How to Run Tax Season Without Losing Track of Clients

Every UK accounting firm runs the same nine-week sprint from November to 31 January, and most run it the same way: a spreadsheet, an inbox, and one manager holding the whole picture in their head. It works — until that manager takes a week off in December and three non-responders go quiet without anyone noticing.

The problem isn't effort. It's that tax season isn't actually one workflow. It's five decisions, repeated for every client, that interact with each other in ways a flat checklist can't represent.

The five decisions that run tax season

1. Client type. Individual, sole trader, partnership, director, or landlord — each needs a different document request at first contact. Send the wrong list and you've already lost a week.

2. Complexity tier. Simple, standard, or complex, screened at intake using a handful of flags: property disposals, multiple income sources, foreign income, capital gains beyond shares, first year in this client type. This decides who prepares the return and how deep the review needs to go.

3. Information status. Complete, partial, or non-responder — and this one doesn't stay fixed. A client who was Complete in November can go quiet by December. Status needs re-checking every time a client sends something, not just set once at intake.

4. Timing risk. This combines calendar time remaining with information status. A partial client with three weeks left is a different risk profile than a complete client with three weeks left, even though both show the same “3 weeks to deadline” on a calendar.

5. Review outcome. Clean, queries raised, or sign-off pending. A return isn't ready to file until it clears this check, no matter how close the deadline is — filing with unresolved queries to hit a date creates more risk than a short extension conversation.

Why a checklist can't hold this

A checklist is a single linear path: do step one, then step two, then step three. It's excellent once you already know which path a client is on. It has no way to represent a client switching paths mid-season — a client type changing, a complexity flag surfacing late, a reliable client suddenly going silent.

A decision tree can hold all of that, because each of the five points above is modelled as a genuine branch, not a single step. The client's position in the tree is visible at any point in the season, and it updates as circumstances change.

Running it week by week

November (weeks 1–2): bulk document requests go out, matched to client type. Complexity screening starts as documents arrive.

Late November–December (weeks 3–6): the preparation queue fills. Chase sequences run for anyone partial or non-responsive. Timing risk gets reviewed weekly.

December–January (weeks 7–10): this is where firms either stay calm or don't. The capacity-triage queue — everyone flagged Late-filer risk — needs checking daily, not weekly, from this point on. Review backlog becomes the priority over starting new preparation.

Final week: only sign-off-pending and late-filer-risk clients should still be actively worked. Extension conversations happen here because they were flagged early, not because the deadline forced the conversation.

What breaks without this

Three patterns account for most of the chaos firms report every January:

• A client who looked simple turns out to have sold a rental property, and nobody re-screens them for complexity until the return is already half-prepared by a junior.
• A non-responder goes quiet with no defined chase sequence, so nobody escalates until the deadline is genuinely close.
• The one person tracking “who's at risk” is unavailable for a stretch of December, and the whole tracking system goes dark with them.

All three are fixable with the same underlying change: make the five-branch logic something the whole team can see and act on, not something held in one person's spreadsheet and memory.

Where this goes next

Building this out properly — full chase scripts, a capacity model, exception handling for the cases that don't fit neatly — is more than one article can cover. The Tax Season Operations Playbook walks through the complete system, including a validated decision-tree file that imports directly into PathPilot if you want the whole thing running as a shared canvas rather than a document everyone reads once and forgets.

If you just want to try the triage logic this week, the free checklist covers the first two branches — client type and complexity — in a single page.