Method and resources

How relative fit is calculated

Each of the 21 requirements receives a score from 0 to 10. Each model family has an illustrative reference profile on the same scales. For each family, take the absolute difference between your score and its profile on each question, then add the 21 differences.

Relative fit (%) = 100 × (1 − total absolute difference ÷ 210).

A total difference of 42 gives 80% relative fit. Identical profiles give 100%. A five-point difference counts the same wherever it occurs. The original calculation gives every criterion equal weight.

The tool ranks the profiles by relative fit and reports ties. It flags individual gaps greater than five points, as the spreadsheet does, and also lets you inspect all differences. A five-point gap can still matter, as can a smaller gap on a condition that cannot be compromised.

All 21 scores must be present before a comparison is calculated. Missing answers aren't zero. The calculation doesn't invent an in-house profile, and it doesn't average the answers into a single measure of autonomy.

What the percentages cannot establish

These are indicative profiles developed for design conversations. They aren't calibrated from a representative dataset of successful and failed services. A fit percentage is neither a success probability nor evidence that a model is legally available, affordable, or desirable.

An aggregate permits one gap to be offset by others. Keep requirements that cannot be traded away outside that arithmetic. A proposed arrangement which cannot meet an essential condition needs redesign or rejection, whatever its overall fit.

Reading the scales

The six control questions separate grip, accountability, ownership, responsibility, authority, and financial control. Competition asks about competition for the work. Contestability asks whether there remains a credible way to challenge or replace provision.

Higher scores for sponsor involvement mean less involvement. Higher scores for community, service-user, and employee involvement mean more involvement. Commercial and workforce questions concern desired freedoms or priorities. There is no consistent ‘higher is better’ direction.

Public edition

This edition retains the 21 criteria, 11 numerical profiles, equal weights, absolute-distance calculation, and gap threshold from the developed spreadsheet tool. Wording follows the later delivery leverage and control guidance. It clarifies inconsistent older scale labels, uses 21 throughout, and makes the absence of a scored in-house option explicit.

The web edition adds a check of one-point sensitivity, downloads, and a decision record. Those features don't change the core fit calculation.

Capability and feasibility

The question isn't only whether a model suits your intentions. It is whether the people and institutions involved can make it work.

• Commissioning and policy: can you understand the wider system, define worthwhile outcomes, establish governance, and sustain relationships?
• Back office and support: can administration, finance, people support, data, and other enabling functions work in the proposed arrangement?
• Delivery body: is there evidence of delivery and self-governing capability, or a credible route to develop it? Some models don't require a distinct body.
• Market or community: do the necessary participants have the willingness, capability, resources, and legitimacy to sustain provision?

Separation can demand more from the commissioning side while changing the support it must provide. Giving up daily control doesn't remove the need to understand what is happening.

Questions for the decision and transition

• What is the purpose, for whom, and how will we know whether the arrangement serves it?
• Which decisions, resources, and responsibilities need to move together? Where would authority and accountability become detached?
• Which parts need a different arrangement, and what joins them back into a coherent service?
• What does a real option cost, including transition, retained commissioning, support, and the cost of failure?
• Which obligations and permissions apply to this actual proposal? Who has checked them, and when?
• Who owns the assets, data, and knowledge needed to sustain or recover provision?
• Who can challenge performance, through what information, and with what consequence?
• If we intervene, can we provide the people and capability to act? What happens to public confidence, and to the provider’s own responsibility, if we intervene too readily?
• What happens if a partner, key employee, funder, or supplier withdraws?
• What will trigger review, and which choices will still be reversible then?

Keep the assumptions and reasons with the decision. They let the next review ask whether the world changed, whether the original judgement was wrong, or whether the arrangement was never implemented as intended.

Read and watch

• The Sourcing Playbook

  Official guidance on assessing delivery options, testing markets, allocating risk, and planning service continuity. Its delivery model assessment considers direct provision, external provision, and hybrid arrangements.

• Delivery model assessments

  A more detailed official account of the assessment process. Use it to develop an evidence-based appraisal of actual options beyond these reference profiles.

• Arm’s-length body sponsorship code of good practice

  Standards for the relationship between a public body and its sponsor, including purpose, assurance, governance, and capability.

• Partnerships with arm’s-length bodies

  The earlier code places the relationship alongside formal oversight, using purpose, assurance, value, and engagement as organising principles.

• Autonomy, accountability, and ambiguity in arm’s-length meta-governance

  Jonathan Hammond, Ewen Speed, Pauline Allen, Imelda McDermott, Anna Coleman, and Kath Checkland examine how formal separation and practical control can diverge. A useful basis for asking who governs the governance. Accessible institutional full text.

• Future commissioning and contestability

  Professor Gary Sturgess’s video presentations, hosted by the Public Service Transformation Academy. Competition, contestability, and commissioning deserve separate attention.

• Contract design: a toolbox approach

  A presentation linked from the underlying delivery-model materials: ways of thinking about what a contract is trying to achieve.

• Consequences regime

  A companion presentation on what follows from performance or non-performance.

The official guidance describes its own scope and applicability. The numerical profiles here are a separate design tool; these sources do not validate its percentages.

Inspect or download the profiles

All 231 model scores are available below and as a CSV file. Columns follow the question order. The question sheet includes the endpoints and space for your own requirements and reasons.

On a small screen, scroll the table sideways. The profile scores describe a broad family; an actual arrangement may differ substantially.