An AI readiness assessment before you commit budget

What does an AI readiness assessment measure?

It measures five areas: data, systems, people, governance and current use. Each gets a score with evidence behind it, so the result is something leadership can act on rather than a general maturity label.

Data covers where key records live, how complete they are and whether anyone can get them out without a manual export. Systems covers whether HubSpot, Salesforce, NetSuite, QuickBooks, SharePoint or Google Drive expose usable APIs and permissions. People covers who is already using AI, who could own a project and how much training is needed. Governance covers policies, vendor terms and logging. Current use covers the tools staff are already paying for, officially or not.

• Data: completeness, structure, access and ownership.
• Systems: APIs, user permissions, integration points.
• People: skills, appetite, available owners.
• Governance: policy, vendor terms, audit trail.
• Current use: approved and unapproved AI tools.

How is an AI readiness assessment different from an AI audit?

A readiness assessment looks forward at what you can start; an AI audit looks at what is already running and whether it is controlled. Most companies need a bit of both, so our assessment includes a light audit of existing tools.

That audit part often surprises leadership. It is common to find a sales team pasting customer emails into a personal chatbot account, a finance analyst using an AI add-on in spreadsheets, and a marketing contractor generating content with no review step. None of this is necessarily wrong, but each one needs a decision: approve it on a business plan with proper data terms, replace it, or stop it.

Why check readiness before an AI pilot?

Because most failed pilots fail on data access and fit with daily work, not on model quality. MIT NANDA reported in July 2025 that only about 5% of custom enterprise AI tools reached production, citing brittle workflows and tools that did not fit how people work.

An assessment catches those problems while they are cheap. If invoice data sits in scanned PDFs across three shared drives, a finance automation needs an extraction step and a review queue before it can work. If the CRM has duplicate accounts and empty industry fields, a lead scoring idea needs cleanup first. Knowing that in week two costs far less than finding out in month four.

How do we run the assessment?

We combine interviews, a short staff survey, read-only system access and sample data checks. Nothing is changed in your systems during the review.

We usually interview department heads and a few hands-on staff, map the systems and data flows, then pull a sample from the records that matter most, for example a month of support tickets or a quarter of purchase orders. We test whether a model can actually work with that sample: can it extract the right fields, answer questions with citations, or classify records at an acceptable confidence level. Low-confidence results show where a human-review step will be needed in any future build.

We also look at volume and timing. A workflow that runs forty times a month rarely justifies a custom build, while one that runs four hundred times a week usually does. Those counts, taken from ticket queues, CRM activity logs or shared mailbox exports, turn a vague sense of busyness into figures you can put in a business case.

What do you get at the end of the assessment?

You get a scored report, a prioritized gap list and a short list of use cases that are ready now. The report is written for leadership and runs to a handful of pages, with the evidence in appendices.

Each gap is written as a task with an owner and an effort estimate, such as adding an API user to NetSuite, moving shared files into one SharePoint library, or approving a business-tier AI plan with data retention turned off. The ready-now list names one or two projects that could begin immediately, with a rough scope so you can decide whether to move straight into a build.

Who should commission a readiness review?

Owners, CEOs, COOs and finance leaders who have to approve AI spending and want an independent view first. It is especially useful when several departments are pushing their own tools, or when a customer or insurer has started asking how the company uses AI.

It also works as a first step before hiring a fractional Chief AI Officer, because it gives that person a baseline to measure against. For smaller firms, the assessment can be compact and focus on one or two departments, which keeps the cost proportionate.