What an AI assessment should hand you (and what to refuse)

If you pay for an AI readiness assessment, these are the four things you should hold at the end, and the warning signs of one that will not help.

An assessment is easy to sell and easy to deliver badly. The poor ones produce a long slide deck that restates what you told the consultants. A good one produces decisions. Here is how to tell them apart before you sign.

Four things you should receive

  1. A scored view of where you stand: data, systems, people and governance, with named gaps and owners.
  2. A ranked list of opportunities, each with its value, its feasibility against your real data, and its risk.
  3. A roadmap with effort and cost ranges, in an order that accounts for dependencies.
  4. One first project scoped tightly enough that an engineering team could begin on Monday.

The fourth item is the test. If the assessment cannot name a first project and say how success will be measured, it has not finished its job.

Questions to ask the vendor

  • Who writes it? The people who would build the work should be in the room, not only strategists.
  • Will you look at our actual data? Feasibility judged from interviews alone is guesswork.
  • Can we take the output to another firm? If the answer is no, you are buying a sales process.
  • What happens if you conclude we are not ready? A vendor who has never said this has not been looking.

Warning signs

  • A promised accuracy or savings figure before anyone has seen your data.
  • A recommendation that happens to match the product the vendor resells.
  • Dozens of use cases with no ranking.
  • No mention of who in your organization will own the system once it is live.

How long it should take

For a defined set of business areas, two to four weeks is enough. Longer usually means the scope was too wide. An assessment that tries to cover the whole enterprise at once tends to describe everything and decide nothing.

An assessment is an informed judgment at a point in time. It should say so, and it should list what it did not look at.

Common questions

What should an AI readiness assessment include?

Readiness scoring for data, systems, people and governance; a ranked list of opportunities with value, feasibility and risk; a roadmap with effort and cost ranges; and a scoped first project.

How long should an AI assessment take?

Two to four weeks for a defined set of business areas. An assessment that tries to cover a whole enterprise at once tends to describe everything and decide nothing.

Can we take an AI assessment to a different vendor to build?

You should be able to. If the output only makes sense to the firm that wrote it, you have bought a sales process and not an assessment.

What if the assessment says we are not ready for AI?

That is a valid result. It should list what has to be true first, usually data access, ownership or a governance decision, so you can fix those before spending on a build.

Talk to us

Want this applied to your organization?

Tell us the business area you have in mind. We aim to reply within one business day with how we would approach it and whether an assessment fits.

Take the AI readiness check

We use these details only to reply to this request. No newsletters, no list. We aim to reply within one business day. See our Privacy Policy.

Book a callTalk to our team