Almost every AI project that disappoints its sponsors fails for the same handful of reasons, and none of them are "the AI wasn't good enough." They fail because of foundations that were never checked before the tool was bought. This checklist is the one we run through, informally, on every engagement — before any conversation about which product to use.
1. Do you actually have the data, and is it trustworthy?
AI needs consistent, reasonably clean input to produce reliable output. Before adopting any tool, ask: where does the relevant data live, who owns it, and has anyone checked it for accuracy recently? We regularly find businesses eager to automate a process built on a spreadsheet that hasn't been fully reconciled in over a year. Fixing that spreadsheet is unglamorous, and it's also the actual prerequisite — no AI layer corrects for input nobody trusts.
2. Is the underlying process actually documented?
You can't automate or augment a process that only exists in one person's head. If the honest answer to "how does this actually work end to end" is "ask Sarah," that's not an AI-readiness gap, it's a documentation gap, and it needs to close first. This is the same finding that surfaces in almost every process efficiency audit we run — the map exists nowhere until someone is asked to draw it.
3. Does the process run often enough to justify the investment?
As we cover in more detail in our piece on where AI actually saves time, volume and repetition are what make automation pay back. A task performed twice a month is rarely worth building an AI-assisted workflow around, however tedious it feels in the moment.
4. Who is accountable when the AI gets it wrong?
Every AI tool, however well-configured, will occasionally produce an inaccurate or inappropriate output. Before adoption, there should be a clear, human answer to: who reviews outputs, how often, and what happens when something's wrong? Businesses that skip this step tend to discover the answer reactively, usually after a client or regulator asks the same question first.
5. Do your people have time and permission to change how they work?
The most underrated readiness factor is capacity, not capability. A team already working flat out has no slack to learn a new workflow, however much time it will eventually save them. Successful adoptions we've run have almost always built in a genuine transition period — a few weeks where the old and new way run in parallel — rather than expecting an overnight switch.
6. Is there a genuinely nominated owner?
Tools bought without a named internal owner tend to be used enthusiastically for six weeks and then quietly abandoned when the person who championed it gets busy with something else. AI adoption that sticks has someone whose job explicitly includes watching how it's performing, retraining or reconfiguring it, and being the point of contact when it misbehaves.
7. Do you know what "working" actually looks like?
Before switching anything on, define the metric that will tell you it's paying off — hours saved, error rate reduced, turnaround time shortened — and measure the baseline first. Without a baseline, six months from now you'll have a strong feeling about whether the tool is helping, and no way to actually check whether that feeling is right.
None of these seven questions are about which AI tool to buy. That's deliberate — the tool is the easiest part of this to get right, and the last decision that should be made.
What to do if you fail most of these
Failing several items on this list isn't a reason to abandon AI adoption — it's a reason to sequence it properly. Fix the data, document the process, and nominate an owner first; the AI decision gets dramatically easier, and cheaper, once those foundations are in place. This is exactly why our Business Health Scan assesses AI readiness alongside process and people — the three are rarely separable in practice.
Get an honest read on your AI readiness
Every Business Health Scan includes this checklist, applied to your actual operation — no product pitch attached.