Topic 31 of 64· Governmental performance and delivery metrics

Service Standards and Transaction Metrics

The GOV.UK Service Standard is the UK government's 14-point checklist for building and running a public digital service, and it comes paired with a small, mandatory set of quantitative transaction metrics — cost per transaction, completion rate, digital take-up, and user satisfaction — that teams must publish for every live central-government service. Together the standard and the metrics are the operational, day-to-day specialisation of the broader public value and KPI frameworks in this repository, aimed squarely at software delivery teams.

Why it matters

The Service Standard, maintained in GOV.UK's service manual, requires every point-in-time assessment (alpha, beta, live) of a government digital service to demonstrate — among its 14 points — that the team understands users' needs, works in a multidisciplinary team, iterates and improves frequently, and evaluates tools, systems, and ways of working. Historically this sat alongside a public Performance Platform where every live service published its transaction data openly; that platform has since been retired, but the underlying obligation to measure and publish these four core metrics persists through the service manual's "measuring success" guidance. The reason this differs from a generic software KPI dashboard is that these metrics were explicitly designed as one linked economic model, not four independent scores: the entire savings case for digital government — the Government Digital Service's Digital Efficiency Report found digital transactions roughly 20 times cheaper than by phone and around 50 times cheaper than face-to-face for comparable local-government services — only materialises if completion rate stays high and digital take-up genuinely rises, rather than merely adding a cheap channel alongside an unchanged expensive one.

The maths

Cost per transaction = total service running cost / number of completed transactions
Completion rate       = transactions completed / transactions started × 100
Digital take-up       = digital-channel transactions / all-channel transactions × 100
User satisfaction     = % satisfied + very satisfied, in-service 5-point survey

Channel-shift saving = transaction volume × take-up shift × (cost per transaction on
                        the old channel − cost per transaction digitally)

Failure-demand cost = (1 − completion rate) × transactions attempted digitally ×
                        cost of the fallback channel those users then use instead

Worked example

Illustrative central-government licence-renewal service, 2 million transactions/year, currently 65% phone (£3.00/transaction) and 35% digital (£0.30/transaction), completion rate 80%. A redesign against the 14-point Service Standard lifts digital take-up to 60% and completion to 92%:

Take-up shift saving = 2,000,000 × 0.25 × (3.00 − 0.30) = £1,350,000/year

Failure-demand cost, before:
  2,000,000 × 0.35 × (1 − 0.80) × £3.00 = £420,000/year (abandoners fall back to phone)

Failure-demand cost, after:
  2,000,000 × 0.60 × (1 − 0.92) × £3.00 = £288,000/year

Net failure-demand saving = £420,000 − £288,000 = £132,000/year

Total annual saving ≈ £1,350,000 + £132,000 = £1,482,000/year

The arithmetic makes explicit why completion rate is not a secondary metric: without the improvement from 80% to 92%, the take-up shift saving would be partly clawed back by failure demand routing frustrated digital users straight back to the expensive phone channel.

Software engineering connection

These four metrics are a working example of a cost-consequence dashboard: one cost metric held apart from three outcome/quality metrics, deliberately never collapsed into a single score — the same discipline argued for in public sector KPIs. For engineers this breaks into concrete, ownable work: completion rate is a funnel-instrumentation problem, and every abandonment point in the journey is, in principle, locatable and fixable; cost per transaction requires genuine unit-cost accounting including staff-assisted and paper-channel costs, not just cloud hosting spend (see cost per transaction and total cost of ownership in government IT); and digital take-up is an equity metric wearing an efficiency costume — the citizens who cannot or will not shift channel are disproportionately older, disabled, or digitally excluded, so aggressive channel closure converts a "saving" into an access harm (see digital inclusion and channel shift savings). The 14-point standard itself is the process specification behind these numbers — see digital service standard for the standard in full, and citizen satisfaction metrics for how the satisfaction figure here relates to broader trust measurement.

Pitfalls

  • Take-up gained by closing the alternative channel: shutting a phone line lifts the digital take-up percentage arithmetically while dumping failure demand onto whatever channel remains (often a more expensive assisted-digital or face-to-face route); always measure total-system cost, not the ratio alone.
  • Measuring completion rate from step two of the funnel: starting the "started" count after the first genuine drop-off point flatters the completion rate and hides the biggest fixable loss.
  • Cost per transaction excluding assisted-digital support: a digital-only unit cost that ignores the staff time spent helping users who can't self-serve understates the true cost of the channel.
  • Publishing metrics without a shared definition across services: "transaction" and "completed" mean different things across different service teams unless the definitions are standardised and versioned, making cross-service comparison unreliable.

Sources