Conferences and Meetups: How to Prepare, Work the Room, and Follow Up

The reason conferences disappoint is that a room full of strangers is not actually a room full of strangers, and behaves nothing like a random mixer. When researchers instrumented a real business networking party and recorded who stood near whom, the guests overwhelmingly spent their evening talking to people they already knew — while nearly all of them had said beforehand that their goal was to meet someone new.

What actually happens in the room

Paul Ingram and Michael Morris of Columbia Business School ran the experiment in “Do People Mix at Mixers?”, published in Administrative Science Quarterly in 2007. They hosted an after-work mixer for roughly 97 executives, mostly executive MBA students, in a 60-by-60-foot reception hall. Each guest wore an nTag — a wearable device detecting other tags within about eight feet — producing second-by-second records of who encountered whom over 80 minutes, with encounters counted from one minute of proximity.

The results are uncomfortable and useful:

• Ninety-five per cent of guests said their goal was to meet new people.
• Pre-existing relationships dominated. The paper reports that a strongly positive pre-mixer relationship made an encounter roughly 223 per cent more likely than between strangers.
• The average guest had about 14 encounters, split roughly half with people they already knew and half with people they did not.
• Demographic homophily was largely absent. Guests did mix across sex, race, education and job type. The barrier was not seeking out people like themselves; it was the pull of existing ties.

So the failure mode is not snobbery or cliquishness. It is that standing next to someone you already know is the path of least resistance, and 80 minutes disappears into it. Everything below is built around that fact.

Choosing events worth attending

This is craft. A few filters that hold up:

• Who else is going, specifically. If you cannot name six people you would want to meet, the event is a lottery ticket. Attendee, sponsor and speaker lists are the evidence.
• The ratio of talks to gaps. A packed single-track programme with 15-minute breaks is a lecture series. Long lunches, hallway tracks and unconference slots are where encounters happen.
• Whether it repeats. An annual event with the same community year after year compounds; a one-off does not.
• Whether you can get a job there. Volunteering, chairing a session or giving a lightning talk converts you from an attendee into someone people have a reason to approach. This is the highest-leverage single change available.
• Size. Smaller events are usually better for making ties; very large ones are better for sampling an industry and for the side events around them.

Canadian examples worth knowing

A few real, current fixtures, offered as illustrations of the categories rather than recommendations:

• Large industry gathering: Web Summit Vancouver, next running 25–28 May 2027 at the Vancouver Convention Centre.
• Mid-sized, community-flavoured: Startupfest in Montréal, held at the Grand Quai du Port de Montréal, which describes roughly a quarter of its attendees as investors and the rest as founders, corporate innovators and ecosystem people.
• Year-round hubs rather than events: the MaRS Discovery District in Toronto, founded in 2001 and describing itself as North America’s largest urban innovation hub, and Communitech in Kitchener, a membership organisation running events and programmes for the Waterloo Region tech community.

Beyond tech, the same structure exists in every profession through provincial and national bodies — accounting, engineering, law, marketing — whose regional chapters are usually the smaller, higher-yield version of the national conference.

Before: the preparation that changes the outcome

1. Pick five to eight names from the speaker or attendee list. Not a type of person — actual names. This is the step most people skip.
2. Write one specific question for each. Something only they can answer, drawn from a talk they gave or a project they shipped. “How did you handle the data-residency requirements?” is a conversation. “What do you do?” is not.
3. Prepare three things you know that seem dull to you and are scarce to others — what a vendor really costs, why a project was cancelled, what a regulator said. You need something to trade.
4. Write your own two-sentence answer: what you work on, and what you are useful for.
5. Plan the gaps, not the sessions. Decide in advance which breaks and side events you will spend on people. Sessions are recorded; hallways are not.

During: joining and leaving conversations

The Ingram and Morris result says the enemy is inertia, so the mechanics matter more than charisma.

Joining. Approach groups of three or more rather than pairs — a pair is often a private conversation, a trio has an opening. Stand at the edge of the circle, catch the eye of whoever is not speaking, and wait for a natural pause. Then enter on what you heard: “Sorry to interrupt — did you say you migrated off it entirely?” Referencing the content proves you were listening and skips the introductions ritual.

The opening question. “What brought you to this one?” outperforms “what do you do?”, because it produces a reason rather than a job title, and reasons are things you can respond to.

Leaving. A bad exit undoes a good conversation. The reliable form has three parts: a signal, a specific compliment, and a next step. “I’m going to catch the next session — but the thing you said about staged rollouts was useful. Can I send you the write-up we did?” Offering to send something specific beats asking to connect, because it manufactures a legitimate reason for the follow-up to exist.

The uncomfortable rule. If you arrive with colleagues, split up. The data says you will otherwise spend the evening with them, and you can talk to them any day.

One reassurance that is evidence-based rather than motivational: people are consistently wrong about how badly these approaches will go. The research on requests to strangers finds compliance underestimated by roughly half, and the research on conversations finds people underestimate how much their partner liked them. Your read of the room is biased in a known direction.

Notes and a CRM-lite that you will actually maintain

You will not remember. By the second day, faces detach from conversations and you are left with a stack of names and nothing to say to any of them. The minimum viable system, again craft:

• Capture within five minutes of the conversation ending, standing in a corridor, on your phone. Not at the end of the day.
• Record four fields: name, where you met and what you discussed, what you promised to send, and anything personal they volunteered. The fourth is what makes the follow-up feel human rather than transactional.
• Use one list, not four. A single spreadsheet with a row per person beats a beautiful tool you abandon. Columns: name, organisation, met at, topic, promised, last contact.
• Mark the ones worth an annual message. Most do not need one. The handful working somewhere adjacent to your world, whom you rarely speak to, are the ones the weak-ties research suggests are worth the upkeep.

After: the follow-up window

The convention is to follow up within 24 to 48 hours. That number is craft, not a research finding — nobody has run the trial. Its logic is worth stating plainly: it is the window in which both of you can still reconstruct the conversation without effort, and before the other person’s inbox has absorbed the rest of the event.

What the research does say is that written requests to people who barely know you are weaker than you think. Roghanizad and Bohns found in the Journal of Experimental Social Psychology in 2017 that people overestimate compliance with emailed requests, because they do not anticipate the suspicion an unsolicited message attracts. The follow-up therefore has to prove it is not cold outreach:

1. Reference the conversation specifically enough that it could not be a template. Name the thing they said.
2. Deliver whatever you promised, in that first message. If you promised nothing, send something anyway — this is what the preparation was for.
3. Ask for nothing in message one. An introduction, a call or a job belongs in message two, weeks later.
4. Make the reply easy. A message that requires no response is fine; one that requires a decision is not.

Then close the loop on your list: mark who you contacted, and diarise one check-in three to six months out for the few contacts worth maintaining.

The practical version

Treat the event as three tasks, not one. Before: choose it because you can name six people you want to meet, prepare a specific question for each, and get yourself a job at it if you can. During: split from your colleagues, approach threes rather than pairs, enter on content rather than introductions, and leave with an offer to send something concrete. After: write notes in the corridor rather than at home, send the promised thing within a day or two with no ask attached, and keep one boring list you will actually update. The measured tendency in every room is to drift back towards people you already know, and a good plan is mostly a set of small decisions that stop you doing that.