The best conversation at the conference gets the worst follow-up email.

The one you didn't write down

Tuesday afternoon, somewhere between the espresso cart and hall B, you have the conversation that justifies the whole trip. Twenty-five minutes standing up. The other person is sharp, working on something adjacent to your problem, and by the end you're both saying some version of let's keep talking. You don't write anything down. Why would you. This one you'll remember.

Friday you sit down to clear the pile. The badge scans get a template, which is fine, because they were always going to get a template. Then you reach the good one and something has gone wrong. You know you liked them. You know it was a good conversation. The specifics have gone soft: the company name is fuzzy, the thing they were stuck on is gone entirely, and the reason the two of you were going to be useful to each other has dissolved into a pleasant feeling with nothing attached to it.

So you write the safest email you've sent all week. Great meeting you at the conference. Would love to continue the conversation. Let me know if you're ever in town.

That email is worse than the templates. The templates were honest about being templates. This one arrives from someone who clearly enjoyed the conversation and clearly cannot remember what it was about.

Warm and blank

Two different things decay at two different speeds after you meet someone. How you felt about them holds. What you learned from them does not.

Call the result the warm blank. Days later you retain a strong, confident, entirely accurate sense that this person was worth your time, sitting on top of an episodic record that has quietly emptied out. The feeling is intact. The content is gone. And because the feeling is intact, you never notice the content left.

Here's the part that makes it a trap rather than an inconvenience. Warmth is the reason you didn't capture anything in the first place. Nobody takes notes on a conversation they're enjoying. Taking notes is what you do at the booth where a stranger hands you a card and you feel nothing, so you scan it, because scanning is what you do with things you know you'll forget. Certainty that you'll remember is produced by exactly the conversations where the stakes of forgetting are highest.

By Friday you're warm and blank. Warm and blank writes a template.

Why the usual diagnosis misses it

The standard read on bad follow-ups is a volume problem. You met 200 people, you're behind, the pile got away from you, buy something that gets through the pile faster. Met's own earlier piece on [the 72-hour memory cliff](/articles/72-hour-memory-cliff/) covers that decay curve, and it's real.

This is a different failure, and it sits somewhere else in the stack. Volume failure is bottom-of-list: the marginal contacts you were never going to work anyway. The warm blank is top-of-list. It hits the five or six people the trip was actually for, and it hits them precisely because you sorted them as unforgettable and therefore skipped the step that would have preserved them.

Which is why a faster scanner doesn't touch it. The category has organized itself around the card: recognition accuracy, batch capture, clean export into a CRM. All of that assumes the valuable object is the contact record, and that the conversation is metadata you'll fill in later. Met has argued before that [OCR is a feature, not the product](/articles/ocr-is-a-feature-not-the-product/). The warm blank is the sharpest version of that argument, because the contacts most likely to have a perfectly parsed name, title, and email are the same contacts most likely to have an empty notes field.

What memory research actually shows about this

The dissociation between feeling and detail is one of the better-documented findings in memory research, and it does not say what most people assume it says.

The common belief is that emotional events are burned in. Talarico and Rubin tested that directly at Duke, comparing students' memories of September 11 against an ordinary personal event from the same week, then re-testing the same people at intervals over the following months. The emotional memories lost consistency at roughly the same rate as the everyday ones. What stayed elevated was confidence: vividness, sense of reliving, belief in one's own accuracy. Their paper is titled for the finding, "Confidence, not consistency, characterizes flashbulb memories."

That gap is the whole problem in miniature. Emotional salience does not protect the details. It protects your certainty about the details.

The supporting structure comes from dual-process work on recognition, summarized in Yonelinas's review of thirty years of research: familiarity and recollection are separable. Familiarity is fast, strength-like, and durable. It's what tells you this person is someone you know and like. Recollection is what brings back context, the specific claim they made, the constraint they were working around, and it's far more fragile. A conference floor is an environment optimized to generate familiarity and hostile to recollection: noise, interruption, forty conversations, no surface to write on.

Same conversation, two emails

Take the Tuesday conversation apart. Three things were actually said, and by Friday each one has degraded differently. Two of them are recoverable with a single sentence typed on the spot. None of them are recoverable from a scanned badge.

The second column is what a good follow-up gets written from. Notice how little was required: not a transcript, not a CRM field structure, just enough of the episode to prove you were present for it. The email on the right is not more polished than the generic version. It's shorter. It's just aimed at a person instead of a category.

Invert the capture rule

If the warm blank is real, the operating rule most people run is backwards. The instinct is to capture defensively: scan the ones you'll forget, trust yourself on the ones you won't. That allocates your capture effort in inverse proportion to how much the contact is worth.

The corrected rule is uncomfortable and short. The better the conversation felt, the faster you write it down.

There's a timing constraint attached. The window isn't that night in the hotel room, because by then you've had eleven more conversations and the interference has already done its work. The window is the ninety seconds after you walk away, while the other person is still in view. One sentence about what they're stuck on. One about why you're relevant to it. That's the entire intervention, and it's the difference between a follow-up that continues a conversation and one that announces a conversation happened.

The cost of being wrong here is asymmetric in an ugly way. Nobody replies to tell you that your email read like you'd forgotten them. They just don't reply, and you file it under bad luck or bad timing.

What a capture tool has to get right

Ninety seconds of note-taking sounds trivial until you try to do it on a conference floor with a phone in one hand and a coffee in the other. That's the actual design problem, and it's why the category keeps solving the wrong half of it. Parsing a card is a solved problem. Getting a human to type one honest sentence about a person while the room is still loud is not.

Met is built for that ninety seconds. Event Mode keeps the capture surface open across a run of conversations so you're not reopening an app between every handshake, and a note goes onto the person while you're still standing there, by voice or by thumb, without a form to fill out or a field structure to satisfy. The follow-up draft is built from what you said about them, not from what their card said about their employer. Contacts live in your iCloud, not on a server we can mine, which matters more than usual when the notes contain things people told you in confidence at a bar.

No notifications nagging you to be diligent, either. Met's position on interruption is that the last thing a person mid-conversation needs is an app asking for attention. The tool's job is to be fast enough that using it doesn't cost you the moment you're trying to preserve.