Research
How firms know what they know.
OrgAtlas began with a piece of organisational behaviour rather than a piece of software. Transactive memory is the study of how a group holds knowledge no single member has, by knowing who to ask. Every firm already runs one. It is made of people, and it leaks. These are the working notes: the theory, what it costs when it fails, and what an honest system built on it has to do.
21 articles · written by Daniel Lambert · sources cited on every page
ATransactive memory
The organisational behaviour theory OrgAtlas is built on: how a group knows more than anyone in it, and what happens when the part that does the knowing walks out.
- A1What is a transactive memory system? A field guide for professional services firmsPillarThe organisational behaviour theory behind every firm that knows more than any partner in it, what its three moving parts are, and why it degrades exactly as a firm grows.
- A2The three failures of a firm's transactive memorySpecialisation, credibility and coordination fail separately and need different fixes. Here is what each looks like from inside a firm, and which one is almost always the real problem.
- A3Your firm's memory has a half-life of 4.1 yearsAverage knowledge-worker tenure is 4.1 years and roughly 42% of what someone knows is held by them alone. Run those two numbers together and a firm's institutional memory has a decay curve.
- A4Transactive memory in the age of AI agents: the directory service problemA forty-year-old theory from organisational psychology has become the leading frame for multi-agent AI. The firms that will get value from agents are the ones whose human directory already works.
- A5Why knowledge management failed and transactive memory did notThirty years of knowledge management produced dead wikis and abandoned precedent libraries. The theory it was built on asked the wrong thing of people. The alternative asks almost nothing.
What it costs a firm when its own experience is unreachable, in the three specific moments where the cost lands.
- B1The three questions no professional services firm can answer quicklyPillarWho has worked with this person, have we done this before, and why didn't we sell that. Three ordinary questions, all unanswerable in under a day, all the same underlying failure.
- B2Walking in cold: what it costs when nobody knows who already knows the clientA CRM contact record is not a relationship history. The difference shows up in the first ten minutes of every meeting with a stakeholder the firm has served for years.
- B3The broadcast email problem: what "does anyone know about X?" really costsThe all-staff email is most firms' actual expertise search. It costs a day, interrupts hundreds of people, and systematically fails to reach the person who knows the answer.
- B4Why your best client bought that service from someone elseNobody decided not to pitch. The gap between what the client needed and what your firm had sold them was simply never visible to one person at one time.
Why every system that asks fee earners to record what they know decays, and what has to be true of one that does not.
- C1Why law and accounting firm CRMs fail, and what replaces the data entryPillarLow adoption is not a training problem. A CRM asks fee earners to convert billable time into administrative time, and in a firm that argument is lost before it starts.
- C2Email is already your firm's richest record. It is just unreadableThe document store holds the artefacts. The reasoning, the relationships and the reason things were done a particular way happened in email, where nothing can find them.
- C3Adoption is the only feature that matters in firm-wide softwareA firm-wide system used by 30% of people is not 30% as good as one used by everyone. It is worthless, because nobody can rely on a partial answer. Here is why the maths is so unforgiving.
- C4What a firm should record about an engagement, and what it never doesA checklist of the twelve things that turn out to matter three years later, why the finance system holds none of them, and how to capture them without a closedown form.
Graphs, evidence and provenance: the mechanics of answering with a chart rather than with confident prose.
- D1Knowledge graphs for professional services firms: a plain explanationPillarWhy the questions firms ask about their own work are relationship questions, why relationship questions need a graph, and what that means in practice rather than in architecture diagrams.
- D2GraphRAG versus vector search when the question is "who knows this client?"Vector retrieval is very good at finding passages that resemble a question. Relationship questions do not resemble their answers, which is why firms get fluent nonsense back.
- D3Reading evidence weight: telling a proven capability from a claimed oneEvery capability statement a firm writes says it can do everything. Weighting each claim by the evidence behind it is what makes a capability record able to discriminate at all.
- D4Mapping a client account: a worked exampleWhat actually happens between forwarding a year of correspondence and having an account you can navigate. Walked through on the demo account, step by step, including what it cannot do.
Relationship intelligence, expertise location, whitespace and the weekly brief, assessed honestly, including where the categories fall short.
- E1Relationship intelligence for professional services: what it is and what it missesPillarA fair account of the category that already sits in most large law and accounting firms, what it genuinely solved, and the half of the question it was never designed to answer.
- E2Expertise location systems: finding the expert, not the documentA category that has existed since the nineties, failed twice for the same reason, and now has the one input it always lacked. What an expertise locator has to do to work.
- E3Whitespace analysis for professional services firmsMost whitespace advice is written for software companies selling seats. Firms sell expertise, which changes what a gap is, how you prove it, and why the twice-yearly workshop cannot find them.
- E4The Monday morning account brief: what should be in oneA system nobody opens is worth nothing. The brief is the answer to that, and there are five things it should contain and several it should not. A working template.
→Next step
Or see the theory running.
The homepage carries a working atlas of a demo account, built the way the articles describe: every person, engagement and capability linked, and every fact tied back to the document it came from.