We're delivering, but need to demonstrate impact
Measuring and communicating value through a people-centred lens
Your work deserves to be seen
If you are delivering real value but struggling to demonstrate it, the gap is rarely the impact itself. It is articulation, or the absence of the baseline data, success metrics and ROI evidence needed to make the case. In some cases, costs were not tracked carefully enough to show clear value for money, and where spend has outpaced benefit, that story is harder still to tell.
This matters more than ever in 2026. The Action Plan: One Year On is explicit that future investment will follow demonstrated value, and the National Audit Office and Treasury are sharpening their expectations of how public sector AI benefits are evidenced.
10 million workers to be upskilled in AI by 2030, measured against business outcomes rather than training completion
(AI Skills Boost, 2026)
Future of Work Unit established in DSIT to track AI’s impact on the labour market and how roles evolve
Cafcass Scribe — hyper-personalised, age-appropriate letters with multilingual and audio options for neurodiverse and accessibility-need children, with full Azure governance and ATRS-compatible auditability
(Version 1)
"Every public sector organisation is under pressure to clearly demonstrate value to senior stakeholders, funders, and citizens. Success is measured by how organisations reimagine value creation, not by the number of pilots."
Version 1 ‘From Playbook to Progress: Delivering AI in the Public Sector’
Understanding the challenge
Demonstrating impact is not just metrics. It is a credible, evidenced human story. Three things tend to go wrong:
- Measuring the wrong things. Technology metrics such as model accuracy, latency and uptime are necessary for operational management, but they are not evidence of value. The metrics that move the audiences who fund you are tied to departmental objectives: speed of service, decisions made better, caseworker time freed, citizens better served. If your reporting pack leads with technical telemetry rather than outcome data, the people who control your budget will not see what your programme is actually delivering.
- Stories are anecdotal, not systematic. A few caseworker quotes are not the same as a structured evidence base.
- The narrative misses the people. The most compelling evidence of transformation is how roles, services and citizen experience are changing — not what the dashboards say.
Three things that will move this forward this quarter:
1. Define a small set of business-linked metrics aligned to your Spending Review priorities. Speed to service, decision quality, staff capacity released, productivity gains and citizen satisfaction. Instrument them now, not at programme close, and make sure you have captured the baseline before you start, so the before-and-after story is credible when it matters.
2. Stand up a structured impact storytelling rhythm. Staff experience, citizen outcome, role evolution, captured continuously. Three credible stories per quarter beat fifty anecdotes a year.Use ministerial communications to amplify your impact.
3. Make responsible AI evidence part of your standing reporting pack. ATRS, transparency assessments, ethical reviews. Built once, shown to ministers, audit and the public on demand.
REACH
At this stage: Hardwire and Habits
Hardwire
Build measurement and evidence into the operating model so impact is captured continuously, not retrofitted at review time.
Habits
Make impact storytelling, peer learning and leadership communication a standing part of how the organisation talks about itself.
What success looks like
Business-linked metrics
KPIs tied to departmental objectives and citizen outcomes — speed, decision quality, staff capacity, rework reduction. Not technical telemetry.
Evidenced impact stories
Structured narratives connecting AI to human outcomes: caseworkers helped, citizens served faster, decisions made more consistently. Stories supported by data.
Executive-ready dashboards
Real-time visibility into adoption, performance, risk and benefit realisation. Visible to leadership and updatable without a slide deck.
Responsible AI evidence
ATRS records up to date, responsible AI assessments documented, transparency obligations met. Required by regulation and expected by the public.
A people-centred story
Staff experience, citizen outcomes and role evolution captured alongside the metrics. The most compelling evidence of transformation is how roles become more valuable rather than redundant — and what the Future of Work Unit will increasingly want to see.
Practical actions for your teams
- Define adoption and value metrics together. Speed to service, decision quality, staff capacity released, citizen satisfaction. Measure adoption and impact, not one or the other.
- Build the evidence base as you deliver. Capture quantitative and qualitative data continuously, not at programme close. The story is easier to tell when the data is already there.
- Link to strategic value. Frame AI initiatives in the language of departmental objectives and Spending Review priorities, not technology categories.
- Document responsible AI evidence. ATRS records, transparency assessments, ethical reviews. Builds trust and meets emerging requirements at the same time.
- Track role evolution. How are jobs and skills changing? Often the most powerful demonstration of positive transformation.
- Share and celebrate. Internal communications, peer-to-peer storytelling, public-facing case studies. Make impact visible to the people who fund and use the service.