Who needs you now.

A profile says who could buy. A buying signal says who needs to buy now. This is what a signal is, how one should be measured, and how we find each of our four on LinkedIn, on X and in company data.

What it is

A buying signal is an observable event that raises the chance a company will buy soon. The word that matters is observable. A signal is something public that can be found and dated: a job post, a funding round, a post a person wrote. An opinion about a market is not a signal.

A signal measures timing, and the profile measures fit. Fit is static: it says an account could buy. Intent is what is changing. Both are needed. A company that fits and shows no intent is a long wait. A company that shows intent and does not fit is a distraction.

How to measure

Measuring a signal takes four things.

The event. It has to be defined exactly enough that two people looking at the same evidence agree on whether it happened.

The source. Every event is visible somewhere specific. Hiring is visible in job posts. Funding is visible in round announcements. A problem is visible in what people write. The source decides how the event is searched.

The window. Intent decays, and a signal is worth most right after the event. So each signal has a number of days after which it no longer counts.

The person. Some events are shown by a person, such as a post, and the person is already known. Others are shown by a company, such as a round, and have to be turned into the people who decide there. An event without a person to contact is not usable.

Then fit is measured separately. Not everyone behind an event is a buyer, so every person found is scored against the profile. A lead needs both: the event and the fit.

Our four

We measure four events, the same four for every company, with the contents written for yours: the people in the buying role, companies that are hiring, companies that raised funding, and people who post about the problem.

An agent writes each one from your profile. For each it writes the situation in one sentence, the pain your product removes in that situation, the window in days, who to contact and with what angle, and the exact inputs of the search. The job titles it uses are the titles in your profile. A hiring or funding event counts for 30 days by default, and a post for 7.

Industry, company size and location are never written into an event. They come from the profile and are applied to every search.

The buying role

The first event is shown by a person: someone holds the position that buys. Its titles are the roles in your profile that sign a purchase or push it through.

The right way to search a role is never by title alone. The same title exists at every kind of company, and a search on the title returns all of them. The title has to be tied to the product's world.

We search LinkedIn profiles. One search runs for each title, and each search pairs the title with a word that a real buyer's own profile would carry and with a country: the tools and the work your product is about, in the short phrases your buyers use. The words and the countries are taken in turn across the searches. Each search returns up to 25 profiles by default. A profile that shows a country or a company headcount outside your profile is dropped.

A second method queries people databases, filtering by title, country, headcount and the same words. In our own runs the two methods returned almost entirely different people. The second runs when the first finds no one new.

Hiring

The second event is shown by a company: it has opened a role that would use your product every day. A company that hires for a problem has decided to spend money on it.

This event has to be measured in two steps, because the person in the job post does not work there yet. First find the companies. Then find the people who decide for the team the new hire will join: the managers, leads, heads and directors over it.

We start with LinkedIn job posts. One search runs for each job title in each country, returning up to 25 posts by default. A post older than the window, or from another country, is dropped. Companies are then ranked by how many matching posts they have, taking the job titles in turn, and the first ten are used by default.

For each company we then search LinkedIn profiles with the company's name and the decision titles. A person is kept only when their current job is at that company, up to three per company.

Company size is measured last, from the people. A job post does not state headcount, and ranking by the number of posts alone puts the largest employers first. So the headcount shown on the profiles of the people found decides it. A company outside your range is skipped, and the next one is searched.

A second method uses company hiring data, where size and country are filters of the search itself.

Funding

The third event is also shown by a company: it has raised a round. The date is what makes it a signal. A round from two years ago says nothing about this quarter.

We search funding-round data on companies. The search asks for companies inside your size range and your markets whose latest round falls inside the window. The type of round can be checked as well, to keep the stages that match the size of company you sell to, such as seed or series A.

Each funded company is then turned into people. We look for the buying roles from your profile at that company. Small companies often have no one with those titles yet, so where none is found we take the founder or the chief executive.

Posts

The fourth event is shown by a person, in their own words: they wrote about the problem you solve. It is searched with short topic phrases of two to four words, never full sentences.

On LinkedIn, each phrase is searched in public posts published inside the window, the last week or the last month. The author of every post found becomes a candidate. Then the comments are read. The posts with the most comments are taken, each phrase in turn, twenty posts by default, and up to fifty commenters are read from each. Every commenter becomes a candidate, together with the text of their comment.

On X, each phrase is searched in posts in the same way. The account that wrote each post becomes a candidate, and the replies to the most answered posts are read like LinkedIn comments.

Writing about a topic does not make someone a buyer. So every author and commenter is compared with your profile, using what the post shows: their name, their headline or bio, the post, and their own comment. Those at 0.5 or more are kept. For the people kept from LinkedIn, the profile is then read, which gives the job and the company that a post does not show.

The score

Fit is measured the same way for every person, whichever event found them.

It is one question, asked about one person at a time: is this someone your company should sell to? The question judges two things together. The role: does the person hold one of the roles in your profile, or the same job under another title. And the company: is it the kind and size of company in your profile, in your markets, and outside the segments you left out.

The answer is one of five described situations. No: neither the role nor the company fits, or the person is a student, a recruiter, an investor, or sells the same kind of product. Weak: one of the two fits a little. Partly: the role fits and the company clearly does not, or the reverse. Mostly: both fit, with one not certain from what is known. Yes: the role and the company are those of your profile.

The position on those five levels becomes a score from 0 to 1. A field that is unknown is judged from the rest. People are scored one per request, because judging them in a batch skewed the answers. A score of 0.6 or more is treated as a fit.