Where Concept fits in CADMID
Concept is the first of CADMID's six phases (Concept, Assessment, Demonstration, Manufacture, In-Service, Disposal). It is the phase where the requirement is defined, candidate solutions are identified at a high level, and the Outline Business Case is developed for the Initial Gate / IPA Gate 1 decision. The phase ends with a decision — to enter Assessment with a manageable option set, to pause, or to terminate. The decision is made by the MoD Investment Approvals Committee and, on Major Projects, with IPA Gate 1 input.
For the controls function, Concept is small in volume but high in leverage. The estimates, ranges and risks set at Concept frame every decision that follows — and the cost estimate produced at Concept is, on the MoD's own data, the single best predictor of the eventual programme outturn. A Concept estimate built honestly survives Assessment; a Concept estimate engineered to look affordable is the root cause of the overrun that becomes visible at Main Gate two years later.
What controls must deliver at Concept
The deliverables at Concept are deliberately lightweight in scope but rigorous in methodology. The cost estimate is an AACE Class 5 — a strategic outline range with explicit uncertainty bounds, not a single-point figure. The expected accuracy at Class 5 is in the order of -50% to +100% on the low and high bounds, which is wide on purpose: at Concept, the solution is not yet defined and tighter ranges would be dishonest. Reviewers familiar with AACE classification will press on whether the range is genuinely consistent with concept-stage uncertainty.
The QRA at Concept is option-level. The job is not to produce a definitive P80 — there is no definitive cost to apply a P80 to. The job is to compare candidate options on a like-for-like uncertainty basis so that the option-set winnowing is rational. A bespoke airframe option and a Military Off-The-Shelf variant cannot rationally carry the same coefficient of variation; the QRA has to expose that difference, not flatten it. Reference-class comparison is the de-biasing technique that HM Treasury Green Book specifically calls for at this stage — every option range should be sense-checked against what comparable programmes have actually cost.
The risk register at Concept is v0.1: strategic risks at the option-set level, not the line-by-line operational risks that appear in later phases. Twenty to forty risks is typical; two hundred is a sign that the team has skipped Concept-level abstraction. Each risk should map to one or more candidate options and indicate which options would carry it forward into Assessment.
What IPA Gate 1 reviewers actually test
IPA Gate 1 — sometimes labelled "business justification" — tests whether the Outline Business Case is fit to enter Assessment. The reviewers are independent; they are not part of the programme team. They read the controls evidence with a specific question in mind: does the cost and time range honestly reflect the uncertainty of the concept, and has the option set been genuinely compared, or has one preferred option been engineered into looking best?
The recurring red flags are predictable. A cost range that is suspiciously close to the available funding envelope at the high end. An option-level QRA where the option-set uncertainties are flattened to similar coefficients of variation. An optimism-bias adjustment bolted on at the end of the analysis rather than built into the input ranges, which reviewers can strip out to expose the underlying numbers. A reference-class comparison that is absent or that uses obviously non-comparable programmes. A risk register whose top risks are not consistent with the option set being recommended.
The controls evidence that passes Gate 1 has three characteristics. The numbers are traceable — every figure in the OBC can be tied back to a model, a reference dataset, or an assumption that is stated explicitly. The methodology is conventional — AACE classification, Green Book optimism-bias, reference-class comparison — so reviewers do not spend their time relitigating methodology choices. The narrative is honest — the case explains where the team is genuinely confident, where they are not, and what work in Assessment is intended to resolve the uncertainty.
The recurring failure mode
The recurring failure at Concept is over-precision. The Class 5 range is presented as narrower than it really is because the team is uncomfortable showing a -50% / +100% spread to a board that wants reassurance. The narrower range is then defended as "more rigorous", which is the opposite of what AACE Class 5 actually means: at Class 5, rigour shows up as honest width, not artificial precision.
The second failure is over-elaboration. The team produces a Class 3 estimate dressed up as a Class 5 because the cost engineering effort runs ahead of the requirement definition. The estimate looks impressive but is technically untethered: the elements being costed reflect a level of solution definition that does not yet exist. Reviewers notice this — the cost breakdown structure is too detailed for the requirement description in the rest of the OBC — and the credibility of the whole estimate suffers.
The third failure is the missing reference class. The estimate is built from first principles without any check against what comparable programmes have cost in practice. Green Book is explicit that reference-class comparison is the most effective single de-biasing technique available; an OBC that omits it invites the reviewer to ask "what does this category of programme actually cost?", and the answer "we did not check" is not a good place to start.
What good looks like
A well-run Concept phase produces a small number of high-quality artefacts. A Class 5 cost range that is honestly wide and explicitly calibrated against reference-class data. An option-level QRA that differentiates uncertainty by technical maturity and gives the board a defensible basis for the option-set winnowing decision. A risk register v0.1 of around thirty strategic risks, each mapped to one or more options, with named owners. A narrative in the OBC that explains where the team is confident, where they are not, and what Assessment work is designed to resolve.
The narrative matters more at Concept than at any other phase. The reviewer is making a judgement about whether the programme should enter Assessment, which is the phase where serious money starts to be spent. The narrative is what convinces them that the team has been honest about uncertainty and has a plan for resolving it — not a series of confident assertions that the team intends to be right about.
How SOMA supports CADMID Concept
SOMA Project Controls works at Concept stage in two engagement shapes. The first is independent QRA on the option set — a short engagement that gives the programme team a reference-class-calibrated, AACE-compliant view of the cost and schedule confidence ranges per option, with a defensible methodology that will survive Gate 1 scrutiny. The second is OBC controls assurance — a structured pre-Gate review of the controls evidence pack, identifying the gaps that an IPA reviewer is most likely to flag and the remediation needed before submission. Both engagements are delivered by senior practitioners who have built controls evidence packs that have passed Gate 1 on real UK defence and government programmes.