What to track for each company you advise

The file you already keep, badly

Every advisor has a mental file per company. What they do, who the founder is, roughly how it is going. It gets refreshed after each call and it decays quietly between them.

The problem is not that the file is thin. It is that it holds the wrong things. What a company does is the fact you are least likely to forget and the least likely to ever produce an introduction. Meanwhile the thing you needed on Tuesday, which of your six companies the person you just met is relevant to, is not written down anywhere.

Three things are worth keeping. The rest you can reconstruct from your inbox in about a minute.

The current need, with a date on it

One sentence. "Looking for a head of sales who has sold to mid-market hospitals." "Needs a design partner in logistics before the raise." "Hiring nobody until December."

Two details make this work and both get skipped.

It has to be current. A company's constraint changes every few weeks. A need recorded in March is not merely stale by June, it is misleading, because you will act on it. This is the one field that genuinely decays, which is why the date matters as much as the sentence: you need to be able to see its age and distrust it accordingly.

It has to be specific enough to match against. "Looking for engineers" cannot be matched. "A backend engineer who has run Postgres at scale and wants to be employee number four" can. The test is whether reading it would make a specific name occur to you, because that is the only thing you will ever use it for.

If you update nothing else after a call, update this line. It is worth more than the rest of the file combined.

Who you have already introduced them to

A list of names and roughly when.

This solves two problems, one obvious and one that quietly costs you. The obvious one is not introducing the same person twice, which tells a founder you are not paying attention at exactly the moment you were trying to demonstrate the opposite.

The quieter one is that you forget your successes. An introduction you made in April that turned into a hire in June is the single most useful thing you know about your own judgement, and if it is not written down you will not remember whose recommendations landed. Over a couple of years that record tells you which parts of your network are actually strong, which is not the same as which parts feel impressive.

The trigger

This is the one almost nobody keeps, and it comes from Derek Coburn in Networking Is Not Working.

Coburn's insight is about profiling. He found that asking people to remember what he did produced nothing, because it required them to spot an abstract category in the wild. So he worked out the observable signals instead: the sentences people say, and the events that happen to them, when they are about to need what he offers.

Turn that on your own portfolio. For each company, write down what you would have to hear, in an unrelated conversation, to know a match had just walked past.

For a company hiring a first salesperson, the trigger is someone saying they are leaving a startup where they were the first commercial hire. For a company that needs a design partner, it is someone complaining about the exact problem that company solves. For one raising in six months, it is an investor mentioning they are looking at that sector.

The reason to write it is that you will not be thinking about that company when the trigger appears. You will be at dinner, talking about something else, and the whole value of having written it is that the sentence has been pre-loaded.

What not to copy by hand

The market, the round, the headcount, the product roadmap. Not because they are unimportant but because they are recoverable. They are in the deck, the investor update, and the last four emails, and copying them into a note is work that produces a tidy record and no introductions.

Meeting transcripts are a different case and worth being clear about, because the instinct to skip them is wrong. Capture everything. A recorded call holds detail you did not know was important at the time, including the sentence a founder said in passing that turns out to be the need, and the tooling that does this automatically has become good enough that there is no reason to take notes by hand instead.

The mistake is treating the archive as the working file. A transcript answers "what did we talk about in March". The three fields above answer "who should this founder meet", which is the question you have on a Tuesday, and no amount of scrolling a transcript answers it quickly. So keep both, and understand what each is for: the transcript is the raw material, and the current need is what you pull out of it while the call is still fresh.

Where it lives

For the first two or three companies, a note each is genuinely fine.

It starts to fail around the fourth or fifth, and the failure has a specific shape. You do not forget the companies. You forget which of them the person in front of you is relevant to. That is a matching problem, not a storage problem, and notes are bad at it because matching requires holding all of them in mind at once, which is exactly what stopped working.

NetworkOS is built around that shape. Each company you advise carries its current needs, and when you describe someone, or a name comes up, it searches your network against those needs rather than making you do the crossing yourself. It also keeps the record of who you have already introduced, so a pair you have connected before does not resurface.

It will also do the extraction described above. Meeting notes that land in your inbox from the common note-takers get read for the asks inside them, and those become introduction suggestions without you writing the need down at all. The limits are worth knowing: it only sees notes that actually arrive by email from a tool it recognises, so a call you scribbled on paper is invisible to it, and what it proposes is a suggestion you confirm rather than something it acts on alone.

The wider system this sits in is the CRM for startup advisors, and how to hold the whole portfolio at different depths is covered in running a portfolio of advisory relationships.

The short version

Keep three things per company. The current need, one sentence, dated, and specific enough that reading it makes a name occur to you. The people you have already introduced, so you never send the same person twice and so you learn whose recommendations land. And the trigger, the thing you would have to overhear to know a match just walked past, written down precisely because you will not be thinking about that company when it happens. Skip copying the market, the round and the headcount, which are all recoverable. Record the call itself, ideally automatically, but treat it as raw material rather than the working file. The need is the only field that decays, and it is the one the work turns on.