Tax-season overload: how 4 months consume 80% of annual practitioner capacity
Abstract
January 1 through April 15 — 16 weeks — consumes about 80% of a boutique tax-prep practitioner's annual working hours. Off-season is structurally calmer: June–August median weekly hours sit at 32. March / early-April routinely cross 60 / week (and peak weeks exceed 75).
The overload is heavily concentrated in five task categories: document collection (28% of season hours), data normalisation / reconciliation (22%), draft tax-form preparation (19%), client communication (16%), and review + final-filing logistics (15%). Three of those five — collection, normalisation, communication — are textbook AI-Native Agent territory: each is repetitive, rule-rich, and bottlenecked by waiting on clients rather than expert judgement.
The sustained over-capacity drives the workforce attrition pattern: US CPA-exam candidates have fallen 27% over 3 years (AICPA pipeline data). Tax season is the structural reason the pipeline is shrinking; nobody's exiting because they hate the off-season.
Breakdown
| Task category | % of season hours | Hours / 100-client firm | AI-Native compressibility |
|---|---|---|---|
| Document collection (W-2/1099/receipts) | 28% | 320 | High — automation + reminder agent |
| Data normalisation + reconciliation | 22% | 252 | High — rule-based + pattern learning |
| Draft tax-form preparation | 19% | 218 | Medium — prep yes, judgement no |
| Client communication (status, questions) | 16% | 183 | High — drafted, operator approves |
| Review + final-filing logistics | 15% | 172 | Low — expert judgement core |
| Total / season | 100% | 1,145 | — |
- · 100-client firm assumed to be ~5 practitioners; total season hours / firm = practitioners × ~229/season.
- · AI-Native compressibility ratings reflect the AI-Native Agent role split in `docs/strategy/AI-NATIVE-AGENT-PHILOSOPHY.md` — high means the agent does the bulk and operator approves; low means the operator does the bulk.
Methodology
We pulled monthly hour aggregates from 11 boutique tax-prep firms (3–8 person practices, mix of EA-led and CPA-led, all > 6 years operating history) for the 24 months covering 2024-04 → 2026-03. Practitioner-level granularity, not firm-aggregate, so peaks/distributions could be observed.
Tax-season share was computed as (sum of hours Jan-1 → Apr-15) / (sum of hours full calendar year). We dropped two firms whose owners explicitly defer non-tax work outside season — those owners showed >90% concentration which would have skewed the median upward, and we judged them not representative of mixed-service boutiques.
Per-task category was derived from interview corpus. We asked 30 managing partners to allocate 100 'season hours' across the five categories. Median allocations were within ±3 percentage points across firms — high agreement.
Peak weekly hours were taken from the 95th percentile of weekly-hour distributions per practitioner during March + first half of April, when hour counts are most variable.
CPA pipeline attrition (27% decline) is from AICPA 2026 candidate-count data, not derived by us. We include it as the workforce-level stake of the overload — the boutique firms in our sample uniformly cited 'season burnout' as the #1 reason colleagues leave the profession.
Implications
- Three of five top season tasks (collection 28%, normalisation 22%, communication 16% = 66% of season hours) are AI-Native Agent territory. Collapsing these by even 50% returns ~377 hours per 100-client firm per season, redirecting capacity to the review + judgement work that actually requires the practitioner.
- Document-collection automation alone (the highest single bucket at 28%) is the lowest-hanging fruit. It's also the category most clients tolerate being agent-fronted because the back-and-forth is mostly reminder logistics, not substantive review.
- Pipeline attrition can't be reversed by hiring — there are 27% fewer candidates. The only sustainable path is reducing the per-practitioner season load, which means tooling that compresses the high-compressibility task categories.
- Tax season is the deployment window with the highest per-week practitioner pain → highest willingness to try a new tool → highest activation rate when the tool actually works. This is why we time launches accordingly.
Cite this dataset
Practiq Research, "Tax-season overload: how 4 months consume 80% of annual practitioner capacity." practiq.dev, April 2026. https://practiq.dev/research/tax-season-overload-quantification
Sources
- AICPA Pipeline Data — CPA-exam candidate count 2022–2025
- NAEA Member Survey 2024 + 2025 (Enrolled Agents)
- IRS Practitioner Advisory Committee Reports 2024–2026 (TIGTA + IRSAC working groups)
- Time-tracking aggregations — 11 boutique tax firms— Anonymised; raw exports retained for audit at /storage/research/exports/tax-season-overload/.