How GBS Procure manages four housing clients in eSourcing OS.

Why traditional procurement case studies are weak evidence

The standard procurement case study has a familiar shape. A problem is described in general terms, a solution is introduced, and a headline saving or time reduction appears in bold. What is almost never present is the thing that would make the claim testable: a defined baseline, a stated measurement method, a scope boundary, and an account of what else changed during the same period. Without those, the number is a marketing artefact rather than evidence, and experienced finance and audit colleagues treat it accordingly.

There are structural reasons for this. A cost avoidance against a supplier's opening price is not the same as a cash releasing saving a budget holder can hand back, and a cycle time reduction on one straightforward competition tells you nothing about a complex works procurement. Selection bias does the rest, because nobody publishes the implementation that stalled or the contract that was renegotiated quietly a year later.

This matters more since the Procurement Act 2023 came into force on 24 February 2025. The regime puts far greater weight on published, structured information about how contracts are awarded and how they perform. That creates an opportunity: instead of relying on anecdote, organisations can build outcome evidence from the same data they are already obliged to publish and record. The discipline of measurement moves from the marketing department to the commercial function, which is where it belongs.

• Ask what the baseline was and who calculated it
• Ask whether the saving is cash releasing or cost avoidance
• Ask what the scope boundary was and what was excluded
• Ask what else changed in the same period, including staffing
• Ask whether the result was still true twelve months later

Baselining before you measure anything

A baseline is simply an agreed picture of the current position, captured before you change anything, using a method you write down. It is unglamorous work and it is routinely skipped, which is why so many improvement claims collapse under scrutiny. The practical minimum is a defined period, usually a full financial year, a defined set of procurements, and a defined set of measures with the rules for calculating each one recorded in plain language so that a successor can reproduce them.

Baselines should capture effort as well as cost. Ask how many procurements were run, of what type and value, how many people touched each one, how long each stage took in elapsed days, how many clarifications were issued, and how much rework occurred. Much of this can be reconstructed from existing records, but expect gaps. Where data is missing, a structured estimate agreed with the people who did the work is better than pretending the gap does not exist.

Be honest about baseline quality. Grade each measure as reliable, partly reliable or indicative, and carry that grading forward into every report you produce. Audit committees respond much better to a modest claim with clear provenance than to an impressive claim with none. It also protects you later, because you will be asked whether an improvement is real or an artefact of better recording.

Admin effort and cycle time from notice to award

Effort per procurement is one of the most useful measures available to a public buyer, and one of the least used. It answers the question a chief executive actually asks, which is not what the software costs but how much of the organisation's scarce professional capacity is consumed by process. Measure it as officer hours per procurement, split by role, and segment by procurement type so that a low value quotation exercise is never averaged together with a complex services competition.

The hours worth tracking cluster in predictable places: preparing and assembling documents, handling supplier clarifications, chasing internal approvals, collating and moderating evaluation scores, preparing award paperwork and notices, and answering internal queries after award. Interpret changes carefully, because a fall in officer hours is only a good outcome if quality holds and risk does not migrate into service departments doing shadow work in spreadsheets. Pair effort with a quality measure such as clarification volume or rework rate.

Cycle time is the elapsed period between defined milestones, and it is only meaningful if the milestones are defined consistently. The most common error is measuring from the day the notice is published, which conveniently excludes the months of internal preparation that often dominate the real timeline. A more honest measure starts at the point a requirement is formally approved to proceed, and runs through to contract signature, with intermediate stages recorded so you can see where time is actually lost.

Under the Procurement Act 2023 the milestone structure is clearer, because the regime relies on a defined sequence of notices across the commercial lifecycle. That makes stage timings easier to capture consistently and easier to compare between similar procurements. Report cycle time as a distribution rather than an average, because a median with an upper quartile tells a far more useful story than a mean that one outlier procurement has dragged upwards. Statutory periods such as standstill are fixed; internal approval queues are not.

Compliance and challenge risk under the Procurement Act 2023

Compliance outcomes are harder to celebrate than savings because success looks like nothing happening. That does not make them less important. Under the current regime the obligations that most commonly create exposure are transparency notices across the lifecycle, the requirement to give unsuccessful suppliers proper assessment information, and the correct operation of the standstill period before a contract is entered into. Each of these can be measured, and each produces a record that either exists at the right time or does not.

Sensible compliance measures are simple counts and timings: the proportion of notices published within the required period, the proportion of assessment summaries issued to all relevant suppliers at the right point, standstill periods correctly calculated and observed, and conflict of interest declarations completed before evaluation rather than after. Add a measure of record completeness, because the practical test in a challenge is whether you can produce a coherent, contemporaneous audit trail showing how a decision was reached.

Challenge risk itself should be tracked as a small set of leading indicators rather than a single lagging count of legal claims, which will thankfully be near zero in most organisations. Useful indicators include the volume and nature of supplier debrief queries, the number of evaluation scores changed at moderation without recorded justification, and the number of awards proceeding with documented deviations from the published methodology. These are where problems announce themselves early.

• Notices published within required timescales
• Assessment summaries issued correctly before standstill
• Standstill correctly calculated and observed
• Conflicts declared before evaluation begins
• Complete, contemporaneous decision records retained

SME and voluntary sector participation

Participation is a policy priority and a legitimate outcome measure in its own right, but it is easy to measure badly. Counting the proportion of contracts awarded to smaller organisations is a weak indicator on its own, because it is driven as much by what you happen to buy as by how you buy it. A more informative approach measures the funnel: who saw the opportunity, who started an application, who completed it, who was excluded and at what stage, and who was successful.

Drop off points are where the insight sits. If small and voluntary sector organisations consistently abandon at the selection stage, the likely causes are disproportionate turnover or insurance requirements, evidence demands that duplicate information already held, or documents that assume a bid team the organisation does not have. If they complete but score poorly, the issue is more often question design and word limits that reward professional bid writing over operational capability. Track the practical enablers too, because they are within your gift: lot structuring and contract sizing, payment terms, publication lead times, and the volume of duplicated assurance information requested.

Social value delivery and post award monitoring

Social value is where the gap between promised and delivered outcomes is widest, largely because the effort concentrates at evaluation and evaporates after award. A commitment scored at tender stage is a forecast, not a result. Treating it as a measurable outcome means carrying each commitment through into the contract as a specific, owned, dated obligation, then monitoring delivery on the same cycle as any other performance measure, with a named contract manager responsible.

Measure delivery against commitment rather than against a general policy aspiration. For each contract, record what was promised, in what units, by when, and what was actually delivered, verified by something more robust than a supplier self declaration in a slide pack. Aggregate that upwards only after it has been checked at contract level. An organisational social value total assembled from unverified returns is precisely the kind of number that damages credibility when a scrutiny committee starts pulling at it.

Expect variance and plan for it. The useful outcome measure is not perfect delivery but the proportion of commitments delivered, actively renegotiated with a recorded rationale, or formally written off.

Contract management outcomes after award

Most of the value in public procurement is won or lost after signature, and most measurement stops there. The practical starting point is coverage rather than sophistication: what proportion of contracts by value have a named manager, a current record with the correct dates, agreed performance measures, and a review that actually took place in the last period. Organisations that measure this honestly for the first time are usually surprised, and that surprise is itself a useful outcome to report.

From coverage, move to performance and risk. Useful measures include the proportion of contracts meeting agreed performance levels, the volume and value of variations, the proportion of contracts renewed or extended without a documented options appraisal, and the number of expiries approaching with insufficient lead time to run a proper competition. That last measure is one of the strongest predictors of poor value, because time pressure narrows options and weakens negotiating position.

Link contract management measures back to the procurement stage. A recurring pattern of variations in a category usually points to specification quality rather than supplier failure, and repeated emergency extensions point to pipeline management.

Building a benefits case an audit committee will accept

A benefits case is a short document with a long memory. It should state the baseline and its date, the measures and their calculation rules, the expected direction and rough magnitude of change, the owner of each measure, and the review points at which the claims will be tested. Write it before implementation, not after. A benefits case assembled retrospectively will always look like a justification, because that is what it is.

Separate the categories of benefit and never blend them. Cash releasing savings are removed from a budget. Cost avoidance reduces a future increase. Capacity released is officer time that can be redeployed, which is real but not the same as money unless a post is removed or growth is absorbed without recruitment. Risk reduction and compliance improvement are usually best expressed qualitatively with supporting counts, rather than converted into a currency figure through an assumption chain nobody believes.

Anticipate the audit questions. Where did the data come from, can it be reproduced, who validated it, and what happened to the benefits that did not materialise. A candid section on unrealised benefits is far more persuasive than uniform success, and it makes the next case easier to approve.

Sector specifics: councils, NHS, housing, education, charities and utilities

The measures are broadly common across sectors but the emphasis and the audience differ. In local government, outcomes are reported to scrutiny and audit committees and to elected members with local economic priorities, so SME participation, social value delivery and contract coverage often carry as much weight as unit price. In the NHS, clinical engagement, standardisation across sites and the interaction with national and regional buying arrangements dominate, and cycle time is frequently constrained by internal governance rather than procurement process.

Housing associations sit in a distinctive position, regulated and often subject to procurement rules while operating commercially, with heavy repairs, maintenance and development spend where contract management performance affects tenants directly. Education buyers, whether a trust central team or a single institution, tend to be capacity constrained, which makes effort per procurement and template reuse the measures that matter most. Charities appear on both sides, buying under funder conditions while bidding for public contracts.

Utilities operate under a different set of rules again. They may run utilities dynamic markets, may charge suppliers membership and award fees, and may run competitions limited to members, none of which is available to a general contracting authority. For utilities, market membership health becomes a genuine outcome measure: application processing times, the range of members by category, and whether member only competitions are attracting sufficient credible responses.

Where Dynamic Markets and G-Cloud 15 change the measurement picture

The Procurement Act 2023 replaced the Dynamic Purchasing System with Dynamic Markets, and the design differences have direct consequences for what you can measure and claim. A Dynamic Market is permanently open, membership cannot be capped, applications must be assessed within a reasonable time, and pending applications must be considered before a competition concludes. That turns market administration into an ongoing service with its own outcome measures rather than a one off setup exercise.

Sensible measures for a Dynamic Market include time from application to decision, the proportion of applications decided within your published service standard, the number and mix of members by category and organisation size, the proportion of members that have actually bid, and the number of competitions run from the market rather than from scratch. That last measure connects the market directly to cycle time and officer effort, which is where the operational benefit case usually lives.

For buyers looking at software rather than services, eSourcing Data is available to public buyers through RM1557.15 G-Cloud 15, with software services listed on the Digital Marketplace alongside cloud support services, and purchases made as call off contracts. The service directory sets out what is covered across sourcing, assurance, evaluation, contract management, governance and reporting.