Short answer
Workflow ROI is the net value of a process improvement divided by its total cost. The calculation is only as reliable as its baseline: measure one defined process before changing it, then compare the same outcome after the change. Count time released, rework avoided and cash effects separately. Do not turn every minute saved into a claimed cash saving.
For a 12-month assessment:
ROI = (verified 12-month benefit − 12-month total cost) ÷ 12-month total cost × 100%
“Verified benefit” means a benefit with a source, calculation and owner who can defend it. If you do not have those inputs yet, report a capacity or service improvement—not a financial ROI percentage.
Step 1: define the process boundary
Write down the event that starts the process and the business outcome that ends it. For example, a supplier invoice exception may start when an invoice fails a matching rule and end only when the approved decision and ERP posting or rejection are both recorded. A task marked complete is not necessarily the end of the business outcome.
Record the period, team, locations, case types and exclusions. Do not mix a simple low-risk request with a complex exception without segmenting the data; the case mix can move the average even when the workflow improves.
Step 2: capture a defensible baseline
Use the system of record, workflow history and a short sample of observed work. If the current process lives in email and spreadsheets, record how the sample was chosen and the uncertainty it introduces.
| Measure | Baseline value | Source and owner | After-change value |
|---|---|---|---|
| Cases completed in the period | |||
| Median elapsed time from request to outcome | |||
| 90th-percentile elapsed time | |||
| Manual follow-up minutes per case | |||
| Cases with rework or missing evidence | |||
| Overdue or blocked cases | |||
| Cases closed with required proof |
The median describes a typical case; the 90th percentile exposes the long tail that customers and managers often feel. Keep the start and end timestamps identical across the before-and-after periods.
Step 3: calculate the benefit without double counting
Time released = monthly case volume × minutes of manual work removed per case ÷ 60.
The published Intelliflow article gives a purely illustrative example: 600 cases × eight minutes of status chasing = 80 hours a month. If half of that chasing is removed, 40 hours of capacity are released. That is not automatically 40 hours of payroll saving. It may let the same team handle more work, improve service or avoid an additional hire; only an approved financial model can value those outcomes.
Rework avoided = reduction in repeat cases or repeat actions × measured cost per repeat. Exclude any time already counted in “time released.”
Verified cash effect = an evidenced reduction in penalties, write-offs, external fees or other actual spend. Do not monetise “risk reduced” merely because a dashboard shows fewer exceptions. Confirm that the cost was real, attributable to this process, and not already included under rework.
For a financial estimate, apply an explicit realisation factor to released capacity. A value of 0% is appropriate when no capacity is redeployed or cost avoided; 100% requires strong proof. The factor is a management assumption, not a product performance claim.
Monthly verified benefit = realised capacity value + rework cost avoided + verified cash effect
Report non-financial outcomes—faster decisions, better evidence and improved visibility—alongside ROI, not hidden inside an invented monetary figure.
Step 4: include the full cost of ownership
Include discovery, configuration, integration, data preparation, testing, training, adoption time, support, licences or subscriptions, and ongoing process ownership. Keep one-off and recurring costs separate.
First-year total cost = one-off implementation cost + 12 × monthly recurring cost
If the benefit ramps up, use the actual expected month-by-month profile rather than multiplying the final month by twelve. A pilot that takes three months to stabilise will not deliver a full year at its mature run rate.
Step 5: calculate ROI and payback
Add each month's verified benefit for the first year, then apply the ROI formula at the top of this guide. A negative result is useful information: it may mean the process is too small, the solution is too expensive, or the benefit has not yet been evidenced.
If monthly net benefit is positive and reasonably stable:
Simple payback months = one-off implementation cost ÷ (monthly verified benefit − monthly recurring cost)
If the denominator is zero or negative, there is no simple payback at that run rate. This simplified formula ignores timing and the cost of capital; finance should use its normal investment method for a material commitment.
Copyable calculation sheet
| Input | Your value |
|---|---|
| Monthly cases in scope | |
| Follow-up minutes removed per case | |
| Released hours per month | Cases × minutes ÷ 60 |
| Realised capacity value per month | |
| Rework cost avoided per month | |
| Other verified cash effect per month | |
| Monthly recurring cost | |
| One-off implementation cost | |
| First-year verified benefit | Sum of actual monthly benefits |
| First-year total cost | One-off + sum of monthly costs |
| First-year ROI | (Benefit − cost) ÷ cost × 100% |
How to test the result fairly
- Compare the same case types and outcome definition before and after the change.
- Record changes in volume, staffing, suppliers, policy and seasonality; any of these can move the result independently of software.
- Check that “saved” time is not merely shifted to another team or to the customer.
- Use an owner from operations and an owner from finance to sign off the inputs.
- Repeat the calculation after the first live cycle and again when the process has stabilised. Label projected, observed and financially verified figures separately.
When not to make an ROI claim
Do not publish a percentage when the baseline is absent, the sample is too small to represent normal work, costs are incomplete, or the supposed saving is an untested assumption. A clearer statement is: “We released an estimated 40 hours of monthly follow-up capacity in this illustrative model; financial benefit is not established.”
For Intelliflow specifically, an approved customer case study needs consent, the actual deployment scope, the measurement method and a finance-reviewed result. An illustrative calculator is useful for planning, but it is not customer proof.
Frequently asked questions
Should ROI include only labour savings?
No. Rework and verified cash effects may matter more. Keep each category separate and prevent overlap. Evidence, service and accountability improvements can be reported without assigning an unsupported rand value.
What if the process improves but ROI is negative in year one?
Show the result honestly. A high setup cost can make the first year negative while later years differ. It may still be worth doing for a required control or service obligation, but that is a separate decision from claiming positive financial ROI.
Can a dashboard prove the saving?
No. A dashboard can show volumes, delays and exceptions. The saving requires a credible baseline, a causal explanation, a cost model and evidence that the benefit was realised.