The three questions no professional services firm can answer quickly

There is a reliable test for whether a firm's institutional memory works. Pick a live client account. Ask three questions, and time the answers.

None of them are exotic. Every one is something the firm demonstrably knows, in the sense that the information exists inside the organisation, held by competent people who would answer correctly if asked directly. And in most firms above about fifty people, none of them can be answered in under a day.

Question one: who has worked with this person?

A partner is preparing for a meeting with a client-side director they have never met. What they want to know is ordinary. Has anyone here dealt with them before? How did it go? What do they care about? Is there anything I should not walk into?

What actually happens is that they check the CRM, find a contact record with a job title and a phone number, ask two colleagues who do not know, and go into the meeting cold.

Meanwhile, two colleagues who worked with that director on a programme in 2022 sit three desks away, entirely unaware the meeting is happening. The relationship history exists. It is in old email that nobody will ever reread, in the heads of people nobody thought to ask.

The firm has the relationship. The partner in the meeting does not, and has no route to it.

The cost is not abstract. A first meeting spent establishing context is a meeting not spent on the thing you came for, and the client experiences it exactly as it is: being asked questions they have already answered, by a firm that has been paid for a decade.

Question two: have we done this before?

A team is scoping a piece of work. It looks familiar. Somebody says, surely we have done this before.

The answer is almost always yes. Finding it is the problem. In most firms, the procedure is a broadcast email to a practice group, or several, followed by a day of waiting. The replies that come back are from the helpful rather than the relevant. The person who actually ran the closest engagement is on a client site, does not read the mailing list, or has left.

So the team rebuilds what the firm already owns. Not out of laziness, out of a rational calculation: three days of rebuilding is more predictable than an unbounded search with no guarantee of a result.

Two things are lost in that transaction, and the second is worse. The firm pays for the same thinking twice, which is a margin problem. And the real expert stays invisible, which is a compounding problem: they do not get asked, so their reputation does not grow, so they get asked even less. The firm's best resource on a subject can go an entire career without the firm knowing.

Question three: why didn't we sell that?

A long standing client procures a service from another firm. It is a service you deliver, competently, for two other clients in the same sector.

Nobody made a decision not to pitch. There was no meeting where the opportunity was assessed and declined. The gap between what the client needed and what the firm had sold them was simply never visible to anyone at one time. The partner who knew the client did not know the firm had that capability. The partner who had the capability did not know the client existed.

This is the most expensive of the three, because it is invisible even after it happens. The other two announce themselves: someone walks into a cold meeting, someone rebuilds an analysis. This one is silent. Revenue that was available and never appeared leaves no trace in any system, which is why firms can lose it for years without it ever reaching a management report.

They are one question

The three look like different problems and are usually owned by different people. Relationship history is business development. Prior work is knowledge management. Whitespace is account planning. Three functions, three budgets, three sets of tooling, no shared diagnosis.

But look at what each one actually requires:

| Question | What answering it needs | |---|---| | Who has worked with this person? | A link from a client-side individual to the colleagues who have engaged them, with evidence | | Have we done this before? | A link from a type of work to the colleagues who have delivered it, with evidence | | Why didn't we sell that? | Both of the above, held against each other, with the gaps visible |

Every one of them is a connection between a person and something they did, backed by proof. That is one data structure. The reason firms cannot answer any of the three is that they have never built it.

They have a CRM, which holds contacts and no capability. They have a document store, which holds artefacts and no people. They have a resourcing system, which holds availability and no history. Each is internally consistent and none of them can answer a question that crosses the boundary, which is all three of these.

Why it gets worse rather than better

Firms usually assume this improves with scale, because a bigger firm has more of everything. The opposite is true, and the reason is structural.

The informal mechanism that answers these questions is a person who has been there a long time and knows everybody. That mechanism scales with the number of pairs of colleagues who have worked together. Headcount grows linearly; shared history does not keep up. At twenty people everybody has worked with everybody. At two hundred, most pairs never have, so most of the connections that would answer these questions were never formed.

Add average knowledge-worker tenure of about four years, and the up-or-out pyramid that guarantees churn, and the informal index is not just failing to keep pace. It is being deleted continuously.

Firms do not outgrow this problem. They grow into it.

What answering them requires

Not a repository, and not a better CRM. Three properties, which fall out of the questions themselves:

It has to assemble itself. Any system whose answers depend on fee earners maintaining records will be empty about the engagements that mattered most, because those were the busiest quarters.

It has to link people to evidence. The answer to all three questions is a person plus the proof. A system that returns a document has not finished the job, and one that returns an unsupported assertion is worse than nothing, because it will be checked once, found wanting, and abandoned.

It has to show gaps as gaps. Question three only becomes answerable if absence is represented. A record that only contains what the firm has done cannot show what it has not.

That is the specification OrgAtlas was built against, and it came from these three questions rather than from a feature list.

Next: the first question in detail, and what a cold stakeholder meeting actually costs