What a good AI Opportunity Assessment actually delivers
Most 'AI strategy' ends in a slide deck. A real assessment ends in a costed, ranked roadmap you can build from — in days, not a quarter. Here is the difference, and what to walk away from.
July 2026 · 6 min read
Almost every enterprise has already had the AI strategy conversation. A firm came in, ran workshops, took hundreds of hours off the business, and left a beautifully formatted deck. And then, on the Monday, nobody could say what to actually do first. That is the failure mode a real assessment has to solve, and it is worth being precise about what “solved” looks like.
Why “AI strategy” stalls
A strategy engagement is usually paid to produce a point of view, so a point of view is what you get: themes, principles, a maturity score, a list of possibilities. All true, all reasonable, and none of it costed or sequenced. The gap between “here are things AI could do” and “fund these three this quarter” is the entire distance between a strategy and a plan, and it is where most AI programmes quietly lose a year.
The output test: a roadmap, not a report
There is a simple test for whether an assessment was any good: could a build team pick up the output on Monday and start? That rules out the static report immediately. What passes is a live, scored portfolio of opportunities you own — every one ranked by the value it creates against the cost and risk of delivering it, with a sequenced build order and an honest recommendation on what to fund, what to park, and what to walk away from. A deck tells you that you have opportunities. A roadmap tells you which three to build.
How opportunities get scored
The way we run it, the AI reads the raw material — your strategy, technology, data architecture, risk and governance policy, and org design — and drafts an AI Product Canvas for every opportunity it finds: the problem, the proposed solution, the value case, the data it needs, and how feasible it is to deliver. Each is then scored on value and feasibility so the portfolio ranks itself. Because the reading is automated, your people spend a couple of focused workshops on it rather than weeks of interviews, and the result lands in days.
From assessment to Build in days
The point of scoring is not analysis for its own sake; it is a build queue. Every top-ranked opportunity arrives with a canvas that a build team can pick up immediately, so there is no dead space between deciding and shipping. This is how the front door of an engagement is supposed to work: it produces the thing that makes the next stage fast, rather than a document that has to be re-interpreted before anyone can move. Downing used exactly this route to find where AI could grow the business without growing headcount, then trained their own people to start building it.
What to walk away from
If you are commissioning an assessment, three things should make you leave. One: it ends in a static report rather than something you can keep, re-run and build from. Two: it takes months, which usually means an army of graduates and a lot of your team’s time, for an answer you needed weeks ago. Three: it hands you ideas without a number attached, because an opportunity you cannot cost is an opinion, not a plan. A good assessment costs less than the wrong build, arrives in days, and tells you plainly what each thing is worth before you spend a penny building it.
Questions we hear next
Maturity and readiness assessments tell you how prepared you are, usually as a score. An opportunity assessment goes further: it tells you which specific things to build, in what order, and what each is worth. Readiness is useful context; the roadmap is the deliverable you can actually act on.