Build a review that requires a choice
This guide is a retrospective analysis of 2025 written from a 2026 perspective. The scenario and amounts demonstrate a method rather than CYTIZEN client results. A benefits review is not a ceremony where the project repeats what it delivered. It checks whether change creates its intended effect, whether that effect is measured properly and what to decide when the gap persists.
An installed application, trained users and a published procedure are outputs. A shorter lead time, avoided expenditure or a more reliable control are outcomes. Their relationship should be described before launch and tested after handover. The benefit owner is the person able to organise the work needed to realise it. A project manager cannot promise savings that depend on an organisational decision outside their authority.
Illustrative scenario: eight minutes promised, two recovered
Lucie, the procurement lead, presents the first results of a new request workflow. The business case promised eight minutes saved per request. Across 10,000 annual requests and an illustrative hourly value of €45, the theoretical benefit was €60,000. The programme delivered on time. Three months after launch, however, users still consider the journey slow.
Karim, the financial controller, separates handling time from waiting time. Handling fell from twenty to twelve minutes for complete requests. Only 60% of requests are complete, while the remainder require fifteen minutes of rework. Current average handling is therefore 12 + 40% × 15 = 18 minutes. The average observed saving is two minutes, or €15,000 of valued capacity before additional operating costs.
The review does not conclude that the tool is useless or that the business has failed. It identifies the variable preventing the benefit: input quality. Lucie receives authority to improve required fields and request guidance. Karim proposes revisiting the completeness assumption after two comparable measurement cycles. The sponsor funds that bounded correction and pauses expansion to another country.
A completed benefit record
| Element | Actionable example |
|---|---|
| Identifier and outcome | B-07: reduce handling effort for a standard purchase request |
| Scope | Domestic non-urgent requests, same purchasing categories before and after |
| Baseline | 20 average minutes from a documented representative pre-launch sample |
| Target and mechanism | 12 minutes through a simplified workflow with sufficiently complete requests |
| Owner | Procurement lead, authorised to change the form and rework organisation |
| Measure | Total handling time including rework; retain volumes and exclusions |
| Value | Capacity rather than committed cash savings; €45/hour is an illustrative internal convention |
| Response to a gap | Improve input quality; reconsider expansion if net saving remains below four minutes |
Attach observation sources, exclusion definitions and the person approving the calculation. If the measure changes, retain the previous version and explain the discontinuity. The benefits register must not become a spreadsheet in which targets are quietly rewritten to match actual results.
A target concerns an outcome rather than a general feeling of improvement. Writing fewer rework events per complete request requires a definition of rework and observation of the journey. Writing optimise procurement allows everybody to interpret success differently. A good record is precise enough for another owner to reproduce the measure using the same data and rules.
Prepare the evidence before convening the committee
The benefit owner provides results and the business explanation. Finance checks comparability and valuation. Applications confirms relevant usage and incidents. Programme coordination prepares dependencies and decisions. The sponsor decides commitments beyond the owner's mandate. This division prevents the programme office from having to manufacture business truth on its own.
In the example, preparation gathers before-and-after times, request volumes, completeness, rework, use of the new workflow and additional costs. Evidence is circulated three days before the meeting with known anomalies. Missing data appears as an explicit status, with an owner and next measurement date. It does not become green simply because there is no proof that performance is poor.
An average alone is insufficient. Include complexity categories and the spread of lead times when they affect the choice. A small number of lengthy requests can explain dissatisfaction despite an improved mean. Compare similar requests; a rise in international volume or urgent work may change the profile without the tool being the only cause.
Preparation should also identify which figures are observed and which are extrapolated. A monthly sample can support an annual estimate, but the estimate must retain its assumptions about seasonal volume and case mix. A benefits review becomes unreliable when projections appear in the same column as realised savings with no distinction.
A forty-five-minute meeting
The first ten minutes establish facts: baseline, period, scope and data quality. The next fifteen address no more than three gaps with a decision consequence. Each owner explains the expected mechanism and where it is failing. Ten minutes compare alternatives: correct the journey, improve adoption, revise an assumption or stop an expansion. The final ten assign decisions, owners and the next evidence required.
This format does not prohibit deeper analysis outside the meeting. It prevents spending forty minutes discussing dashboard colours and five minutes requesting vague actions. Gaps needing no escalation remain with accountable teams and return to the sponsor only when their authority or agreed limits are exceeded.
A completed decision for B-07 could state: extend completeness checks to standard requests, with Lucie accountable; test on 300 requests and compare rework and total handling time; correction budget capped at €8,000; Karim approves interpretation on the twentieth of the month; country expansion remains paused until net saving reaches at least four minutes without increased abandoned requests. Every number is an illustrative parameter rather than a universal rule.
Classify benefits without counting them twice
Cash savings require expenditure actually reduced or avoided: an external service removed, a contract resized or a purchase cancelled. Released capacity measures available time that must be reassigned to produce an outcome. Quality can be measured through rework, defects or successful controls. Lower risk is explained through exposure and safeguards without always forcing it into an artificial monetary value.
In this case, €15,000 is a capacity valuation. If that capacity avoids €10,000 of external assistance, show the avoided expenditure separately and do not add the same portion of time again. If it reduces a critical delay, measure that delay and explain why it matters. Financial and service tables can coexist without forced conversion between them.
Negative benefits also belong in the record: added checking work, a new dependency or additional incidents. A new process may improve one function while transferring effort to another. Follow the whole journey, rather than only the team sponsoring the investment. Where owners disagree about the transfer, make the disagreement a decision requiring reconciliation instead of choosing whichever number best supports the programme.
Understand the gap before moving the target
Separate four causes. The solution may fail to deliver its technical effect. The change may not be adopted. An external condition may have changed, such as volume or regulation. Finally, the original assumption may have been wrong. These call respectively for a product correction, an organisational action, a new context assessment or an explicit revision of the business case.
For B-07, poor completeness may stem from an unclear form, data unavailable when the request is made or an unnecessary control. The same training intervention does not solve all three causes. Observing a few requests and recording information available at each step often costs less than issuing repeated general reminders.
The review keeps the original target and adds a revised forecast. If expected savings change from eight to four minutes, explain why, who accepts the reduction and its investment consequence. A target can legitimately evolve. Transparency distinguishes that adaptation from erasing the gap. Keep changes connected to approved decisions so that the final outcome can be understood without reconstructing informal conversations.
The outcome still has an owner after project closure
At closure, every unrealised benefit transfers to an owner with authority, access to data and a review date. The programme sponsor cannot simply hand over a list and consider follow-up complete. Operational management must accept the work and allocate the necessary capacity. Otherwise the project has handed over an aspiration rather than a managed result.
Frequency follows the benefit mechanism. A high-volume daily journey may support monthly measurement; a quarterly close or annual renewal needs another calendar. Frequent reviews of a slow outcome create noise, while an annual review of rapid deterioration arrives too late. Choose the period in each benefit record rather than imposing one schedule on every case.
Close a benefit when its measure is confirmed over the agreed period and any continuing monitoring enters ordinary management indicators. Retain evidence and decisions. Recurring savings still need review when contracts, volumes or operating conditions change. The closure record should state which indicator now carries the result and who responds if that indicator deteriorates.
Adapt the review to actual sector mechanisms
In finance, invoice automation is assessed through reconciliation, exceptions and closing deadlines. The number of automatically processed documents is insufficient if rejected cases require more work than before. The finance owner should also confirm whether deferred exceptions compromise the period's accounts.
In a regulated industry, quality benefits retain necessary controls: shorter approval times, recovered traceability or fewer revisions. Acceleration is not success if it removes a required review or weakens document control. Define the quality measure together with the authorised function rather than assuming that a faster workflow is always better.
In IT services, distinguish fewer incidents from less reporting. Compare tickets with failed journeys and targeted feedback. A more difficult support portal may reduce recorded tickets without improving the service. Similarly, a new self-service channel may initially increase visible demand by exposing previously hidden problems; interpret that change before declaring a failure.
Sources, method and limitations
Primary sources consulted on 4 October 2026: APM on benefits management and project success; Department of Finance introduction to benefits management; HM Treasury Magenta Book. They inform the discussion of realisation and evaluation; the completed records are teaching constructions.
This guide proposes a decision-oriented review rather than a universal accounting method. Amounts, thresholds, samples and durations require adaptation to organisational rules. Measuring an outcome alone does not establish causality: comparable conditions and external factors still matter. The review should help the sponsor choose, including reducing or stopping an expansion, rather than turn every delivered project into an assumed success.