Twelve questions, roughly five minutes. The score is not a grade — it is a map of where your business is exposed and where automation would pay for itself fastest.
Score each question 0 to 2: 0 = no, 1 = partially, 2 = yes.
Section 1: Your current tooling (questions 1–4)
1. Do you know every AI or automation tool your staff currently use?
If tools are being used that the owner has never seen, spend is invisible and data exposure is unmanaged.
2. Can you name the monthly total of your software subscriptions?
Writing the number down is usually the first surprise. Subscription creep is the most common recoverable expense in an automation audit.
3. Do any two tools do substantially the same job?
Overlapping tools are redundant spend. If Zapier and a CRM automation both move the same lead, one of them is paying for nothing.
4. Has anyone measured the return on any AI tool you already pay for?
Tools bought but never measured are the first place to look for waste.
Section 2: Your data (questions 5–8)
5. Do you know which documents your staff paste into cloud AI tools?
If nobody knows, assume client financials and employee data are going out.
6. Is any customer or financial data stored in a system that exports it to a third-party AI service by default?
Check the vendor’s training and retention settings — most defaults favor the vendor.
7. Could you produce a list of every process where sensitive data touches an AI tool?
A score of 0 here is the single largest risk flag in the whole assessment.
8. Do your contracts with clients prohibit third-party processing of their data?
If yes, and you use cloud AI, you may already be in breach without knowing it.
Section 3: Your processes (questions 9–12)
9. Is there a task that takes more than five hours per week and is largely repetitive?
This is the highest-probability automation candidate. If none exists, AI is not your bottleneck — don’t buy any.
10. Does anyone know exactly how long that task takes, in hours, measured rather than estimated?
Savings claims without a baseline are projections, not results.
11. If an automated process produced a wrong output, does anyone know who fixes it and how fast?
No failure plan means the automation will be abandoned at the first error.
12. Could your team describe what runs a key process without you in the room?
If only one person understands it, that is a single point of failure whether the process is manual or automated.
How to read the score
0–8: awareness stage. The cheapest first step is an inventory — every tool, every subscription, every data flow. Inventory usually turns up duplicated tools and idle seats before anything gets automated.
9–16: real candidates, no controls. Prioritize questions 5–8 (data) before question 9 (automation). Automating a process you cannot secure just moves the risk faster.
17–24: ready to build. Pick the single highest-friction repetitive task, prove the return on it in 30 days, then expand. Do not automate everything at once.
What this scorecard does not do
It is a self-diagnosis, not an audit. It cannot price your savings, because pricing requires your actual documents, exports, and time records — not descriptions of them. That is the difference between a scorecard and an assessment: the scorecard tells you where to look; the assessment puts a dollar figure on what you find.
If you want the dollar figures
Thryft Advisors’ AI Readiness Assessment is a scoped, paid engagement that produces exactly this: a written opportunity list with current costs (hours per month, dollars per year), a ranked shortlist by return against effort, and a reject list of processes that look like good AI candidates but aren’t. It is time-boxed, fixed-fee, and confirmed in writing before work begins. You keep the output regardless of what you decide next.
Book an AI Readiness Assessment