What a free AI scan tells you

Search "ai readiness scan" or "free ai scan" and you get a long list of two-minute questionnaires, each one ending in a score out of a hundred and a handful of generic recommendations. They are free, they are fast, and there is a real question hiding underneath all of them: does a two-minute quiz actually tell you where AI would help your business, or is it something else wearing that label.

It is something else, mostly. Not useless, but not what it looks like either. Here is what a quickscan actually measures, where the score stops being trustworthy, and what changes once a person sits down and looks at your business instead of a form.

What the questionnaire is actually measuring

A typical quickscan asks ten to twenty questions: do you use AI tools today, how many people are on the team, do you have a data strategy, is leadership on board. You answer from memory, in a couple of minutes, without checking anything against reality. The tool maps your answers to a score and a set of recommendations pulled from a template library, sorted by industry or company size.

That is a reasonable way to measure awareness. It tells you, roughly, whether AI is already part of how your company thinks, and whether the topic has reached the people who would need to approve a project. What it cannot measure is your actual workflow: which task eats the most hours, where the handoffs between people break down, where the source data is too messy for anything to read reliably. None of that shows up in a multiple-choice form, because none of it is something you can accurately report from memory.

Where the score gets it wrong

Self-reported data is the first problem. Ask a founder how much time the team loses to manual data entry and you get a guess, usually a low one, because nobody tracks it and the real number is spread across five people doing it in small chunks all week. Ask the same question with an actual timesheet in front of you, and the number almost always comes out higher, sometimes by a lot.

The second problem is that the recommendation is generic by construction. A scan built to serve every visitor cannot know that your invoices arrive as photographed receipts in three different formats, that your CRM has not been touched since 2024, or that the process everyone agrees is broken is quote writing, not the one the scan happened to ask about. You get a category of advice ("start with a pilot project", "invest in data quality") that is true for almost any company and specific to none.

The third problem is worth saying plainly: most of these scans exist to generate a sales lead, not to inform you. That is a legitimate business model, marketing is not a dirty word, but it means the score is built to end in a phone call, not to be the deliverable itself.

When a quickscan is still worth the ten minutes

Honesty over hype: a quickscan does have a real use, just a narrower one than the score suggests. For a leadership team that has never seriously discussed AI, a score out of a hundred is a cheap way to open the conversation. It costs nothing, it takes ten minutes, and it puts something on the agenda that might otherwise never get there. Use it for that. Do not mistake it for a plan.

What changes when a person looks instead

The difference is not AI versus human, it is depth. A quickscan runs on what you type into a form. An hour with someone who has done this before runs on what your business actually does: the invoices you show them, the process you walk them through step by step, the tool nobody remembers is still switched on. A person notices the thing you did not think to mention, because you stopped seeing it as a problem a long time ago.

It also means the plan that comes out the other end is specific rather than generic. A scan tells a marketing agency and a logistics company the same three things about "process automation". A person who has actually looked knows that the agency's bottleneck is proposal writing and the logistics company's bottleneck is order confirmations, and that the fix for one does not transfer to the other. That specificity is the entire point, and it is exactly what a form cannot produce.

That is also why our own AI audit is built the way it is: one hour, one on one, through your real workflow, not a form. Whatever turns up goes into a report with every gap, its monthly cost, and the fix, written for your business specifically. If nothing worth automating turns up, the report says that too. Free, no barrier, because the point is to find out whether we are a fit, not to sell the session itself.

Putting a number on it

Before booking anything, quickscan or otherwise, run one exercise yourself. Pick the task your team complains about most. Time how long it actually takes this week, not from memory, with a stopwatch or a timesheet. Multiply the weekly hours by a realistic internal hourly rate. That number, not a score out of a hundred, is what tells you whether the problem is worth fixing. A ten-minute task done twice a day is a different conversation than a two-hour task done every Monday, even if a generic scan would flag both the same way.

If that number turns out to be worth acting on and you want it mapped properly, with an actual look at your workflow instead of a form, that is what the audit is for.