A B2B matchmaking event is easy to describe and hard to run. People sit at tables, each one gets a few minutes to talk, everybody moves, and it starts over. Anyone gets the idea in thirty seconds.

What nobody tells you is that almost every decision that determines whether it works is made weeks earlier, at a desk, and most of those decisions are arithmetic. This guide walks through them in order, with the math shown.

What a B2B matchmaking event is, and what it is not

A matchmaking event is a structured session where companies talk in short, controlled blocks and change counterparts at the end of every block. Three properties define the format.

Time is the same for everyone. Each participant gets the same minutes, in the same order, on the same clock. That is what stops the loudest person at the table from taking half the conversation.

Nobody meets twice. Two people who have already sat together should not sit together again. No repeat pairings is the line between a matchmaking event and a cocktail hour with assigned seats.

The schedule is built in advance. Who sits at which table, in which round, is settled before the doors open, not improvised at the check-in desk.

And what it is not: not a keynote with networking afterward, not a trade show floor with booths, not a happy hour with badges. In those formats the people who already know each other talk to each other, and whoever walked in alone walks out alone. Open networking puts fifty people in a room and hopes the right two find each other. Matchmaking schedules it.

It is also not one single thing. A chamber of commerce session for members, an Enterprise Europe Network brokerage event pairing companies across borders, a hosted buyer program bolted onto a trade show, an economic development organization putting suppliers in front of one large buyer: four different designs. The intent decides the numbers, so the intent comes first.

Decide the goal, and who is in the room

There is one question that solves half the problems of an event, and almost nobody asks it: what has to have happened for you to call this a success?

The plausible answers lead to different events.

Direct business between the people present. Complementary supply and

demand in the same room, and enough minutes for a proposal to start. Smaller tables, longer turns.

More contacts per person for a defined group. Volume of meetings. Bigger

tables, shorter turns.

Suppliers in front of one large buyer. An anchor company at a fixed

table, with everyone else cycling past it.

Activating a community that already exists (an association, a coworking

space, a franchise network). Recurrence and new faces each edition matter more than volume in any one edition.

Once the goal is set, the guest list mostly writes itself. And here is the unpopular part: turning down a registration beats accepting everyone. Sixty companies from the same niche, with nothing to sell each other, produce sixty polite conversations and zero deals. Curation is the cheapest filter you have.

The sizing math

This is the section most organizers skip and later wish they had not. Four numbers talk to each other constantly: participants, tables, seats per table, and rounds. You choose three. The fourth is a consequence.

Seats per table

Start here, because it moves more than any other number and gets discussed the least.

At a table of N seats, each person holds N minus 1 meetings per round. A table of 5 gives 4 meetings. A table of 8 gives 7. A table of 10 gives 9.

That looks obvious and has a consequence that is not. Bigger tables do not make the event messier. They make it shorter, because they deliver more meetings per round and need fewer rounds for the same coverage. The price is the speaking turn, which has to shrink so the round does not become half an hour in a chair. Smaller tables do the opposite: depth, real negotiation, a longer day.

There is no correct table size in the abstract, only consistency with what the invitation promised.

Tables, and the capacity check

Tables times seats is your capacity per round. If that number is smaller than the number of people registered, somebody is left waiting every round. It is not the end of the world, the waiting can be spread evenly, but it is a person standing idle at an event they paid for or cleared their diary to attend.

With 40 participants and tables of 5, you need 8 tables. With tables of 8, you need 5. The difference shows up in the room rental, in the catering cost and in the circulation space, which is what almost everybody forgets to measure when they go to choose the venue for the event. The number that matters on the site visit is not how many chairs fit. It is how many tables fit with people moving between them without a crush.

The ceiling on rounds, which nobody can break

Here is the most useful piece of arithmetic in this article.

If each round delivers N minus 1 new meetings, and there are P minus 1 people to meet in total, then the maximum number of rounds without repeating anybody is:

(P minus 1) divided by (N minus 1)

Go back to the 40 people at tables of 5. Each participant has 39 people to meet and holds 4 meetings per round. 39 divided by 4 gives 9.75. In other words: in the best imaginable case, 9 rounds fit, and even then each person finishes the event without having talked to 3 of the other 39. Full coverage does not close in that design, full stop.

The same 40 people at tables of 8 hold 7 meetings per round. 39 divided by 7 gives 5.57: 5 rounds at most, with 35 people met. Fewer rounds, a much shorter event, practically the same reach.

Two honest warnings about that arithmetic:

It gives a ceiling, not a guarantee. The ceiling is what the arithmetic

permits. Whether the plan actually exists, with nobody repeating and no table left short, depends on the arrangement closing. That is why the number of rounds is the last thing to be announced, and not the first. The subject has an article of its own: how many rounds fit your event.

"Everyone talks to everyone" is the privilege of specific designs. In our

software's tests, 25 people at 5 tables of 5 reach full coverage in 6 rounds; 49 people at 7 tables of 7 reach it in 8; 100 people at 11 tables of 11 reach it in 12; 300 people at 19 tables of 19 reach it in 20. Notice that the right answer is almost never to invite more people. It is to change the table size.

How long all of this takes

Do the arithmetic with every piece, because the forgotten piece is always the same one.

A round runs: (seats per table times the speaking turn) plus the gaps between turns plus the table changeover.

An example with tables of 6 and 3 minutes per turn: 18 minutes of speaking, plus 5 gaps of 15 seconds between turns, which comes to 1 minute and 15 seconds, plus 3 minutes for the whole room to stand up and find the new table. A total of 22 minutes per round.

Seven rounds at that pace add up to 2 hours and 34 minutes. Add 20 minutes of opening and check-in and 20 minutes of break in the middle, and your event runs 3 hours and 15 minutes. That is the number that goes in the invitation, in the venue contract and in the catering brief, and it is what sustains a matchmaking schedule that holds.

The classic error is timing the round and not timing the changeover. With 100 people, moving between tables takes two to three minutes. Multiplied by ten rounds, it is half an hour nobody put in the spreadsheet and that arrives all at once at the end of the evening.

The data that makes meetings worth having

An event with a perfect seating plan and no information about who is who delivers random conversations with excellent punctuality. Tidy, and thin.

What turns a seat into an opportunity is two fields on the registration form: what the company offers and what the company is looking for. They let you seat a seller next to a buyer instead of seating whoever registered first next to whoever registered second. That is the whole idea: fit between supply and demand becomes the criterion for building the table, and compatibility gets scored per meeting rather than guessed at.

Ask for industry, job title, and one interests field. Ask for nothing else. A long form kills registrations, and a field nobody reads only annoyed the person who filled it in.

One small rule pays for itself: keep colleagues from the same company off the same table. Two people from the same employer taking two of a handful of seats is pure waste, and they will talk at the coffee station anyway.

Registration and promotion

The invitation has to state three things precisely: who the event is for, how many conversations each person will have, and how long it will take. You have already calculated the last two. Promise exactly what the arithmetic delivers, and never write "you will meet everyone in the room" without checking first.

Open registration with its own link, collect the offer and demand fields from the very first sign-up (chasing them later is a guaranteed headache), and confirm attendance at least twice. If the registration export comes out in the exact format your scheduling tool imports, nobody retypes a list the night before.

Plan the no-shows into the count. For a free event, the gap between the registration list and the attendance list is always material, and a margin built in early is what prevents half empty tables next to full ones.

The day before

The day before is not the day you build the seating plan. It is the day you check the plan you already have. If the schedule only exists the night before, a step got pushed. What has to be finished:

The final list, with cancellations already removed.
The seating plan for every round, no repeat pairings, no table left short.
Badges with a number on them, not just a name. The number is what makes

the table change fast.

The table map printed, for you and for the floor staff.
The timings locked: turn length, gap between turns, changeover, break.
The audio cues tested in the actual room, at a volume that beats ninety

people talking.

And one test worth the whole evening: simulate an absence. Pull five names off the list and see what happens to the plan. If the answer is "I would rebuild everything by hand," you have a problem waiting at nine in the morning, because no-shows are not a hypothetical. They are routine.

Event day

The day has fewer decisions and more discipline. A short list actually matters.

Check-in that records attendance for real. Who showed up is not who registered, and the plan has to hold for the first list.

A clock the whole room can see. Without it, every table finishes at a different moment and the next round starts ragged. A second screen showing the round and the time remaining beats any amount of announcing over a microphone.

An audio cue before the end. Fifteen seconds is enough for a conversation to close instead of getting cut off mid sentence.

Announce the move, not the end. "Round 3 is over" leaves ninety people looking around. "Table 4 goes to table 9" moves the room.

Log incidents as they happen. Who was absent, who arrived late, which table ended up with one person at it. Nobody remembers afterward, and that is the material you want when you plan the next edition.

After: what to prove, and to whom

The rotation ends when the last table stands up. The event does not. It ends when you can say what happened there.

Record the plain numbers: how many people attended, how many meetings were scheduled, how many actually took place, and between how many companies. Those numbers carry three conversations: the sponsor who wants to know what they bought, the board that approves next year's budget, and the participant deciding whether to come back.

Send the post-event survey while the memory is fresh, and follow up on the connections that started at the tables. The value of a matchmaking event appears or evaporates in the two weeks afterward.

The mistakes that cost the most

Not the most common ones. The ones that cost the most when they happen.

Announcing the number of rounds before checking the design. You promise six, discover that four fit, and choose between disappointing the room and repeating pairings in front of everyone.

Promising full coverage without doing the division. "Everyone meets everyone" is a beautiful sentence and a very specific claim that most designs do not honor.

Treating the registration list as the attendance list. The plan has to work for the people who came through the door, and it has to be rebuildable fast.

Putting the break at the end. A break in the middle, when the energy dips, rescues the second half. At the end it is a line for the parking garage.

Collecting data you will never use, and not collecting offer and demand. That is what turns a matchmaking event into a seat lottery.

Building the plan by hand at scale. In a small room it is tedious and feasible. As the room grows it stops being an organizing task and becomes a combinatorial one, where the error is invisible in the spreadsheet and perfectly visible in the ballroom.

Where the software fits

Look back at what this guide asked of you, and what it asked of arithmetic.

You decide the strategy: the goal, the audience, the curation, the tone of the room, how long each person speaks, what the invitation promises. Nobody does that better than the person who knows their own market.

The rest is checking and combining: confirming the design closes, building the plan with no repeats, rebuilding it when five people do not show, running the clock, and counting what happened. Repetitive, sensitive to human error, and invisible until the day it fails in front of everybody.

That second layer is what Rodada de Sucesso handles. Describe the room and it tells you what that layout supports before the invitation goes out, builds the schedule under your rules, runs the timing on the day, and hands you the numbers at the end. It does not replace your curation. It takes off your shoulders the part that eats the whole night before and that should have been settled four weeks earlier.

How to run a B2B matchmaking event: the complete guide