Independent assurance that the business case survives delivery — for ERP, AI and major digital programs. From approval, through the decisions that quietly erode value, to benefits actually banked.
of expected digital value is actually captured
McKinsey, 2024of large projects land on budget, on time and with the promised benefits
Flyvbjerg, 16,000+ projects, 2023of recent ERP initiatives will miss their original business case by 2027
Gartnerof companies have achieved AI value at scale
BCG, 2025Every major program has a PMO governing time, cost and scope, a QA function checking the solution works, and audit checking the controls. None of them owns the question the money was approved on — and the one closest to it reports to the person delivering. Value assurance sits outside the program and answers to the sponsor.
Does the investment still hold against the strategy, the market and its own business case — tested continuously, not once at approval.
Every design, scope, sourcing and sequencing decision traced back to the value drivers that justified the spend — and priced before it is committed.
Adoption, ownership and measurement built in, so benefits keep arriving after the program closes and the team disperses.
Programs report cost burn-down. Almost none report value burn-up beside it. The gap between the two curves is the earliest warning signal available — and the only one that translates directly into a decision.
We turn that gap into a live register: every threat to the business case, quantified, with an owner, a mitigation and a decision date. Not a RAG colour. A number.
Illustrative. On a live engagement both curves are fed by actual operational and financial data.
A value tree where every benefit carries a metric, a data source, a measured baseline, a named owner and a date. Benefits that cannot survive this are removed from the case.
Ranges not point estimates. Reference-class forecasting, explicit disbenefits, whole-of-life run costs, and a confidence rating on every line.
Independent, evidence-based recommendations to the sponsor at each gate — proceed, proceed with conditions, pause or stop.
A live value ledger fed by real operational data — process mining, transaction and asset data — not self-assessment.
Every material trade-off tested against the value tree and priced in dollars before it is committed, not explained afterwards.
Post-go-live verification against baseline with a stated counterfactual, formal handover of benefit ownership, and findings fed into the next investment.
ERP cases are built on process-level claims — faster close, lower cost-to-serve, fewer manual touches, better working capital. ERP programs are then governed at module and milestone level. The measurement never meets the promise.
AI breaks the assumptions most value frameworks rest on. Output is probabilistic. Unit economics change with usage. And the failure mode is not that the pilot fails — it is that the pilot works and nothing changes.
The discipline does not change with the technology. Only the value drivers, the data source for the baseline, and the obligations you are assured against.
The delivery contract for the program we assure sits with someone else. That separation is the entire point: our finding stands free of any commercial consequence attached to it.
The independent review of a halted $2.8bn Commonwealth program found that letting the program select its own independent assurer "may limit the independent assurer's impartiality." Structural independence is not a nicety.
A value assurance function typically runs at a low single-digit percentage of program cost, and can be structured with a risk-shared component tied to assured outcomes rather than days consumed.
These are engagement commitments. Program-specific results are provided as references under NDA.
A rapid independent read on a program already underway. Value tree reconstructed, benefits tested, leaks found, Value at Risk quantified.
Start hereBefore the money is committed. Ranges, reference-class forecasting, disbenefits, run costs, confidence ratings, owners and baselines.
EnquireA small independent team running gates, the value ledger, decision assurance and Value at Risk reporting through to realisation.
EnquirePost go-live. What was actually delivered, measured against baseline with a stated counterfactual — and a recovery plan for the gap.
EnquireProject assurance asks whether the program is being delivered properly. Value assurance asks whether what is being delivered will produce the value that justified the investment — and keeps asking after go-live, when most benefits actually arrive.
Benefits realisation management is a core component of it. Value assurance adds independence, stage-gate decision rigour, value testing at the level of individual delivery decisions, and challenge of the business case itself before it is approved.
Before the business case is approved — the cheapest value ever recovered is the value recovered before commitment. For a program already underway, the highest-return entry point is a diagnostic ahead of the next major gate.
It front-loads decisions rather than adding them. Programs with agreed value criteria decide faster, because the basis for deciding is already settled. The slow programs are the ones relitigating scope in month fourteen.
Yes, and it should. We align to your mandated gates rather than duplicating them, and strengthen the value evidence those reviewers ultimately assess.
Particularly to AI programs. The research is consistent that AI value is determined mainly by workflow redesign, executive ownership and measurement discipline rather than model choice — and those are precisely the things value assurance governs.
A diagnostic takes two to four weeks and tells you three things: whether the business case still stands, where value is leaking now, and what it will cost to stop it.