Advisory firm operations: a guide for firms between $75M and $500M

Operations is the least discussed function at an advisory firm and the one most likely to be capping it. Compliance gets outsourced, finance gets outsourced, and operations stays with whoever has been carrying it, usually after hours.

This is a working guide to what operations means at a firm of one to fifteen people, what tends to break, and how to work out which part is costing you money. It is written for principals rather than for operations staff, because the decisions in it are commercial decisions.

What operations covers

At this size, operations is not a department. It is four distinct layers that most firms have never separated, which is why the problem feels vague when you try to describe it:

• **Workflow.** The recurring processes: onboarding, annual reviews, RMDs, distributions, account maintenance. What the steps are, who owns each one, and what happens when one fails.
• **Data.** Which system holds the authoritative version of a client fact, and how it gets to the other systems that need it.
• **Documents.** What a file is called, where it lives, how long it is kept, and whether somebody who did not file it can find it.
• **Knowledge.** What is written down versus what lives in somebody’s head, and what that means when they are on holiday or gone.

Most firms have bought good software for the first three and have done nothing deliberate about the fourth. That is the usual shape of the problem.

What breaks first, and in what order

Operational failure at this size is predictable. It arrives in roughly the same sequence at almost every firm:

Around five people, the principal is still the escalation point for routine questions, because they are the only person who knows every answer. It feels like being busy rather than like a structural problem. A firm being stood up from nothing gets one chance to avoid arriving here at all.

Around eight, coordination that used to happen in the hallway stops happening. Work starts getting dropped between people rather than by them, and nobody can say how many things are open. An acquisition accelerates this sharply, because two firms arrive with two ways of doing everything.

Around twelve, the firm hires an operations person, and the process moves out of the principal’s head into theirs. The principal’s week improves and the firm’s dependency does not. We wrote about that trade in hire an operations manager or outsource it.

None of these is a software problem. Each one is what happens when a process that was never designed is asked to carry more volume than the person who invented it can hold.

How to find out what it is costing you

The instinct is to ask the team where the time goes. Do not, at least not first. People describe the work they remember rather than the work they repeat, and the answer will point at whatever went wrong most recently rather than at whatever costs the most.

Take one recurring process and follow a single live case end to end. Onboarding is usually clearest because it touches every system you own. Write down:

• Every system it touches, in order
• Every point where a person types something a machine already knows
• Every point where it waits on somebody, and how long
• Every decision made without looking anything up, which is undocumented process
• What happens on the exception path, when a form is rejected or a custodian asks for something unusual

Do that for three processes and you will have a defensible picture of the operation. You will also usually find that the delays are not inside any one product but in the spaces between them, which is the argument in why another platform will not fix your operations.

The four decisions that do most of the work

Whatever the diagnosis, the fix tends to reduce to a short list. These are not purchases. They are decisions the firm has to make, and no vendor can make them for you:

• Which system owns which record, so there is one authoritative copy and the others read from it
• What triggers what between systems, and who owns each automation by name
• One naming convention and one filing structure, with retention applied
• What is written down, and where, so the procedure sits inside the system where the work happens. How to write ones that get used

The detail of the second one is in why your tech stack does not talk to itself. The fourth is the one firms skip, and it is the one that determines whether any of the others survive a departure.

Build, hire, or migrate

Three options get considered, usually in the wrong order.

**Migrating** is the most expensive and the least often correct. If the delays are between products rather than inside one, a new platform reproduces the problem with a different logo. The real numbers are in what a CRM migration actually costs.

**Hiring** is usually right eventually and usually wrong first. A role filled before anyone has written down what it covers relocates the dependency rather than removing it.

**Building** means designing the process, configuring what you already license to match it, connecting the systems, and writing it down. It is the least glamorous option and the one that changes the marginal cost of the next household, which is what capacity actually turns on.

What good looks like

A firm with its operations in order does not feel dramatically different day to day. The tells are quieter than people expect:

• A new hire learns the job from documentation rather than from whoever has time that week
• Anybody can say how many onboardings are open and which one is stuck
• A client fact entered once appears everywhere it is needed, without anybody retyping it
• A week of somebody’s absence is inconvenient rather than disruptive
• When compliance asks how something happened, the system holds the trail

That last one matters more than it sounds, and it is the whole of preparing your operations for an examination. The same documentation is also what a buyer reads, which is covered in what a buyer is actually pricing.

Where to start

Start with measurement rather than with a solution. Almost every expensive operations decision at this size gets made from an anecdote, and the cheapest thing you can do is replace the anecdote with a number taken out of your own systems. If you are choosing when, the fourth quarter is the window most firms actually have.