A credible ROI model starts before purchase
Workplace technology ROI cannot be rescued by a confident percentage after rollout. The model begins with a specific operating decision, a baseline that describes the present condition and a small number of outcomes the organisation is prepared to measure. Without those elements, licence cost is precise while the claimed benefit is imaginary.
Separate three layers: adoption signals, operational outcomes and financial consequences. Logins or reservations can show whether people reached the tool. They do not prove that space, support, collaboration or employee experience improved. Financial value appears only when a measured operational change affects a real cost, risk or capacity decision.
Treat the first model as a set of testable assumptions. Record the owner, evidence, period and confidence for each one. The purpose is to decide whether to proceed, adjust or stop, not to manufacture a business case that can only move upward.
| Layer | Example measure | Decision it can support |
|---|---|---|
| Adoption | Successful completion of a priority task | Whether the workflow is usable |
| Operation | Support demand or avoidable booking conflicts | Where process friction remains |
| Resource | Space or service capacity used by time and location | Whether capacity can be changed |
| Experience | Task-specific satisfaction with a defined sample | Which journey needs improvement |
| Financial | Verified cost avoided or capacity redeployed | Whether realised value exceeds full cost |
Define the decision and counterfactual
State what decision the measurement must unlock. A facilities team might need to decide whether to consolidate a floor, change a booking policy or extend a pilot. IT might need to decide whether support and integration effort are sustainable. Finance might need to decide whether the realised benefit justifies the next contract period.
Then describe the counterfactual: what is likely to happen without the investment. “Current cost” is rarely static. A lease event, headcount change, existing tool renewal or policy shift may alter the baseline. Record those changes so the technology does not receive credit for a benefit created by another decision.
Use a time boundary long enough to observe the outcome but short enough to influence action. A two-week pilot may reveal task failure, while a space decision may require seasonal evidence. Do not combine periods with different occupancy patterns unless the model explains the adjustment.
Build a baseline people can audit
Collect only data that serves the decision. Useful baseline evidence may include successful task completion, no-shows under a defined policy, support tickets by cause, time spent on manual administration, available capacity and the cost of systems or processes being replaced. Define every measure so two people would calculate it the same way.
Describe coverage and gaps. Badge data, booking data, Wi-Fi signals and surveys observe different things and carry different privacy, accuracy and inclusion limits. A reservation is not presence. A network connection is not productive work. A voluntary survey may not represent employees whose journey is most difficult.
The GOV.UK Service Standard guidance on defining success recommends choosing metrics that reveal whether a service solves its problem and establishing a baseline. The public-service context differs, but the discipline is directly useful: know what good means before interpreting a dashboard.
Model the full cost of ownership
Include subscription, implementation, mapping or hardware, identity and calendar integration, migration, privacy and security review, training, communication, support, administration, analytics work and exit. Add internal time when the decision compares the technology with other uses of the same capacity. Avoid false precision where time estimates are uncertain.
Separate one-off and recurring costs, then show which costs grow with people, locations, resources or data. Model the same operating scenario across alternatives. If one quotation excludes implementation and another bundles it, normalise the scope before using either in the calculation.
Microsoft’s application modernisation assessment guidance connects inventory, cost, readiness and business planning. It is publisher-owned guidance rather than independent ROI evidence, so use the assessment categories while validating costs and benefits in the organisation’s own environment.
Measure adoption as a pathway, not a launch event
People cannot adopt a workflow they cannot access, understand or fit into real work. Track the path from awareness to successful repeated task completion. Look for permission failures, duplicate entry, abandoned attempts, workarounds and support requests. Overall active-user counts can hide one role or location that is effectively excluded.
The Microsoft 365 Adoption Guide separates envisioning, onboarding and driving value, including business scenarios, early adopters, feedback, usage and ongoing engagement. It is a product publisher’s framework, so it should guide questions rather than serve as proof that a particular investment will pay back.
Compare cohorts and tasks rather than ranking employees. A lower rate may reflect a role that rarely needs the workflow, inaccessible design, missing mobile access or a policy conflict. The useful question is what prevents the intended outcome, not how to force every person toward the same usage number.
Convert operational change into financial value carefully
A measured time saving has financial value only if the organisation can explain what happens to that time. It may increase service capacity, reduce overtime, prevent a hire, shorten a customer journey or simply remove frustration. Each outcome is useful, but they are not interchangeable and should not all be booked as cash savings.
For space, distinguish theoretical capacity from an executable property decision. Lower average occupancy does not automatically release cost. Lease terms, peak demand, team patterns, accessibility, meeting space and growth scenarios determine whether a floor can actually be changed. Record the decision date and the constraints that make the saving real.
Show benefit ranges based on verified volumes and conservative conversion assumptions. Keep avoided cost, cash saving, redeployed capacity, risk reduction and experience improvement on separate lines. A single total can be presented, but the decision maker should be able to see which elements are realised, forecast or still unquantified.
Run a measurement cycle that can change the plan
Review leading signals weekly during a pilot: successful priority tasks, failed integrations, support causes and policy exceptions. Review outcome and financial measures at the cadence needed to observe change. Give every measure an owner who can explain its collection, limits and next action.
Use thresholds agreed before results arrive. Decide what evidence supports scaling, what requires a change and what should stop the rollout. A threshold can include quality and risk conditions, not only a target number. For example, adoption may be adequate overall while an accessibility failure still blocks expansion.
A bounded enterprise software pilot provides the right place to test the model. If task completion, support or data quality cannot be measured in the pilot, a portfolio-wide ROI percentage will not become more credible after deployment.
The investment review
Bring the decision statement, baseline, coverage limits, full cost, outcome measures and current evidence to one review. Ask whether the technology caused the observed change, whether the change can be sustained and whether the organisation can act on it. Record alternative explanations instead of burying them in a sensitivity note.
For desk and space decisions, the hybrid-office booking systems comparison offers a procurement lens, while the service desk pathway helps frame support and recurring demand. Link the economic model to both the buying decision and the operating system that follows it.
A credible ROI conclusion may be “continue measuring” or “the value is operational but not yet cash-releasing”. That is stronger than an inflated number. The model succeeds when it makes the next investment or operating decision clearer, not when it produces the largest percentage.
