Delivery model assessments: the official guidance on make, buy or mix, explained

The key facts

• A delivery model assessment, previously called a make or buy decision, is an evidence based method for deciding whether to deliver a service in-house, buy it from the market, or use a hybrid model.
• The note builds on chapter 3 of the Sourcing Playbook and is consistent with the options appraisal approach prescribed in the Green Book.
• The process has eight steps and all projects go through all of them, though the depth of each step varies with the project.
• An initial assessment of mid-level complexity can take three to six months, and is ideally conducted during the Strategic Outline Case then iterated through the business lifecycle.
• Assessments are required for new public services, for significant new developments to an existing service, and where an existing delivery model needs re-evaluating, and are good practice for all projects.
• Options are wider than in-house or private market: arms length bodies, local authorities, the third sector, joint ventures and GovCos are all viable alternatives.
• A Should Cost Model is good practice for all procurements and shall be produced where a complex service is being considered for outsourcing.
• The five strategic and operational criteria are strategy, people and assets, delivery, market and suppliers, and risk, each weighted for the specific service.
• Reputational risk can never be outsourced, and where a service is insourced the full risk profile returns to the department.
• Where a service is being outsourced for the first time, there is a presumption that a pilot will be run before full implementation.

What a delivery model assessment is and when you need one

A delivery model assessment is an analytical, evidence based approach to reaching a recommendation on whether a department should deliver a service, or part of a service, in-house, procure it from the market, or adopt a hybrid solution. It was previously known as a make or buy decision. The guidance builds on chapter 3 of the Sourcing Playbook and is aimed at helping departments run a structured process rather than settling a strategic question by instinct.

The option set is broader than it first appears. Alongside departmental resources and the third party supply market, the guidance points to the wider public sector such as arms length bodies and local authorities, regulatory bodies, the third sector, and alternative commercial vehicles including joint ventures and GovCos. The note is equally clear about what an assessment is not: it is not a target operating model, not a budgeting mechanism, not a sourcing model assessment or supply strategy, and not a supplier evaluation. You are assessing mechanisms for delivery, not individual suppliers.

Assessments are required in three scenarios and are generally good practice for all projects: on the introduction of new public services, where a significant new development to an existing service such as a new technology requirement has been identified, and where there is a need to re-evaluate the delivery model of an existing service, for example because quality is deteriorating or there has been a major policy change, cost reduction or shift in strategic direction. It is also good practice to revisit the assessment periodically as markets and technology mature.

Timing and proportionality both matter. The assessment should be proportional to the criticality, complexity and size of the project, and an initial assessment of mid-level complexity can take three to six months depending on service complexity, data availability and resourcing. It is ideally conducted during the Strategic Outline Case and iterated as new data arrives. For complex projects, departments should consult the Cabinet Office before beginning, and the Sourcing Programme works with the Complex Transactions team who provide independent facilitation.

The eight steps, and getting the foundations right

The guidance sets out an eight step process. All projects go through all eight steps, but the time and depth required for each will differ, and elements of steps two to four are often run in parallel because you cannot understand data needs without understanding the evaluation criteria, and the Should Cost Model has to adapt as options evolve.

Step one is to frame the challenge. That means putting the right people in place: a dedicated core team of suitably skilled individuals drawn from finance, commercial, programme and operations, technical experts and central government teams such as the Infrastructure and Projects Authority. Complex and higher value projects need proportionally greater expertise and benefit from independent facilitation, because assessments can evoke strongly held views and a structured, independently facilitated process helps produce something all stakeholders can buy into. The core team should engage key stakeholders including end users, and settle governance and decision making from the outset.

Step two defines the service and the delivery options. A clear service definition comes before the detailed technical or outcome specification that will be produced later for tender, and it needs to carry enough information to work out whole life cost and to compare models against the criteria. Teams should consider which operational components must be delivered together, which are effectively out of scope because they must always be in-house such as policy creation, and where a problem statement approach might draw innovation from the market. Market engagement should shape the definition from the earliest stages, and market consultations must be transparent and non-discriminatory. A long list of delivery models is then shortlisted against critical success factors and practical limitations, and signed off by the programme board or senior responsible owner before evaluation begins.

Evaluation criteria, data and the Should Cost Model

Step three sets the strategic and operational evaluation criteria. The guidance offers five headings. Strategy asks whether the delivery model aligns with medium and long term organisational strategy, noting that once a service is contracted out it can be difficult and costly to revert to public provision, and that wider government requirements such as social value must be considered on both sides of the comparison. People and assets asks where capability and resources are best placed, with TUPE, asset transfer and pension liabilities needing legal and commercial advice. Delivery asks who is best placed to deliver the service definition and maintain continuity, including transition and exit planning. Market and suppliers asks whether a viable market exists or can be created, supported by comprehensive market analysis and benchmarking of costs and processes. Risk asks how overall risk can be minimised, and reminds teams that reputational risk can never be outsourced.

Criteria are weighted to reflect what matters for the specific service, and should be signed off by the sponsor and project board before evaluation begins. Sensitivity checks on weightings are worthwhile to ensure the assessment actually differentiates between options. The guidance also warns against too many criteria, which makes the process laborious, and against importing criteria used to assess tenders, since you are comparing delivery models, not bidders. A supporting table sets out service and market characteristics to factor in, including the nature of the service, clarity of service definition, degree of commoditisation, ease of disaggregation, intellectual property and data ownership, the need for transformation, maturity of technology, transition capability and availability of specialist expertise.

Data is a critical success factor. Key data inputs should be identified early, supported by a thorough external market health assessment, because the availability and quality of data drives both the design and the robustness of the Should Cost Model. Where independent sources do not exist, for example for novel services or first generation outsourcing, departments should get as close to objective as they can through structured interviews with potential suppliers, service customers or market analysts.

Step four builds the Should Cost Model to understand whole life cost of delivering or transforming the service. It is good practice for all procurements, and where a complex service is being considered for outsourcing a Should Cost Model shall be produced. The model should be revisited after the strategic and operational evaluation, since that evaluation often surfaces additional cost drivers.

Reaching a recommendation, and what happens next

Step five is the evaluation itself. Each potential delivery model is scored against the weighted criteria using a numerical scoring methodology with clear descriptions of what each score means, agreed with key stakeholders. The guidance lists the critical success factors: appropriate governance and sponsorship with suitably qualified individuals, good availability of input data, informed criteria supported by subject matter experts and independent facilitation, and realistic timelines. On timelines it is specific that published commercial pipelines should look ahead three to five years to be truly effective, with a minimum of 18 months, because complex outsourcing projects can need at least that long in preparation.

Step six aligns the cost and non-cost evaluations, bringing the scored criteria together with whole life cost for each option to reach a recommendation. Departments are encouraged to draw on colleagues across government and the private sector who have run similar programmes to sense check that the recommendation reflects real world experience. The potential outcomes are in-house delivery, a mixed make and buy model, outsourcing, or an alternative commercial vehicle such as a joint venture or GovCo, where expert advice should be sought before proceeding.

Step seven covers recommendations and approvals. Assessments and their underlying assumptions should be documented so outcomes can be audited or reviewed if circumstances change. Assessments are expected to be iterated in line with the Green Book business case process: the initial assessment informs the Strategic Outline Case, is reassessed ahead of the Outline Business Case, and its assumptions validated and carried into the Full Business Case. Once signed off, there should be a clear implementation plan, a plan for monitoring service delivery and quality, and a link to benefits realisation.

Step eight is piloting and implementation. Where a service is being outsourced for the first time there is a presumption that a pilot will be run before full implementation. Specifications should be tested with the market before being finalised, contract management mechanisms must be in place, and there must be plans for contract exit whether planned or unplanned, because government remains responsible for continuity of public services if a supplier fails. Crucially, in-house delivery is held to the same standard: the Project Delivery standards on objectives, business justification, governance, accountabilities, lessons learnt, planning and control, validated outcomes, capable teams and managed transition apply either way, and where similar services are both insourced and outsourced there is an expectation of comparable KPIs so the models can be fairly compared.

How eSourcing Data helps

A delivery model assessment lives or dies on evidence, and the guidance says so directly: assessments and their assumptions must be documented so they can be audited or reviewed as circumstances change, and they are expected to be iterated across the Strategic Outline Case, Outline Business Case and Full Business Case. That is a records problem as much as an analytical one. eSourcing Data gives teams one place to hold the service definition, the shortlist and its sign off, the weighted criteria and their sign off, the scoring, the supporting evidence and the final recommendation, so the version approved at each business case stage is retrievable rather than reconstructed.

Much of the input the guidance asks for comes from the market. Early and ongoing market engagement shapes the service definition, market health assessment feeds the criteria and the Should Cost Model, and consultations must be transparent and non-discriminatory. eSourcing Data supports structured market engagement with a timestamped record of who was invited, what was published, what was asked and what was answered, which both improves the evidence base and demonstrates that the engagement was run fairly. Where the recommendation is to go to market, the same platform carries the service definition through into specification, tender and evaluation without the thread being lost.

After the decision, the guidance expects monitoring, benefits realisation, contract management and exit planning, plus comparable KPIs where similar services are delivered both in-house and externally. Supplier and contract records, performance data and reporting across a portfolio make those comparisons possible instead of anecdotal. eSourcing Data does not make the delivery model decision or build your Should Cost Model, and it is no substitute for legal advice on TUPE, pensions or intellectual property. What it does is keep the analysis, the evidence and the approvals in one auditable place, so the next iteration starts from what you already know.

What to do about it

1. 1Start the assessment during the Strategic Outline Case and allow three to six months for a mid-level complexity service, contacting the Cabinet Office Sourcing Programme first if the project is complex.
2. 2Appoint a dedicated core team spanning finance, commercial, programme, operations and technical expertise, and consider independent facilitation to keep the assessment objective.
3. 3Produce a clear service definition before you start evaluating, identifying components that must stay in-house, components that must be bought, and where the service can be disaggregated.
4. 4Develop a long list of delivery models including hybrid and alternative vehicles, shortlist against critical success factors, and get the shortlist signed off by the programme board or SRO.
5. 5Weight the strategic and operational criteria for the specific service, run sensitivity checks on the weightings, and secure sponsor and project board sign off before evaluation begins.
6. 6Identify key data inputs early, commission a market health assessment, and build the Should Cost Model, revisiting it once the evaluation surfaces new cost drivers.
7. 7Document the assessment and its assumptions so it can be audited, and reassess it ahead of the Outline Business Case and Full Business Case.
8. 8Plan for implementation from the start: pilot a first generation outsource, put contract management and exit plans in place, and set comparable KPIs whichever model you choose.