Connecting on LinkedIn Is Not a Follow-Up

Article · 5 min read

Connecting on LinkedIn is an address change, not a follow-up.

The connect request delivers the feeling of closure without any of the structure a follow-up needs: context, a thread to pull on, and a reason for the other person to reply.

The handshake that ends in a QR code

Eleven minutes into a good conversation at the edge of the expo hall, someone reaches for a phone. QR code up, camera up, request sent, request accepted. Both of you say some version of great to meet you, I'll follow up. And then neither of you does, because it already feels finished. The tap closed the loop.

Three weeks later their name surfaces in a notification about a work anniversary. You can't remember which company they'd just moved to, what they were building, or why the conversation was good enough that you stood there while your coffee went cold. They're in your network. They're nowhere in your memory.

The directory illusion

Call it the directory illusion. The connect action delivers the psychological closure of a follow-up while providing none of the structure one actually requires.

A follow-up has to do three jobs. Prove you were paying attention. Hand the other person one specific thread they can pull on. Arrive at a moment when replying costs them nothing. A connection request does none of that. It carries no record of what was discussed, no signal of intent beyond a tap that looks identical whether the conversation ran eleven minutes or eleven seconds, and no mechanism that brings the person back to you at an interval you chose.

Being stored in someone's network and existing in someone's memory are separate states. Only one of them produces a reply.

A directory is working correctly when it buries you

The standard advice is to connect on the spot so you don't lose them. Reasonable enough, and it quietly treats storage as retention.

LinkedIn is built for discovery. Search, ranked feed, recommendations, degrees of separation: every core surface answers the question of who else is out there. Maintenance is a different problem and it needs different machinery. A field for what was actually said. A resurfacing schedule you set. A decay signal that flags a contact going cold. None of that lives in a connection graph, and arguably none of it should. LinkedIn is doing what a directory does, which is hold names accurately at enormous scale. Using it as a relationship system is a category error, and the error belongs to us.

The feed makes it concrete. Whatever your new contact posts gets ranked against hundreds of other candidates for the same attention, most of them from people you already know far better. Within hours, they're gone.

What a follow-up requiresWhat a connection graph gives youWhere it has to come from instead
Context of what was discussedNo field for itA note taken in the room
Proof you were paying attentionA tap, identical for everyoneOne detail only the two of you heard
Timing you controlFeed ranking decidesA date set before you leave the venue
A reason for them to replyNone implied by acceptanceA specific ask, offer, or answer
Survives ninety daysA name in a list of thousandsThe person and the reason stored together
Category-level comparison: what a follow-up has to carry versus what any connection graph is designed to store.

Why conference LinkedIn connections don't lead to anything

Search behavior tells on us. Google's People Also Ask box surfaces variations on how to follow up with LinkedIn connections after a conference, which is a strange thing to type if the connection were the follow-up. Nobody searches for how to follow up after sending a follow-up.

The numbers underneath the behavior aren't subtle. LinkedIn lets a single member hold up to 30,000 first-degree connections. Robin Dunbar's work puts the count of stable social relationships a person can actually maintain at roughly 150. Conference contacts disappear into the gap between those two figures, and no amount of scrolling closes it.

There's a category-level version of the same failure in capture tools. Across business-card and badge-scanner apps, the complaint that repeats in reviews has surprisingly little to do with scan accuracy. A cleanly parsed contact still lands in a list nobody reopens. Met's earlier pieces on the 72-hour memory cliff and on OCR being a feature rather than a product both circle the same drain: the capture step gets solved, the context step doesn't.

30,000
LinkedIn's cap on first-degree connections per memberLinkedIn Help Center
~150
Stable social relationships one person can maintainDunbar 1993

Two versions of the same Tuesday

Take a payments PM you met in the last hour of a fintech conference. She spent four minutes describing a reconciliation problem her team had been fighting for two quarters, and you named a vendor category she hadn't tried.

Version one: you connect in the aisle, she accepts before dinner. Tuesday morning she opens LinkedIn, sees eleven new connections from the same three days, recognizes maybe four faces, and moves on. You exist there as a name and a job title. There's nothing to respond to, so she doesn't.

Version two: in the ninety seconds after you part, you write down reconciliation, two quarters, tried building it in-house first, plus the vendor you promised to send. Tuesday morning that note surfaces. You send four sentences that open with her problem in her own words, then the link, then one question. She replies, because replying is easy and because the message proves the conversation actually happened.

Same person. Same event. The whole difference is a note that took ninety seconds and a date you set before you left the room.

A connection is an address. A follow-up is a reason to use it.

The work moves earlier than you think

If the thesis holds, the decisive moment isn't Tuesday morning at your desk. It's the ninety seconds after the conversation ends, while you still hold the details that make a follow-up land: the problem in their phrasing, the name they dropped, the thing you offered, the constraint they mentioned twice.

Wait until you're back at the hotel and you'll keep the job title and lose the specifics. Wait until Monday and you'll have a list of strangers with correctly spelled names. Context decays fastest right after the moment it's most abundant, so the capture has to happen in the room rather than in the recap.

So connect on LinkedIn anyway. It's a fine address book and a fine way to still be findable in two years. Just stop scoring the tap as contact made.

What Met is built around

Met's Event Mode exists for that ninety-second window. You stay in the conversation, capture the card or badge, and add two lines of context while the room is still loud. The person and the reason stay attached to each other instead of splitting across a contact list and your memory, and follow-up drafts start from what you wrote down rather than from a template that could have gone to anyone in the hall.

Storage is iCloud only. No ads, no notification spam, no contact data sold on. The people this is built for have been pitched a thousand times and can smell a data-broker business model through a landing page.

None of this makes LinkedIn worse at what it does well. It stops asking a directory to do the one job it was never built for.

Capture the context while you're still in the room, not after the room forgets. Get Met.

Met was built for the person who leaves a two-day conference with forty conversations and a stack of badges. Event Mode captures the card and the context in the same motion, keeps the record on your own iCloud, and hands you back the specific detail that makes a follow-up worth replying to. No ads, no notification spam, no contact data sold on.

Get Met in your inbox

Field notes on conference networking, follow-up timing, and what we ship next. No spam, no AI hype.

No spam. Unsubscribe anytime. Replies go to support@sailquery.com.