Reporting Value Honestly Under Pressure

The first lesson built the parts of a value case. This one is about using them when results arrive, when several things changed at once, and when a sponsor or finance partner wants a clean number. The pressure to round up is real, and a leader's job is to resist it without becoming unhelpful.

Attribution when several things changed

Pilots rarely run in a quiet laboratory. Kestrel Valley Council, an invented local authority, piloted an assistant that summarises public consultation responses for its planning team. In the same quarter the team also gained a new case officer and moved to a revised report template. When turnaround improved, three explanations competed, and preference cannot settle that.

Useful options include comparing with a similar team that did not get the tool, comparing the same team's work with and without the tool in the same period, or looking at the steps the tool touches rather than the whole process. Where none of these is possible, say so: report the improvement, list the other changes, and describe the tool's contribution as plausible but not isolated. Claiming the whole gain for the tool because it is the newest change is the usual mistake.

Name a benefits owner

A benefit with no owner does not happen. Time saved is spread across individual diaries, and someone has to decide whether it becomes faster service, a smaller backlog, less overtime or nothing at all. The benefits owner is a named manager who holds the budget or target the benefit would land in, agrees the measure and baseline, and is accountable for reporting the result, including a disappointing one.

The owner is not the initiative lead by default, and not whoever happens to be available. It is usually the line manager of the work rather than the technology team. At Kestrel Valley the head of planning took the role, which meant the question of what to do with released time finally had someone to answer it.

The sponsor value story

When you report to a sponsor, lead with the evidence: what was measured, against what baseline, over what period and sample. Then say what the evidence suggests and, separately and plainly, what remains unproven or unmeasured. Finish with a specific decision request, such as extending the pilot to a second team for a stated period with the same measures, funding integration work, or stopping. A sponsor can act on a decision request, but cannot act on enthusiasm. A story that leads with the vision and tucks the caveats away invites the sponsor to repeat the vision without them.

Tailoring the view for a finance partner

A finance partner needs a one-page view in their terms: costs, timing, what is committed, what is estimated, and what depends on conditions. Tailoring changes order, vocabulary and level of detail. It must not change the evidence, soften the uncertainty or drop the conditions under which a benefit holds. If the time saving holds only when released hours go to measured work, the finance page says so. If a cost is an estimate, it is labelled as one. A page that hides gaps costs credibility on every later request.

When the evidence supports only a limited claim

Often the honest result is narrow: one team, one workflow, one period, a modest sample. State the claim at the size of the evidence. For consultation summaries in the planning team over one quarter, turnaround was shorter and corrections were no higher is a claim. The council speeds up with AI is not. A limited claim is still useful because it supports a bounded next step. Inflating it to win support creates a commitment the evidence cannot carry.

Comparing options honestly

The alternative is rarely the tool against nothing. It may be a different tool, a process change without AI, extra staff time, or waiting. Compare options on the same measures, the same cost categories and the same time horizon, and give the non-AI option its real cost rather than a weak version of it. Record the options you rejected and why. If the evidence cannot separate two options, say that and propose the cheapest test that would.

Do not import benchmarks as your result

Results reported by other organisations, whether by vendors, conference speakers or peers, describe their work, data, staff and starting point. They can help you form a hypothesis and decide what to measure. They are not evidence about your organisation, and they should not appear in your value case as an expected saving. If an outside figure is quoted at all, label it as external, give its source, and state how your situation differs. Better still, replace it with your own measurement as soon as you have one.

Common failure modes

• Reporting usage because outcome data is harder to obtain.
• Setting the baseline after the pilot, from recollection.
• Turning a time saving into money or satisfaction nobody measured.
• Crediting the tool with gains that other changes may explain.
• Leaving the benefit with no named owner.
• Asking a sponsor for general support instead of a specific decision.

None of these needs bad faith, because a clean story is easier to tell than an honest one. The remedy is a habit: say what was measured, what it shows, what it does not, and what you want decided.