AI Readiness Scorecard: 10 Questions Before You Spend a Dollar on AI
Most mid-market AI projects fail because the company wasn't ready — not because the tech didn't work. Use this 10-point scorecard before your next AI initiative.
George Fassett, Jr. · July 8, 2026
The mid-market AI failure rate is somewhere north of 70%, and it's almost never a model problem. It's a readiness problem: the company bought an AI tool before it had the data, the process, or the governance to make the tool useful.
Here is the ten-question readiness scorecard we run before we recommend a single dollar of AI spend.
The ten questions
- Do you have a written AI usage policy? If employees are pasting customer data into free ChatGPT, you have a data-leak problem, not an AI opportunity.
- Is your customer, financial, and operational data in structured systems, not spreadsheets and email? AI amplifies whatever data it can reach — including the mess.
- Can you name the single business outcome the AI project must move? "Efficiency" and "innovation" are not outcomes. Revenue per employee, cycle time, and gross margin are.
- Do you have an executive sponsor who will kill the project if it doesn't hit the number? Projects without a kill switch become expensive science fairs.
- Is there a process owner accountable for the workflow the AI touches? AI without a process owner is a demo, not a deployment.
- Have you documented the current-state process end-to-end? You cannot automate what you cannot describe.
- Do you have a data governance model — who owns which fields, who can change them? AI decisions are only as good as the data lineage behind them.
- Is your identity and access model tight enough that you can revoke AI access in one place? Shadow AI accounts are the new shadow IT.
- Do you have a security review process for third-party AI vendors? SOC 2 Type II is table stakes, not a differentiator.
- Are you prepared to retrain 10–30% of the affected team? The tool is 20% of the change; the people work is the other 80%.
Scoring
Give yourself one point per yes. Below 6, do not buy AI tools yet — invest the money in data, process, and governance. Between 6 and 8, run one narrow pilot with a hard ROI target. Above 8, scale.
The pattern we see
The companies winning with AI in the mid-market are not the ones with the most sophisticated models. They are the ones who cleaned up their data and processes first, then pointed AI at the highest-value bottleneck.
Get the full scorecard with scoring rubrics and vendor-selection criteria in the AI Readiness Scorecard.