Data Readiness Checklist for AI

AI works from the material you give it. If that material is locked in someone's inbox, inconsistent across spreadsheets, missing key fields, or not allowed to be used, the output will be unreliable no matter how good the tool is. Data readiness is simply whether the source material behind a workflow is good enough for AI to help.

The point of a data readiness checklist is not to clean every record in the business. It is to confirm that the data behind one specific workflow is ready, so your first AI project is not undermined before it starts. For the broader picture, see our business AI readiness checklist.

1. Access: can the people running the workflow actually get to the data without chasing it across personal devices, locked accounts, or one staff member's memory? 2. Quality: is it reasonably consistent — same formats, few duplicates, not riddled with typos or stale entries? 3. Completeness: are the fields the workflow depends on actually filled in most of the time? 4. Structure: is it organized enough — labeled columns, named documents, consistent records — that a tool or person can find what they need? 5. Permission: are you allowed to use this data for AI, given customer expectations, contracts, and privacy rules?

Score each from 1 to 5 for the workflow you have in mind. A workflow where all five score 3 or higher has data ready enough to start. A single low score tells you precisely what to fix first.

You do not need analysts to run a data readiness assessment. Open the actual source — the spreadsheet, the document folder, the CRM export — and look at twenty real records. Count how many are missing the fields you care about, how many are duplicates, and how many are obviously out of date. That rough error rate tells you most of what you need.

If twenty records are clean and consistent, the data is probably fine for a first project. If half are missing key fields or formatted three different ways, fix that before involving AI — the tool will faithfully amplify the mess.

Tackle access and permission before quality. There is no point cleaning data you are not allowed to use, and no point improving quality on a source nobody can reach. Get the data into one accessible place with clear permission to use it, then improve consistency and fill the critical gaps.

Keep the scope narrow. You are preparing the data behind one workflow, not running a company-wide data project. A short cleanup of the fields that one AI use case depends on is usually enough to unblock a pilot.

Some AI use cases need very little data to start. Drafting emails, summarizing meetings, or writing first-draft copy mostly need a few good examples of your tone. These are excellent first projects when your structured data is not ready yet.

Use cases like internal knowledge search, support automation, or data cleanup depend more heavily on accessible, organized source material. If those are your goal, the data readiness checklist matters more. Our small business AI use cases guide maps which use cases need solid data and which do not.

Pick the workflow you most want AI to help with. Run the five-point data check on just that workflow. If it scores well, move on to the AI implementation checklist and start a pilot. If it does not, your data readiness plan is simply the one or two low-scoring areas — usually access or quality.

For a quick overall read on where your business stands, the free AI readiness quiz factors data accessibility into its score in about two minutes.