Practical guide
How to measure adoption and behaviour change that enable benefits
Distinguish training and access from correct, sustained practice, then use adoption evidence to choose operational action and review the intended benefit.
Wovex
Summary
Measure adoption by defining the behaviour the benefit depends on, identifying eligible people or work and observing whether the practice is used correctly over time. Keep availability, training attendance, use, competent application and business results separate. Explain denominators, exceptions and source limitations before reporting percentages. Use weak evidence to investigate practical barriers and choose an operational response. Adoption can enable a benefit; it does not establish that the intended improvement occurred or that the change caused it.
Define the practice the benefit depends on
Adoption is useful to a benefits review when it shows whether people use a new capability or practice in the way the intended improvement requires. Availability, training attendance and login counts can explain parts of that story, but none establishes the complete change in behaviour.
Start with a specific statement: which people need to do what differently, for which work, and why should that help? A new case-handling procedure might require staff to check a shared record before requesting information again. The important behaviour is the check and the resulting handling decision, not simply opening the application.
Identify the conditions needed for that practice. People may need usable information, clear instructions, time, access and support. Apparent resistance may instead reflect a practical barrier or a process that does not fit the work.
The Magenta Book distinguishes questions about implementation from questions about an intervention’s impact. Use that distinction to investigate whether the change is operating as intended while continuing to collect separate evidence of the business result.
Use an adoption-to-benefit evidence ladder
The following ladder is a management aid. It separates useful observations rather than assigning a universal maturity score or assuming every change progresses neatly through the same stages.
Evidence level | What it can establish and what remains open |
|---|---|
Availability | The capability or process is accessible to the intended users. It does not show use. |
Preparation and reach | Relevant people received guidance or attended training. Attendance does not establish competent practice. |
Actual use | The new approach is used for eligible work. Access or login activity alone may not establish this. |
Correct application | The important steps are applied appropriately, with relevant quality checks and exceptions understood. |
Sustained practice | Appropriate use continues across later periods and ordinary operating conditions, not only during a supported launch. |
Outcome and benefit | Performance or experience improves in a way that matters. The observation still needs interpretation and consideration of other influences. |
Choose evidence that can challenge the connection between levels. If people use the new process correctly but the intended improvement does not emerge, the explanation may lie in the design, another dependency or an external influence. More adoption activity is not automatically the right response.
Use the outcomes management guide for the distinctions between outputs, use, outcomes and benefits. This ladder helps decide what to observe between delivery and a defensible result.
Fix the population and observation method
Define who and what is eligible before calculating an adoption percentage. The population might be staff performing a particular task or cases for which a procedure applies. Those are different denominators and should remain recognisable.
State exclusions and valid exceptions. An urgent case legitimately following another procedure should not be marked as failed adoption simply because it bypasses the routine route. Equally, excluding difficult cases without explanation can make the new practice appear more successful than it is.
Use proportionate process evidence: suitable operational records, permitted sample checks, discussion with users and observation of how the work is performed. Agree the purpose and information needed. This guide concerns learning about the change process, rather than ranking individuals or building employee surveillance.
Check what the source really establishes. A completed checkbox may show that somebody recorded a step, not that the step was performed competently. Missing documentation is an evidence gap; it should not silently become proof of either correct or incorrect practice.
Compare preparation with application in a fictional example
A fictional service introduces a procedure intended to reduce repeat requests for information. Staff are expected to check existing case information and apply an exception rule before contacting the customer again.
Training attendance is high, but the operational review asks whether eligible cases are handled in the intended way. The figures below are fictional observations, not customer results, universal thresholds or a statistically validated sample design.
Evidence cohort | Illustrative finding and interpretation |
|---|---|
Preparation cohort | 180 of 200 eligible staff attended training: 90% attendance. This is a people-based measure. |
First case review | Of 100 reviewed eligible cases, 60 have documented use of the procedure and 48 of those meet the agreed application checks. Correct documented application covers 48% of all reviewed cases, or 80% of the 60 recorded uses. |
Later case review | Of 120 reviewed eligible cases, 96 have documented use and 84 meet the application checks. The corresponding figures are 80% documented use and 70% correct documented application across the reviewed cases. |
The first review therefore does not support describing the service as 90% adopted. Attendance concerns people; application concerns reviewed cases. Even the case percentages answer different questions depending on whether the denominator is all reviewed eligible cases or recorded uses.
The later observations suggest better application within those reviewed cases. Check selection, coverage, case mix and consistency before treating the change as representative of the whole service. The reviews do not follow a fixed set of individual staff and should not be presented as a personal performance comparison.
Keep the business result separate. The service still needs evidence about repeat requests, service quality and customer experience before claiming that the intended benefit has occurred.
Connect weak evidence to an operational response
Discuss the findings with the people doing the work. In the fictional example, they might explain that reference information is difficult to find or that the exception rule is ambiguous. Treat these as hypotheses to investigate, not established causes inferred from a low percentage.
Use a short response sequence:
Identify which expected practice is unsupported or inconsistent, using the defined eligible work.
Examine source limitations, valid exceptions and practical barriers with the operational team.
Choose a proportionate response, such as clearer guidance, improved access or a change to the process.
Name the person responsible, the work and support required, and the next observation point.
Review application and the business result together, retaining adverse effects and alternative explanations.
A refresher session may help a knowledge gap. It will not repair missing source information or an unsuitable process. Record why the selected intervention addresses the evidence rather than automatically prescribing more training.
Continue observation beyond the supported launch
Check whether the practice persists when initial coaching ends, demand changes or new people join. Choose intervals that fit the work and available evidence. A launch-week observation may not describe normal operations.
The Teal Book provides context for maintaining benefits work through delivery and operational use. Connect the continuing adoption observations with the relevant benefit owner and review, rather than running an unrelated usage-reporting exercise.
Use the customer and service outcomes guide for the final service measures. For AI-enabled change, the AI and digital investment value guide considers operating costs, checking and realised value beyond a technical pilot.
Suitable software may organise records and review information, but this evidence ladder does not imply native usage telemetry, behavioural analytics or automatic collection from another application. Keep the observation method and its limits explicit, whichever tools are used.
Sources and context
HM Treasury: Magenta Book, Central Government guidance on evaluation.
UK Government Project Delivery: Teal Book Chapter 19, Benefits management.
Sources checked 20 October 2026. The official sources provide UK evaluation and project-delivery context. The evidence ladder and cohort figures are Wovex-authored fictional illustrations, not validated sampling guidance or customer results. No employee-surveillance method, native telemetry or automatic external collection capability is asserted.