The most expensive process nobody measures
Ask an accounting firm where their administrative burden sits and they will say document processing. Ask them to write down the hours spent onboarding a new client and most cannot answer. That gap is why onboarding stays manual: it is painful but invisible, spread across partners, managers, and administrators rather than concentrated in one role.
In our experience onboarding is the single highest-return automation in a mid-sized firm, for a simple reason. It is high-volume in exactly the moments when the firm is busiest, it is almost entirely rules-based, and it involves a lot of correspondence that clients judge you on. Automating document filing saves paper. Automating onboarding saves the client relationship at its most fragile point.
Where the first thirty days actually go
When you map it honestly, the time goes into five things:
Collection. Chasing ID documents, prior-year returns, bank details, entity structures, and engagement letters. Usually across several rounds of email, because the initial request never lists everything.
Verification. Confirming the client is who they say they are and that the structure matches what was described. Necessary, compliance-driven, repetitive.
Setup. Creating the client in practice management software, the ledger, the document management system, and any portals. Often with re-keying because the systems do not share data.
Engagement documentation. Letters of engagement, scope definitions, and fee agreements, frequently personalised by hand from a template.
Communication. Explaining what happens next, what is needed when, and answering the same first-week questions every client asks.
Only the second item requires judgement. The other four are process, and process is what automation handles.
The trap: automating in the wrong order
Firms typically try to automate setup first, because it is the most visible and the most annoying. That is the wrong starting point. Setup depends on data that arrives during collection, so automating the destination before the source means the same manual effort simply moves.
The right order is collection, then chasing, then setup. Get documents arriving in a consistent form with a consistent naming convention before you automate what happens to them.
What a working onboarding workflow looks like
A client accepts the engagement. The system generates a complete document request rather than a partial one, sends it, and tracks what has arrived against what is required. Missing items trigger reminders automatically rather than when someone remembers. Received documents are classified, filed to the correct client and entity, and logged against the compliance checklist.
The administrator's role shifts from chasing and filing to handling exceptions — the client with an unusual structure, the document that does not look right. Setup can then be automated because the inputs are finally consistent.
The measurable outcome is not "hours saved". It is days-to-complete for onboarding, the number of reminder emails sent, and the number of onboarding steps that required manual intervention. Those are numbers you can put in front of a partner group.
Measuring it properly
Before you change anything, record for ten new clients: days from acceptance to complete documentation, total staff hours involved, and who spent them. That is your baseline. Without it you will have an argument rather than a result at the end.
Then measure the same three things after. In firms we have worked with, the first metric typically improves by more than half — not because work happens faster, but because nothing waits for someone to remember it.
Where AI is actually needed
Worth being straight about this. Much of the value above comes from ordinary automation — sequencing, reminders, and structured filing. AI earns its cost in classification of documents that vary in format, in reading correspondence to detect what a client is actually asking, and in flagging the exception rather than processing the routine.
If your client base is uniform and your documents are consistent, a simpler workflow tool may deliver most of the benefit for less. That is worth knowing before anyone sells you a platform.
What to do next
Time ten recent onboardings. If the total is under five hours across all ten, the case is weak and you should automate document processing instead. If it is twenty or more, onboarding is your highest-return project and you should start with collection.
Our AI for accounting firms page covers how we scope this work, document management for accounting firms covers the related process, and the readiness assessment will tell you whether your foundations support it. Our implementation approach explains what delivery involves.
Book a free audit and we will map your onboarding process and tell you honestly which single change would return the most.