Legacy IT is a procurement choice: what the DDaT Playbook is really about
Every legacy system in government was once a shiny new procurement. That is the uncomfortable insight running through the Digital, Data and Technology Playbook: technical debt is rarely created by technologists, it is created at contract signature, when nobody priced the exit, pinned support commitments or planned for the day the contract ends. Version 2 of the playbook, refreshed in June 2023, reads as an attempt to make an entire sector buy technology like something that has to be maintained, secured and eventually replaced.
The definition that changes the conversation
The playbook defines legacy IT precisely: systems and their component software and hardware outside vendor support, on extended support or on bespoke support arrangements. Notice that this is a contractual condition, not a technical one. A system becomes legacy the day its support arrangements decay, which means the countermeasures are contractual too: requiring software to stay in mainstream support for the duration of the contract, and planning early for contract end, including extensions.
That is why exit planning gets its own chapter at the far end of the lifecycle. Transition, data exchange and extension decisions made in a hurry are how authorities end up trapped on bespoke support, paying more for less security. The playbook's answer is to make the end of the contract a design input at the start.
Eleven policies, one theme: do the thinking early
Pipelines published at least 18 months ahead, market health assessments, delivery model assessments with Should Cost Models, cyber assessment during selection. The pattern across the key policies is that the expensive mistakes are made before procurement formally begins, and the playbook keeps pushing effort upstream. Even the resolution planning requirement for critical suppliers is early thinking: preparing for a supplier insolvency you hope never happens.
The same logic explains the mixed delivery model the playbook champions, with government as service integrator, in-house ownership of products, and market expertise supplementing agile teams. That model only works if authorities genuinely assess delivery options rather than defaulting to the big outsourced programme, which is precisely what the delivery model assessment policy exists to force.
AI procurement grows up
The refreshed AI content deserves attention because it cuts through vendor fog with one question: what exactly are you buying? The model, the algorithm or the data set. Each carries different intellectual property, value and management implications, and a buyer who cannot answer the question cannot evaluate the deal. The playbook also warns that AI increasingly hides inside services that are not sold as AI, so the burden is on market engagement to surface it, along with the risks it imports.
Generative AI gets a characteristically sober treatment: capable of very plausible content, driven by probability, subject to bias from its training. For contracting authorities, this is a reminder that the question is not whether AI is impressive but whether it is the right solution, and government guidance now exists specifically to test that.
What suppliers, especially SMEs, should take from it
The playbook is a buyer document, but it is also a map of what government will ask of its suppliers: open and interoperable data, APIs to CDDO standards, software kept in mainstream support, cyber expectations under the Cyber Assessment Framework, resolution planning for critical contracts, and sustainability obligations. Suppliers who arrive with these answers prepared are pushing at an open door.
For SMEs the direction of travel is favourable. The playbook commits to levelling the playing field, maps how cloud layers create SME routes to market, and flags procurement reform measures that remove blockers like mandatory audited accounts and pre-award insurance. The demand signal is there too, in the requirement for pipelines published well in advance. The practical move for smaller suppliers is to watch those pipelines and engage early, because early engagement is where this playbook says the real decisions happen.
The takeaways
- Legacy IT is defined by support status, so prevention is contractual: mainstream support for the term and exit planned from day one.
- The playbook pushes effort upstream: pipelines, market health checks and delivery model assessments before procurement begins.
- In AI procurement, always establish whether you are buying the model, the algorithm or the data set.
- Comply or explain applies to central departments and ALBs, with controls biting above 20 million per transaction.
- Suppliers who pre-answer the playbook's asks on interoperability, security and sustainability gain a real edge.
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.