The difference between recording work and understanding it.

Every company has systems that record work.

Very few have systems that understand it.

This distinction sounds abstract until you try to answer a simple question that every People Team faces at some point: who in the company would be the right person to lead this project?

Not formally -- not who is the right person according to the org chart. Actually -- who has the skills, the relationships, the credibility and the bandwidth to make this particular thing succeed?

Your payroll system doesn't know. Your HRIS doesn't know. Your performance management tool knows who scored highest in last year's review cycle, which is a different question entirely.

The answer to "who is actually the right person for this?" lives in the heads of the people who have worked with them. Distributed across dozens of conversations, Slack messages, project retrospectives and institutional memories that exist nowhere in formal systems.

This is the recording vs. understanding gap.

What recording systems know

Recording systems know facts. They are very good at what they do.

Your HRIS knows who works at the company, when they started, what their job title is and what they're paid. It knows when someone was promoted and when someone left.

Your project management tool knows what tasks were assigned to whom and whether they were completed on time.

Your performance management tool knows what was said in the annual review and what the outcome was.

These are valuable facts. But they are not the same as understanding.

What understanding requires

Understanding requires something different: the ability to interpret signals from how people actually work, not just what is formally recorded about them.

When a colleague sends a congratulations message to someone on a different team, that is a signal -- a weak one, but a real one -- about a relationship that exists between them. When a manager approves a document request on behalf of a direct report, that is a structural signal confirming a reporting relationship. When someone presents a project at a company demo day, that is a contribution signal about their expertise in the subject matter of that project.

None of these signals appear in a recording system. They are the exhaust of real work -- the residue that real collaboration leaves behind.

Understanding organisations requires the ability to collect those signals systematically, weight them by reliability, and draw inferences from the patterns they create.

Why this is hard

The challenge is not that signals are unavailable. They are everywhere -- in every message, every approval, every recognition, every project.

The challenge is that no single signal is reliable on its own.

A Moonboard message between two colleagues might reflect a close working relationship. Or it might reflect that they happened to work together once and the sender was being kind. You cannot tell from the signal alone.

This is why intelligence systems that rely on single signals produce unreliable results. The LinkedIn endorsement problem is the classic example: skills endorsed by one person, once, at some point in the past, with no mechanism for decay or validation. The signal exists, but its reliability is so low that it is nearly useless.

Reliable organisational understanding requires multiple independent signals, accumulated over time, weighted by source and recency, and subject to decay when they are not reinforced.

That is a different kind of system from a recording system. It is a confidence engine.

What changes when you can understand your organisation

When a company has a genuine understanding of how its people work -- not just who they are and what they're paid -- the quality of decisions that depend on people improves substantially.

Succession planning that relies on who the CEO already knows is replaced by succession planning that relies on who the data shows is ready. Mentorship matching that relies on the HR team's memory is replaced by mentorship matching that accounts for skill overlap, relationship distance and learning objectives. Talent risk that only surfaces when someone hands in their notice is surfaced earlier, when the signals of disengagement are detectable but not yet decisive.

These are not futuristic possibilities. They are practical outcomes of having better data about how people actually work -- rather than just better data about who they are.

The graph is not a database

The organisational graph is not a new database for storing HR facts. It is a representation of the living, dynamic reality of how a company operates -- who works with whom, who is trusted for what, who has demonstrated expertise, who is connected to whom across team and department boundaries.

It is built not from data entry but from the natural residue of real work. It is maintained not by administrators but by the ongoing activity of the organisation.

The difference between recording work and understanding it is the difference between a photograph and a film. A photograph tells you what existed at a moment in time. A film tells you how things moved, changed and related to each other over time.

Understanding organisations requires the film.