A lead going cold is usually described as a timing problem. You followed up too slowly, the budget moved, a competitor got there first. Sometimes that is exactly what happened.
But there is a quieter failure that happens well before the follow-up, and most teams never diagnose it, because the evidence of it looks like success. You have the card. You have a valid email address. Your CRM shows a new contact. And you have no idea what to say to them.
Data survives. Context does not.
A business card is a remarkably durable object. The name, the company, the phone number: all of it is still perfectly accurate three months later. Nothing about that information decays.
What decays is everything that made the person worth contacting in the first place. That they were evaluating three vendors that quarter. That the actual decision sat with their colleague. That they were unhappy with their current setup for a reason they described in some detail. That they asked you to come back after a particular date because of a contract renewal.
None of that is on the card. It was in your head, and its shelf life was about a day.
The contact details are the least valuable part of the exchange, and the only part that gets stored.
So when you finally sit down to follow up, you are working from a name and a company. The email you write is therefore generic, because generic is all the surviving information can support. And a generic email to somebody who described a specific problem reads as though you were not listening, which, functionally, is what happened.
What the context actually contains
It is worth being concrete about what gets lost, because “context” sounds soft and optional. It is not. It is the difference between a record you can act on and a record you can only file.
The record on the left is what a card scanner alone produces. It is accurate and it is nearly useless, because every follow-up written from it will be interchangeable with every other one.
The record on the right contains a role, a timeline, a next action and a second stakeholder. Not one of those four things appears anywhere on the card. All four came from ten seconds of talking to yourself after the conversation ended.
Why the obvious fixes do not hold
Every team knows they should write notes. Almost none do it consistently, and the reasons are structural rather than a matter of discipline. This is worth understanding, because if you diagnose it as laziness you will try to fix it with reminders, and reminders will not work.
- Typing competes with talking, and loses. The correct moment to record context is immediately after a conversation ends, which is also the exact moment the next person walks up. Anything needing both hands and thirty seconds of attention loses that competition every time.
- Batching does not work. “I will write them all up tonight” assumes the memories are still separable at nine in the evening after eighty conversations. They are not. Somewhere around the twentieth person, details begin attaching themselves to the wrong faces, which is worse than having no note at all.
- Writing forces premature structure. To type a note you first have to decide what the note says and which field it belongs in. Speaking does not. You can be vague, hedge, correct yourself mid-sentence and trail off, and still preserve the useful part.
- The CRM notes field is the wrong shape. It exists, it is free-text, and it is three screens away from where you are standing. A field you can only reach from a desk is not a capture mechanism for something happening in a hallway.
The habit that survives a busy day
Say it out loud, immediately, in whatever form comes naturally. Ten seconds, into whatever device is already in your hand.
“That was the operations lead from the logistics company, they are replacing their current system in Q1, she said to follow up after the fifteenth, and her colleague Ravi actually signs off on it.”
That sentence is unstructured, slightly rambling, and worth more than any form you could have filled in. Two properties make it work where written notes fail. It takes about as long as saying goodbye, so it does not compete with anything. And it requires no decisions, because you do not have to know which field the information belongs in. You only have to say it before you forget it.
The mechanics matter less than the timing. A voice memo app works. A message to yourself works. What does not work is any version of this that happens more than a minute after the conversation ended, because by then you are already reconstructing.
If you want it attached to the contact automatically rather than sitting in a separate voice memos app that nobody ever opens again, that is the specific thing Karo is built around: the WhatsApp capture flow takes the card photo and the voice note in the same ten seconds and puts the transcript on the same record.
What a good voice note sounds like
Since the whole practice rests on ten seconds of speech, it is worth knowing what separates a useful recording from a useless one. The difference is not eloquence.
Too thin: “Good chat, seemed interested, follow up.” This is a note in form only. Every word of it would apply to any of the other eighty conversations, which means it distinguishes nothing and helps nobody.
Better: “Logistics company, unhappy with their current setup, wants pricing.” Now there is something to write about. It is still missing the two details that decide whether this goes anywhere: when, and who signs.
Genuinely useful: “Operations lead at a mid-size logistics firm, they are replacing their current system in Q1 because it cannot handle their volume, she asked me to come back after the fifteenth, and her colleague Ravi actually approves the spend.”
The third one is barely longer to say than the second. What makes it work is that it answers the only question that matters later: what would I need to know to write this person a useful email? Four things tend to cover it. Who they actually are beyond their title, what specific problem they described, anything time-bound, and who else is involved in the decision. If you catch those, the recording has done its job even if the grammar is a mess.
The objections, honestly
“I will remember the important ones.” You will remember roughly the top three, and you will remember them less precisely than you think. The problem is not the top three. It is leads four through forty, which are collectively worth far more than the top three and which are exactly the ones that blur.
“We use the organiser’s badge scanner.” Badge scanners solve data capture, not context capture. They hand you whatever the attendee typed into a registration form months ago, with no record of what any of them said to you. They are worth using, they do not address this problem at all, and the trade show guide covers how the two fit together.
“My reps write good notes.” Some do. Check the timestamps. Notes written the same hour tend to be specific and useful. Notes written days later tend to say things like “interested, follow up”, which is a note in form only.
“This is only a problem at big shows.” It is worse at big shows, but it applies to any event where you have more than about a dozen conversations in a day. That includes most conferences, and it includes the evening drinks afterwards, which is where a surprising amount of the useful detail gets shared. The booth checklist covers the operational side of this in more depth.
Where this breaks down
Voice capture is not universally better, and it is worth naming the cases where it struggles rather than pretending otherwise.
Loud halls. A large expo floor at peak time is genuinely difficult audio. Turning slightly away from the aisle and holding the phone closer helps more than speaking louder does. If the hall is truly unworkable, a short typed note beats a recording nobody can make out.
Mixed languages. If your note switches between two languages mid-sentence, which is completely normal in a lot of markets, transcription quality drops. The audio is still kept, so nothing is lost, but you may end up listening rather than reading. Teams working bilingual events tend to hit this most, which we go into on the bilingual capture page.
Very long notes. A two minute recording defeats the purpose. If you find yourself narrating the entire conversation, you have moved from capturing context to writing a report, and it will not survive as a habit. Ten to fifteen seconds is the sustainable range.
Anything confidential. If somebody tells you something in confidence, a transcript that lands in a shared dashboard is the wrong place for it. Use judgement about what goes on the record at all.
The one-week test
You do not need a process review to find out whether this is happening to you. You need one minute and one lead.
Open a lead from an event you worked last month. Try to write the follow-up email without guessing anything.
If you can, your capture is working, whatever tool produced it, and you should keep doing exactly what you are doing. If you cannot, if you are looking at a name and a company and reaching for a template, then the capture step lost the only part that mattered, and no amount of follow-up speed will recover it.
Do this with five leads rather than one and you will have a better read on your event process than any dashboard will give you.
What this actually costs
It is worth putting this next to what a stand costs. Exhibiting at a mid-sized trade show involves the space, the build, the travel, the accommodation and several people away from their normal work for three days. Against that, the marginal cost of capturing context is ten seconds per conversation.
The asymmetry is the argument. You are not deciding whether context capture is worth ten seconds. You are deciding whether the entire cost of the show should produce records you can act on or records you can only count. Teams routinely spend a great deal on getting the conversations and then nothing at all on preserving why they were worth having.
The one thing to take away
Say why they mattered, out loud, before the next conversation starts. That is the whole practice. The tool you use to record it is a detail, and any tool that adds friction to those ten seconds is the wrong tool regardless of what else it does.
Karo exists because we think the voice note is not a nice extra alongside the card, it is the part that decays, so it should be captured in the same motion and land on the same record. Twenty free scans is enough to test that on one real event.