Before you buy an AI tool or hire a consultant, you should be able to answer 12 questions. Most mid-market teams can't answer more than six or seven of them clearly. That gap is usually the main reason AI investments underperform — not the technology, not the vendor, not the team's intelligence or ambition. They started before they were ready.
"Ready" doesn't mean perfect. It means having enough of the foundation in place that an AI initiative has a reasonable chance of succeeding rather than a predictable chance of stalling. Here's the checklist, organized by category.
People readiness
1. Who owns outcomes?
Not who owns the tools or the budget. Who is accountable for whether people actually use AI and whether it produces results? This person needs to have both the authority to make decisions about the rollout and the bandwidth to drive it week over week. If the answer is "whoever has time," the initiative doesn't have an owner.
Ready: A named individual with clear accountability and protected time to drive the initiative.
Not ready: Shared responsibility across a committee, or ownership assigned to someone without capacity.
2. Do you know which roles will benefit most?
AI creates leverage for specific types of work: drafting, summarizing, data manipulation, research, pattern recognition. Some roles have high concentrations of this work; others don't. Rolling out AI to everyone simultaneously without knowing this wastes training resources and produces low adoption in the roles where AI isn't naturally useful.
Ready: You've identified 2-3 roles where AI has the highest potential impact and can describe specifically why.
Not ready: You plan to roll out broadly and let people figure out what's useful.
3. Have you talked to the people who will actually use it?
This sounds obvious. Most organizations skip it. Leaders design AI rollouts based on their understanding of what the work looks like — which is often different from what the work actually looks like. The people doing the work know where the friction is, what takes too long, and what they'd automate if they could. They're also the people who will tell you, honestly, whether they'll use a tool or not.
Ready: You've had real conversations with people across the roles you're targeting, not just interviews with their managers.
Not ready: The rollout plan was designed top-down with no ground-level input.
4. Is there a plan for resistance?
Some teams will be skeptical. Some individuals will feel threatened. Some managers will deprioritize AI adoption because it's not in their performance review. None of this is unusual — it's present in every large-scale technology rollout. The question is whether you have a plan for addressing it, or whether you're assuming it won't be an issue.
Ready: You've identified likely sources of resistance and have a communication and engagement plan.
Not ready: You're assuming enthusiasm will be sufficient.
Process readiness
5. Are your core processes documented?
AI works best when it's augmenting a defined process — not an ad-hoc one. If a workflow isn't documented, it's hard to know where AI fits, hard to train people consistently, and hard to measure whether AI is actually changing anything. Undocumented processes also tend to have more variation than anyone realizes, which creates problems when you try to automate them.
Ready: The workflows you're targeting are documented at a level where a new employee could follow them.
Not ready: Workflows are in people's heads, and "it depends" is the most common answer when you ask how things work.
6. Have you identified specific use cases?
Not "we'll use AI to improve productivity." Specific: "Proposal writers will use AI to generate first-draft sections from client briefing notes, cutting initial draft time from 4 hours to 90 minutes." That level of specificity. Without it, training is generic, adoption is optional, and results are unmeasurable.
Ready: 3-5 specific, workflow-level use cases defined for each priority role.
Not ready: General aspirations about AI improving work quality or speed.
7. How will you measure success?
Before you launch, you need a baseline and a target. Baseline: current state metrics for the workflows you're targeting (time per task, error rate, volume capacity, etc.). Target: what those metrics should look like at 30, 60, and 90 days. Without these, you're managing the rollout by anecdote.
Ready: Specific metrics defined, baseline data captured, 90-day targets agreed on.
Not ready: Success defined as "people are using it" or "leadership is happy."
8. Is there a rollout sequence?
Who gets access first? In what order? A phased rollout — pilot with 10-20 people, iterate, then expand — is almost always better than a big-bang launch. The pilot surfaces problems before they affect everyone. It also creates advocates: early adopters who can speak to the tool's value from experience, not from the announcement.
Ready: A pilot group defined, a rollout sequence planned, a timeline agreed on.
Not ready: Plan to make the tool available to everyone on the same day.
Technical readiness
9. Does the tool integrate with how people actually work?
A tool that requires context-switching won't become a habit. If your team does their core work in Salesforce and the AI tool lives in a separate interface, adoption will be low — not because people don't want to use it, but because the friction of switching is enough to make the default behavior "don't." The best AI integrations are invisible: they're in the tool you're already in.
Ready: The AI tool integrates into the primary systems where target roles spend their time.
Not ready: Users will need to copy-paste between the AI tool and their work.
10. Is your data in shape?
This matters most for AI applications that will be trained on or work with your organization's data — custom models, retrieval-augmented systems, AI that accesses your CRM or knowledge base. If your data is inconsistent, siloed, or poorly structured, the AI will reflect that. Garbage in, garbage out applies here at high speed.
Ready: Data relevant to the use case is clean, consistently formatted, and accessible.
Not ready: "Our data is a mess" — which is fine to know, but means you should fix data before, not after, deploying AI on top of it.
11. What are the security and compliance requirements?
For most mid-market companies, this comes down to: what data will the AI tool process, and what are the rules around that data? If you're in healthcare, financial services, legal, or HR tech, the answer has real complexity. Vendor SOC 2 compliance, data residency, what the vendor can use your data for, and how to configure the tool to prevent sensitive data from leaving your environment all need answers before you sign a contract.
Ready: Security requirements documented, vendor compliance verified, data handling policies agreed on.
Not ready: Security review hasn't happened yet ("IT will figure it out").
12. Who handles it when something breaks?
AI tools have bugs, outages, unexpected outputs, and edge cases. The vendor has a support process. Your IT team has a support process. The question is whether those two things connect clearly, and whether the people managing the rollout know what to do when something goes wrong. "Someone will figure it out" is not a support plan.
Ready: Clear escalation path documented: who owns what when the tool misbehaves, how users report issues, what the SLA is for resolution.
Not ready: Users will email the AI owner, who will email IT, who will email the vendor.
How to use this checklist
Don't use this as a gate — "we need to be fully ready before we start anything." Use it as a diagnostic: where are the biggest gaps, and which gaps will most directly damage the rollout if they're not addressed first?
The most critical items, roughly in order of impact on rollout success: ownership (1), specific use cases (6), success metrics (7), and tool integration (9). If you can't answer those four clearly, the other eight won't save you.
If you scored yourself and found significant gaps: the right move isn't to delay indefinitely. It's to start with the AI Readiness Audit — a structured engagement that surfaces exactly these gaps and gives you a prioritized plan for addressing them. Most gaps can be addressed in 4-8 weeks with the right focus.