# AI value discovery > Decide where AI is worth pursuing before anyone builds anything. A short engagement that ends with a prioritised portfolio of opportunities, each with a value hypothesis, a measure and a recommendation to proceed, reshape or stop. Source: https://level9.net/ai-consulting/value-discovery Publisher: Level9 Networks (https://level9.net) Updated: 2026-11-03 **Who it is for.** COOs, CIOs and business-unit leaders who have budget pressure to "do something with AI" and want to spend it on the right thing. **The problem.** Most organisations have a list of AI ideas and no reliable way to rank them. The list is usually technology-led, so the first project is the easiest demo rather than the most valuable change. **Target outcome.** A ranked set of opportunities that leadership can fund with confidence, each tied to a business measure and a baseline. ## Measures - Number of opportunities with an agreed measure, baseline and owner - Estimated value and cost range per opportunity - A clear proceed, reshape or stop decision for each ## Inputs - Access to the people who run the workflows under review - Existing performance data, even if partial - Known constraints on data, security, procurement and regulation ## Outputs - Prioritised opportunity portfolio - Value hypotheses with measures and baselines - Delivery options, dependencies and risks for each - Recommendation to proceed, reshape or stop ## Risks - Interviews reveal that the measure nobody owns is the one that matters - Data needed for the best opportunity does not exist yet - Stakeholders expect a tool recommendation rather than a decision ## When this engagement fits Value discovery is the right first step when you have more ideas than evidence. It fits when a leadership team wants an AI programme but does not yet agree on what it is for, or when an earlier pilot produced a demo and no decision. It is the wrong step when you already have one well-defined problem with an owner and a measure. In that case go straight to a [proof of value](/ai-consulting/proof-of-value). ## What we do We interview the people who do the work and the people accountable for its results. We map the workflow as it actually runs, including the exceptions and the workarounds. For each candidate opportunity we write down the decision or step that would change, the measure that would move, and the baseline today. We then review data readiness, integration dependencies, security and regulatory constraints, and the operating change each opportunity would require. The result is a ranking that reflects value, cost, risk and time to evidence. ## What you receive A short written portfolio. Each entry states the value hypothesis in one sentence, the measure and baseline, the delivery options, the dependencies, the risks and a recommendation. The recommendation is one of three words: proceed, reshape or stop. ## Decision gate Discovery ends with a leadership session. For each opportunity you decide whether to fund a proof of value, change the scope, or close it. Nothing proceeds without a named owner for the measure. ## Questions people ask **How long does value discovery take?** Typically three to six weeks, depending on how many workflows are in scope and how quickly we can reach the people who run them. **Do we need clean data first?** No. Part of discovery is finding out what data exists and how usable it is. Data readiness is an input to the ranking, not a precondition. **Will you recommend specific AI products?** Only where a product choice changes the economics of an opportunity. The output is a decision about where value is, not a vendor shortlist. --- Level9 Networks is an authorised UK partner of Radiator Software Oy. Radiator is a trademark of Radiator Software Oy. Level9 Networks Limited, United Kingdom.