What happens to our support staff if the admin goes?

What is actually inside a support role?

Before you can say what happens to it, be precise about what it contains. In most firms, a support role is usually six or seven separate jobs bundled into one person because each of them was classed as not fee earning. A legal secretary types up dictation, formats documents, opens files, chases identity evidence and takes the call from the client who cannot reach their fee earner. A practice administrator in a surveying firm chases timesheets, assembles the monthly fee report, books inspections and keeps the template library from falling apart.

Nobody designed that bundle. It accreted, one task at a time, usually at the moment a partner needed something off their desk. That matters, because when you automate part of it you are not removing a role. You are removing three or four of the tasks inside it and leaving the rest standing.

Which parts does automation actually remove?

The parts that are mechanical and repeated. Moving the same information between a case system, a spreadsheet and a document. Reformatting a report into house style. Assembling a bill from time entries and a narrative. Producing the same pack every month in the same shape. Those are the jobs behind processes like invoice preparation, timesheet capture, client onboarding and email triage.

What stays is everything that needed a person in the first place. Ringing the client who has ignored three polite reminders. Knowing that one particular client still wants hard copies and will complain if they do not arrive. Judging which of forty exceptions this week is the one a partner needs to hear about today. Keeping a fee earner to a deadline they have already slipped twice. None of that is admin in the sense people mean when they say the admin is going. It is coordination, and firms are usually short of it.

Why does the remaining work get harder?

Because the repetitive tasks were the easy part of the day. They were also the buffer. When a support person had forty minutes of formatting ahead of them, that was forty minutes of low-concentration work they could do while fielding interruptions. Take it away and what is left is a shift made entirely of judgement calls and awkward conversations.

There is also new work. A rebuilt process needs someone watching it: spotting when output starts drifting from what the firm wants, noticing when a template has been quietly edited, flagging the case the system handled wrongly. That ownership has to sit with a named person, as we argue in the process nobody owns. In most firms the obvious candidate is the administrator who ran the manual version, because they are the only person who knows what good output looks like.

What should the freed hours absorb?

This is the decision firms skip, and it is the whole question. Before the build starts, name the work the recovered capacity will go into. Realistic candidates in most firms:

• Billing hygiene: chasing narratives, getting bills out on time, shortening the gap between work done and cash in, the problem described in the bill nobody can raise.
• Bid coordination: owning the schedule, the evidence pack and the submission checks on proposal and bid production, which partners currently do badly at eleven at night.
• Template and knowledge upkeep, so the firm stops paying the tax set out in why templates rot.
• Compliance evidence gathered as work happens rather than reassembled before an audit, which is the point of compliance evidence.
• Client contact: proactive updates, the calls that stop a complaint forming, the follow-ups nobody has time for.

Pick one or two and write them into the role. A vague promise that people will do more valuable work is not a plan, and everyone can tell.

What goes wrong if nobody decides?

The question answers itself, badly. The commonest outcome is that the freed hours get swallowed by noise within a month and the firm concludes the rebuild achieved nothing, because no one defined what the capacity was for. The second commonest is drift: a vacancy is not replaced, then another, and two years later the support function is thin, the knowledge has walked out, and nobody made a conscious decision at any point. When knowledge leaves covers what that costs.

The third outcome is the most expensive. The best administrator in the firm, the one who actually understood how the work flowed, gets bored or spooked and resigns. You have kept the headcount and lost the capability.

How should a firm talk about this before the build?

Plainly, and early. Say which tasks are going. Say what the role is expected to contain afterwards. Say whether headcount is on the table, and if it genuinely is not, say so in a way you can stand behind in twelve months. If you do not know yet, say that too, with a date by which you will. People handle uncertainty far better than they handle discovering they were managed around. The job security question goes further into that conversation, and getting a team to adopt AI covers what happens next.

What should a managing partner do first?

Take one support role and break the bundle apart on paper. List every task, how often it happens, and roughly how long it takes in a normal week. Mark each one as mechanical, judgement or contact. The mechanical column is what a rebuild can take. The other two columns are the real job description, and they are usually longer than anyone expected.

Then work out what the recovered hours are worth and where they should go. The capacity calculator gives you the arithmetic, and when the constraint moves explains why the bottleneck rarely stays where you left it. If you want the firm-wide view of where the hours are disappearing first, start with the audit.