Use this worksheet to connect a workflow improvement to its economic consequences. Keep operational capacity, realized cash effects, and financial reporting treatment separate. The worksheet supports an investment discussion; it does not determine accounting policy.
Choose one period, such as a month, and use it consistently. Document every input as observed, estimated, or assumed, with a source or owner.
The seven steps below form a procedure. Each step produces something written down, and the last produces a decision. The worked example uses the same illustrative inputs as the economics chapter and the support-reply chapter; they are invented teaching numbers, not benchmarks. The companion workbook, linked from the pilot kit, implements Steps 1 to 4 with blank inputs.
Before you start: gather the inputs
For your finance lead
For each input in Step 1, record four things in the workbook's Inputs sheet or an equivalent table:
| Record | Why it matters |
|---|---|
| Value and period | Mixing a weekly volume with a monthly cost silently scales the result |
| Status: observed, estimated, or assumed | An assumption is a question the pilot must answer |
| Source or owner | Someone must be able to defend or correct it |
| How it was measured | Effort measured from timestamps, diaries, or sampling behaves differently |
Measure baseline effort before the tool is chosen. Once an assisted process exists, people's memory of the old effort shifts.
Step 1: Define the population and effort
For your finance lead
| Input | Symbol | Your value | Definition |
|---|---|---|---|
| Total units per period | V | To enter | Complete requests, documents, or other agreed units |
| Eligible share | E | To enter | Share within the approved scope, from 0 to 1 |
| Adoption share | A | To enter | Share of eligible units actually assisted, from 0 to 1 |
| Baseline human minutes | B | To enter | Complete effort per comparable unit |
| Assisted human minutes | T | To enter | Effort including review, correction, and follow-up |
| Hourly capacity value | R | To enter | Explicit assumption for valuing human time |
| Recurring costs | C | To enter | Costs not already counted in handling effort; mark which part is new cash and which is existing staff time |
| One-time setup | S | To enter | Integration, preparation, initial evaluation, and rollout |
If different types of work have materially different effort or value, calculate them separately. A blended average can hide that assistance helps one category and increases effort in another.
What to measure locally for each input:
- V from system logs over a representative period, net of duplicates and spam.
- E by classifying a sample against the pilot's written eligibility rules, not by estimating what share "looks routine".
- A from actual use during supervised operation; count units where a suggestion was offered and declined, because review still took time.
- B and T over the same observation window and unit of work, including follow-up and rework. T must include review: that is where savings most easily disappear.
- R as an explicit, documented value per released hour. It is a valuation choice, not a fact about payroll.
- C as continuing spending and continuing work not already inside T: software and usage, source upkeep, quality sampling, integration monitoring.
- S as one-time integration, preparation, initial evaluation, and training.
Step 2: Calculate capacity
For your finance lead
Assisted units = V × E × A.
Hours released = assisted units × (B − T) / 60.
Capacity value = hours released × R.
Capacity value after recurring cost = capacity value − C.
Retain negative results. If review and correction make T greater than B, the process consumes additional human capacity. Do not floor that result at zero.
Include recurring maintenance, evaluation, and training refresh in C where they are not already in T. Avoid counting the same labor twice. Include per-unit software charges, retries, and other usage costs using the same adoption scenario.
Decision point. If capacity value after recurring cost is negative under the base inputs, stop here and ask whether a different scope, a simpler alternative, or better measurement would change the inputs. Do not proceed to cash in the hope that realization will rescue a negative capacity case; it cannot.
Step 3: Identify realizable financial effects
For each benefit, name the operational mechanism, financial amount, evidence, timing, and responsible decision-maker. Examples include an avoidable contractor invoice, reduced overtime, or additional contribution from work that can actually be sold and delivered.
Do not assume that every released hour reduces payroll. Do not add a realized expense reduction to the full capacity value if both describe the same hours. Keep additional revenue hypotheses separate until attribution and contribution costs can be assessed.
Net recurring cash benefit = evidenced incremental cash benefit − incremental recurring cash costs.
Simple cash payback = setup cash cost / positive net recurring cash benefit per period.
If net recurring cash benefit is zero or negative, this formula provides no finite payback. It also ignores timing within periods, uncertainty, discounting, taxes, and other considerations needed for a fuller investment analysis.
Record each mechanism in a table like this, one row per mechanism:
| Mechanism | Evidence that it exists | Amount and timing | Who can make it happen |
|---|---|---|---|
| Avoided contractor hours | A contract that can be reduced, and its notice period | Rate × hours no longer bought, from the end of the notice period | Budget owner for the contract |
| Reduced paid overtime | Overtime records for the affected team | Overtime actually not paid after rollout | Team manager and payroll |
| Avoided planned hire | An approved requisition that would otherwise be filled | Loaded cost of the role from its planned start | Hiring manager and finance |
| Additional sold work | Demand that is currently turned away, and capacity to deliver it | Contribution after delivery cost | Commercial owner |
If no row can be completed with evidence, the cash benefit is zero, and the capacity value describes time that may be used well but is not a saving. The finance leaders guide shows how three proposals look under this test.
Step 4: Work a base and downside case
Suppose V = 5,000, E = 0.60, A = 0.80, B = 8 minutes, T = 5 minutes, R = $45/hour, C = $2,000/month, and S = $12,000. Assisted volume is 2,400 and released time is 120 hours. Capacity value is $5,400; after recurring cost it is $3,400.
If only 70% of that valued capacity becomes a demonstrably avoidable expense at the assumed rate, gross cash benefit is $3,780. Treating the whole of C as new cash spending, net cash benefit is $1,780 monthly; if existing staff absorb part of C, that part is consumed capacity, not cash, and is reported separately. Simple setup payback is approximately 6.7 months. The 70% is an illustrative realization assumption requiring evidence, not a default conversion rate.
In a downside case, review increases T to seven minutes. Released capacity falls to 40 hours, worth $1,800. Capacity value after recurring cost is negative $200. With the same 70% cash-realization assumption, net cash benefit is negative $740. There is no positive simple payback under those assumptions.
Choose the downside from the input most likely to be wrong, not from a uniform percentage haircut. For assisted work, the effort inputs are the hardest to know before a pilot, which is why Step 5 tests every input. Keep the base and downside side by side in the decision memo; a base case reported alone invites the reader to assume it.
Step 5: Find the assumption that matters
For your finance lead
For a capacity-only comparison with fixed C, the required minutes saved per assisted unit are 60 × C / (V × E × A × R). With the base inputs above, about 1.11 minutes are needed just to cover recurring cost in capacity terms. At a 70% realization assumption, about 1.59 minutes are needed to cover it in the simplified cash scenario.
These thresholds exclude setup and require positive volume, adoption, value, and realization where applicable. If costs vary with volume, recalculate them for each scenario instead of treating C as fixed.
Which assumption moves the answer most
- Result falls
- Result rises
- Base result $3,400
Show the data
| Input | Low input | Result at low | High input | Result at high |
|---|---|---|---|---|
| Baseline minutes | 6.4 min | $520 | 9.6 min | $6,280 |
| Assisted minutes | 4 min | $5,200 | 6 min | $1,600 |
| Monthly volume | 4,000 | $2,320 | 6,000 | $4,480 |
| Eligible share | 48% | $2,320 | 72% | $4,480 |
| Adoption | 64% | $2,320 | 96% | $4,480 |
| Hourly value | $36/h | $2,320 | $54/h | $4,480 |
Then move each input by the same proportion, one at a time, and rank the effect on the result. In the example, a 20% change in baseline minutes moves capacity value after recurring cost by $2,880 either way and a 20% change in assisted minutes by $1,800, against $1,080 each for volume, eligibility, adoption, and hourly value. The two effort inputs lead because they act on the three-minute difference between them. Per minute, the two have the same leverage; the baseline bar is longer only because 20% of eight minutes is more than 20% of five.
A uniform step ranks leverage, not importance. What matters is leverage times how uncertain the input is: an assumed input with a wide plausible range deserves attention before an observed one with a longer bar. Use each input's status from the inputs record to weigh the bars. In this example the effort inputs are both the highest-leverage and the assumed ones, so the instruction holds: measure baseline and assisted effort before refining the volume forecast.
The ranking depends on the inputs. A workflow with a large difference between baseline and assisted effort, or a small recurring cost, can rank differently; recompute rather than reuse this order.
Step 6: Check non-labor value separately
For your finance lead
Inventory release changes cash tied up in stock; it is not the same as annual carrying-cost reduction. Revenue should be evaluated through incremental contribution after delivery cost. Risk benefits need a credible loss mechanism and uncertainty, not an invented precise avoided-loss number.
Work each non-labor effect in its own units and keep it out of the capacity total:
- Working capital. A one-time release of cash when average inventory falls, plus an annual carrying value at an explicit carrying rate. The forecasting chapter works an example and shows why the two must not be added.
- Revenue. Incremental contribution after delivery cost, with an attribution design such as a holdout. Under time-and-materials billing, faster work can reduce billed hours; the professional services guide shows how the billing model changes the sign.
- Risk. A named loss mechanism, its frequency evidence, and the control's effect on it. Without those, describe the risk qualitatively.
Replace assumptions with pilot measurements as they become available.
Step 7: Write the decision
The worksheet is finished when it supports a decision someone can act on. Record it in a fixed shape so later readers can see what was known:
| Item | What to write |
|---|---|
| Decision | Fund a bounded pilot, fund measurement only, revise the scope, or decline |
| Capacity result | Base and downside, after recurring cost, with the period stated |
| Cash result | Each realization mechanism with its evidence, or a statement that none is evidenced |
| Dominant input | The input from Step 5 that moves the result most, and how it will be measured |
| Setup and payback | Setup cost and, only where net cash benefit is positive, simple payback with its caveats |
| Non-labor effects | Working capital, revenue, or risk, each in its own units and kept out of the capacity total |
| Next observation | What measurement would change the decision, who will make it, and when |
A decision to fund measurement only is legitimate. When the dominant input is unmeasured, a short baseline study can cost less than a pilot and answer the question that matters. The finance leaders guide shows that outcome for one of three proposals.
Common errors
| Error | How it appears | Correction |
|---|---|---|
| Review counted twice | Agent review inside T, and again as a separate cost in C | Count each activity once; put periodic quality sampling in C only if it is not inside T |
| Capacity reported as cash | "Saves $5,400 a month" with no named mechanism | Report capacity and cash on separate lines; cash needs a row in Step 3 |
| Continuing work ignored | Source upkeep and evaluation left out of C | List continuing activities and their owners; existing staff time is consumed capacity even if not new spending |
| Blended averages | One effort figure across categories with different savings | Calculate categories separately |
| Setup inside the monthly result | One-time cost spread silently over a year | Keep S separate and state any payback as a separate calculation |
| Optimistic adoption | A taken as the share of cases that could be assisted | Measure actual use during supervised operation |
| Revenue added to capacity | Released hours valued once as capacity and again as sold work | Choose one treatment for the same hours |
Using the companion workbook
For your finance lead
The workbook's Economics sheet takes the Step 1 inputs per workflow column and computes released hours, capacity value after recurring capacity cost, and net cash benefit from separately entered avoidable expense and incremental cash cost. It returns no payback when net cash benefit is zero or negative. The Inputs sheet records each input's basis, source, and measurement; the Evidence sheet tracks the pilot's requirements; the Evaluation sheet holds case records for the pilot. The workbook contains no measured data and models no ramp-up, tax, financing, or demand change, so a decision that depends on those needs a fuller analysis.
Related chapters: economics foundation, finance leaders, pilot design, inventory forecasting.