Make, buy or mix: the decision that shapes everything else, and why it gets rushed
By the time a tender goes out, the most consequential decision has already been made and it usually was not the one anyone argued about. Whether a service is delivered in-house, bought from the market, or split between the two sets the ceiling on what any subsequent procurement can achieve. The government guidance on delivery model assessments exists because that decision is too often taken at speed, on thin evidence, by people with strongly held views. Its warning is unambiguous: insufficient oversight and assurance on this fundamental analysis, particularly where a project moves fast, creates significant operational, commercial and reputational risk.
Why this decision carries so much weight
Outsourcing done well frees management to focus on core priorities, leverages scale from the market, brings dynamism from a diverse supplier base, draws on expertise unavailable in the public sector and, with the right contractual flexibility, adapts to change. In-house delivery done well gives flexibility to react without being restricted by contract terms or procurement law, provides greater control over how a service is delivered, exploits internal synergies and stays aligned to core purpose. Both lists are real. Neither is a default.
What makes the choice hard is that it is close to irreversible in practice. The guidance notes that once delivery has been contracted to the private sector it can be difficult and costly to revert to public provision, whether on expiry, early termination or step in. And insourcing is no lighter a lift: it is a substantial transformation that brings the full risk profile back to the department, along with TUPE and pension exposure, backroom capability demands and the challenge of attracting specialist expertise within public sector pay.
There is also a market you affect by deciding. Exiting a market can push suppliers out of it, with knock on effects on other public services, which is why the guidance asks departments to assess that impact and engage early. The decision is never purely internal.
What goes wrong
The first failure is treating the assessment as a formality to clear before the real work starts. Three to six months for a mid-level complexity assessment sounds like a lot until you compare it with the length of the contract it determines. Complex outsourcing projects can need at least 18 months of preparation, which is why the guidance ties commercial pipelines to a three to five year look ahead.
The second is a weak service definition. If a service cannot be well defined with clear target outcomes, external providers will struggle to bid or will price in a risk premium. If a service has outcome based specifications but task level activity is poorly understood, insourcing gets hard for the mirror image reason. Either way, an unclear definition means the whole life cost comparison rests on sand.
The third is evaluation theatre: criteria chosen without weighting, weighted without sensitivity checks, or scored without agreed descriptions of what each score means. If the criteria do not differentiate between options, the process produces a number that ratifies a view someone already held. Independent facilitation exists precisely because these assessments provoke strong opinions.
The fourth is confusing this with other exercises. A delivery model assessment is not a target operating model, not a budget, not a supply strategy and not a supplier evaluation. Teams that let it drift into any of those end up with a document that answers a question nobody asked and leaves the make, buy or mix question open.
Running one that actually holds up
Get the people right first. A core team spanning finance, commercial, programme, operations and technical expertise, with independent facilitation on anything complex, and governance settled before analysis begins. Sign off the shortlist and the weighted criteria before evaluation, not after, so the process cannot be tuned to a preferred answer.
Then be honest about data. Identifying and gathering key data inputs is a critical success factor, and the quality of what you have determines how robust the Should Cost Model can be. For novel services there may be no independent market analysis at all, in which case structured interviews with potential suppliers, service users and market analysts are the next best thing. What you cannot do is score criteria on assertion and call it evidence.
Keep the option set genuinely open. Mixed models are often the right answer, since a service can be disaggregated so that, say, a technology platform and customer contact are delivered differently. Mixed models also demand extra care on where responsibilities lie, what rights each party has, what liabilities sit where, and how governance works across the parts.
Finally, hold in-house delivery to the same standard as anything you buy. The guidance is explicit that internal delivery should follow the same expectations set out in the Sourcing Playbook, with clear objectives and outcomes, proportionate governance, defined accountabilities and a focus on benefits realisation, and that comparable KPIs should be set where similar services are delivered both ways. Otherwise the comparison you make once at the outset can never be tested against what actually happened.
The takeaways
- The delivery model decision sets the ceiling for everything the procurement can achieve, so give it real time.
- The options are wider than in-house or outsource: hybrid, arms length bodies, the third sector, joint ventures and GovCos all count.
- No clear service definition means no reliable cost comparison and a risk premium in every bid.
- Sign off criteria and weightings before evaluation, and sensitivity check them so the result actually differentiates.
- Hold in-house delivery to the same standards and KPIs as an outsourced service, or you will never know whether the decision was right.
Want the full breakdown?
The complete explainer covers the key facts, the requirements in detail and a practical action list, free and printable in the Procurement Library.