AI saved your tax team time. How much reached profit?

Start with one kind of completed work

Choose a return type and a comparable group of engagements. Give each engagement one ID that connects its agreed scope, fee, work records, review outcome, invoice, and payment. A preparation tool's task counter won't tell you how much effort the whole engagement required.

Follow the work through client follow-up, preparation, review, corrections, and delivery. Record active time by role, including partner time that never appears on the client's bill. Keep time spent waiting separate. Six days waiting for a document and six hours preparing a return need different responses.

Use a fee-and-effort sheet with these fields. Your practice-management exports may already supply most of them.

Compare the same scope and quality standard before and after the change. Keep unfinished engagements visible so they don't disappear from the effort total just because they missed the reporting cutoff. When the mix changes, compare within each work type before combining the results.

That detail matters in a value-priced practice. A fixed fee tells you the agreed amount. Value pricing concerns how you set that amount around the client's value. Neither tells you whether a particular engagement takes twice the effort you expected. The effort record helps you see which scope, delivery, or pricing decision needs attention.

Give the released hours a destination

Suppose preparation gets faster but a manager still has to correct the same work. The saving may never reach accepted returns. Count all the checking and correction time before treating any hours as available. If review remains the limit, use the tax-review queue check to locate the cause.

Once time is available, assign an owner and a use. For extra client work, name the engagements, expected fees, required skills, and review slots. A list of prospects isn't sold work, and a preparer's spare afternoon doesn't replace a partner's judgment. For expense reduction, name the overtime, outside support, or other payment that can actually stop.

You can also choose a shorter busy season, training, or time to improve the practice. Those choices may serve the business goal. Record them as the use of capacity, with a measure that fits, instead of giving every saved hour a billing rate and calling the total profit.

Follow the same 200 hours through three outcomes

Let's say a tax team completes 100 comparable engagements in a month at an agreed fee of $2,000 each. Total delivery effort is 1,000 hours. An AI-assisted process reduces that effort to 800 hours, including review and corrections, while meeting the same quality bar. The team has released 200 hours, and the new system costs $2,000 a month to run.

In this example, existing salaries and overhead stay the same. The monthly revenue for the original work remains $200,000. That means the 200 hours alone haven't reduced payroll or increased revenue. What the team does next determines the change.

These are alternative uses of the same capacity. In the extra-work option, 20 engagements at eight hours each use 160 of the 200 released hours. The remaining 40 hours give the team room for variation. That option only works if the team has the right mix of preparation and review time, and the additional work is sold, completed, and earns the assumed fee.

The $30,000 is additional revenue less the extra costs in the example. It isn't the whole firm's profit. Finance still needs to account for setup, training, integration, and any other changed costs. Record those separately from recurring operating results so a good month doesn't hide what it cost to get there.

In the outside-support option, the team uses its released time to take over that work and the $12,000 payment actually ends. If the contract continues, the saving hasn't happened. If those hours are needed for outside-support work, they aren't also available for the 20 extra engagements.

This is where a time-savings estimate can go wrong. Multiplying the 200 hours by a notional $80 hourly cost produces $16,000, but salaries haven't changed. Adding that $16,000 to the $30,000 would count a second benefit from the same released capacity without a second financial change. Keep the hours in the capacity record and the dollars in the revenue-and-cost record.

Let the effort record improve the pricing decision

A lower cost per completed return gives you useful options, but it doesn't set the right fee for every client. Compare the agreed scope with the work actually delivered. If repeated follow-up, additional entities, or extra advisory work explain the effort, a faster preparation tool may leave the commercial problem in place.

For each engagement that falls short of the practice's target margin, have the pricing owner choose a response. The firm might change the delivery process, clarify what's included, agree a fee for added scope, or revise the next engagement's terms. Use the effort record to explain the choice. Don't assume every difference in effort calls for a price increase.

For hourly work, model what happens when billed hours fall. For a fixed fee, check whether credits, scope changes, or additional costs absorb the delivery gain. For a proposed new service, confirm that clients want it and that someone can deliver and sell it before assigning the released time to advisory revenue.

A fee cut also needs its own calculation. In the example, reducing the fee by $200 on the original 100 engagements gives up $20,000 in monthly revenue. That would reduce the extra-work option's $30,000 contribution gain to $10,000 if everything else stayed the same. It might still be a deliberate commercial choice, but the time-saving report can't make that choice for the partners.

Build the weekly record before buying another dashboard

Start with exports from the systems that hold engagements, time entries, invoices, payments, and tool costs. Finance owns the dollar definitions. The tax-practice leader owns completion and quality. The person assigning work owns the capacity plan. Agree on a weekly cutoff and use the same engagement IDs across all three views.

Use fixed calculations to total hours and money. An AI assistant can help group free-text review or scope notes into your firm's approved categories, but each suggestion should link to its source and wait for an owner's correction or acceptance. It shouldn't invent missing time entries, decide that an invoice is earned revenue, or turn an open sales opportunity into completed work.

At the weekly review, the example's month-to-date record would read like this: “Twenty added engagements completed and accepted. $40,000 of revenue verified by finance. $10,000 of added delivery and system costs. $30,000 of incremental contribution. Invoice and payment status tracked separately.” The work scheduler can then check whether the team used the planned 160 hours and whether another queue grew elsewhere.

Before relying on an automated report, reconcile a sample to the underlying records. Include a reopened return, a credit, a missing time entry, and work still waiting for acceptance. A missing value should remain unknown until its owner resolves it. Test that a corrected invoice updates the total once and that a reopened engagement doesn't appear as a second completion. If a spreadsheet answers the decision reliably, keep using it. Add an integration when maintaining the record becomes a repeated, material burden.

Expand when the next use of capacity is supported

At each review, compare the original plan with accepted work, actual fees, changed costs, and quality. Keep a forecast for the next period beside those observed results. If the hours are real but demand is missing, invest in a supported route to new work or choose another use for the time. If review absorbs the gain, address that step before adding more preparation capacity.

That gives the next investment a clear test: which additional work or removed expense will pay for it, who owns that action, and what evidence would make you stop. The firm can then expand a process that improves its economics without treating every productivity estimate as a profit result.