose the incidental contact — hallway conversations, shared lunches, chance encounters — that builds workplace relationships without anyone planning them. Gallup's 2024 data puts 1 in 5 employees worldwide experiencing loneliness the previous day, and distributed teams have no default mechanism to counter it. Structured pairing is one way to rebuild it deliberately.

Is coffee roulette the same as random coffee?

Yes. Different names for the same practice, along with virtual coffee, coffee lottery, and randomised coffee trials.

How many people do you need?

At least 15–20 for the randomness to produce genuinely novel pairings. Below that, you don't need a program.