Too often, firms ask the wrong question first: "Can AI fix busy season?" The right question is really, "Can my pain be addressed with AI?" The difference matters.
AI adoption in accounting jumped from 9% to 41% in a single year per Wolters Kluwer's 2025 Future Ready Accountant survey, yet 56% of executives across industries reported no measurable financial return from AI investments per PwC's 2026 CEO Survey. The shortfall isn't due to AI’s capabilities, but rather it is due to prepurchase diagnosis.
Busy-season pain falls into two categories: mechanical and structural. Mechanical pain is process-driven, such as document handling, data entry, communication loops, classification errors. It scales with volume, and AI excels at it. Structural pain is decision-driven, such as pricing that doesn't reflect scope, clients who don't respect boundaries, judgment calls that require experience. AI cannot solve structural pain, and when it is applied here, AI becomes shelfware in 18 months. Using the five diagnostics below can help you classify your pain points now, so you make decisions in July on which tools offer the right answers.
How long it takes: Two to three hours.
How to do it: Use the methods that your firm already uses to capture time to get the data. Pull timesheets for your highest-billing staff across the peak four weeks of busy season, run a quick survey of the team, or, if you're a one- or two-person shop, reconstruct your own week from your calendar and inbox.
Categorize every hour into five buckets:
Most partners estimate 60% to 70% of busy-season hours go to pure tax preparation. Time audits typically reveal that it is closer to 40% to 45%. The difference is mechanical work that AI can displace. If document handling plus client communication exceed 25% of total hours, you have a mechanical problem with a technological solution. If pure tax preparation dominates but rework is over 15%, run Diagnostic 4 to understand whether those errors come from data extraction or professional judgment.
How long it takes: One to two hours.
How to do it: Randomly sample 50 completed client files. For each, count total documents received, documents that required follow-up requests, documents that required manual data extraction, and documents where you had to call the client for clarification. Calculate three ratios:
Document automation delivers measurable returns when clean intake exceeds 60% and manual extraction stays below 30%. Outside those ranges the technology works, but time savings get consumed by exception handling. A clean intake rate below 60% is a client education problem, not a technology problem.
How long it takes: Two hours.
How to do it: Map your busy-season workflow as a series of handoffs from "client engagement" to "return filed." For each step, identify who does the work, how long it takes, what causes delays, and where client interaction is required.
Most firms find their bottleneck falls into one of these three patterns:
Each pattern points to a different AI category:
Buying the wrong category for your actual bottleneck is the most common AI spend mistake in the profession right now.
How long it takes: One to two hours.
How to do it: Review every amended return and significant correction from the most recent busy season. Categorize each error: extraction (wrong numbers pulled from documents), classification (right numbers, wrong tax treatment or form placement), judgment (correct extraction and classification, wrong strategic decision), or client (incorrect or incomplete information from the client).
Calculate what percentage of total rework hours each represents. If extraction errors exceed 40% of rework time, document AI solutions may deliver immediate ROI. Machines read numbers more consistently than humans under time pressure. If judgment errors dominate, AI may become a drafting assistant, but it doesn't eliminate the rework.
Firms at which classification errors exceed extraction errors typically have a training issue, not a technology issue. Staff understand the documents, but they struggle with tax code application under pressure.
How long it takes: Two to three hours.
How to do it: Export all client communication from busy season and rank clients by total touch volume. A small share of clients typically drives a disproportionate share of total communication.
For the top-volume clients, categorize each interaction:
If your highest-communication clients are process-driven, automation would eliminate most of those touches. If they're complexity-driven, automation helps but doesn't eliminate them. If they're relationship-driven, automation can make the problem worse by removing the human connection these clients expect.
You may find an uncomfortable truth in this audit: clients who generate high relationship-driven volume are often unprofitable once you factor true time cost. The solution isn't software. It's pricing or a difficult conversation about the client roster.
When you run all five of these diagnostics a pattern emerges. Most small and midsize firms find that intake and document handling drive 30% to 40% of busy-season hours; rework comes primarily from extraction rather than judgment, and a long tail of high-touch clients drives outsized communication volume. If that matches your results, the order of operations is documents first, communication second, workflow last, and the ROI is measurable. If your results look different – judgment errors dominate rework, communication is primarily relationship-driven, pure tax preparation actually accounts for 60%+ of hours – you have a structural problem. AI would be at best a drafting assistant.
The leverage from this audit comes from pricing, scope, staffing, or client mix. The five hours you spend diagnosing a few weeks after tax season will save months of implementation effort and software costs that don't deliver returns. Diagnose first. Choose tools second. Only then will April look different.
Chris Farrell is the CEO of Liscio, the File Intelligence platform for accounting firms. Liscio publishes accuracy benchmarks for tax document classification at liscio.me.
Sign up for PICPA's weekly professional and technical updates by completing this form.
Statements of fact and opinion are the author's responsibility alone and do not imply an opinion on the part of the PICPA's officers or members. The information contained herein does not constitute accounting, legal, or professional advice. For actionable advice, you must engage or consult with a qualified professional.