AI Readiness Assessment
Most AI projects fail because the organisation wasn't ready, not because the technology didn't work. Ten questions tell you which is your problem before you spend a dollar.
- No signup required
- 100% free
- Built from 100+ intakes
How would you describe your data quality?
Think about how your data is stored and organized
What is AI readiness?
AI readiness is whether your data, infrastructure, team, and exec sponsorship can carry an AI project from pilot to production. Most companies score 30–60 out of 100 on first pass, which means a pilot can start but a launch shouldn't. The score isn't a verdict; it's a map of what to fix first.
01 Reading the number
What your score actually means
The score is diagnostic, not pass-fail. Each band has its own next step, and skipping the step usually shows up later as a stalled pilot or a launch that quietly gets shelved.
Score 0–40
Foundation work first. AI is the wrong next step.
Pattern we see at this score: customer data spread across 6–8 systems with no single owner, no exec sponsor beyond a curious VP, no use case written down with a measurable outcome. Spending on a model here burns budget. Spend the first 8–12 weeks on a data audit and one use-case sprint. Then re-score.
What we'd do next
Data engineering and a 2-week use-case sprint.Score 40–70
Pilot-ready with real risks. Most teams land here.
Roughly 6 in 10 teams we score sit in this band. You have enough data and enough buy-in to ship a pilot, but not enough of either to launch in production yet. Pick one use case where a wrong answer is recoverable. Time-box it to 8 weeks. Run on real data, not synthetic. Don't add a second use case until the first one is in production.
What we'd do next
Scoped AI pilot with a fixed exit criteria.Score 70+
Launch-ready. Watch for the failure modes that hit at this stage.
You can move to production, but the failure modes shift. Scope creep kills the timeline (every team lead wants their feature in v1). Model drift degrades accuracy 3–6 months in if you don't monitor it. Governance gaps surface when a model decision lands in a customer-facing dispute. Plan for all three before launch, not after.
What we'd do next
Production-grade build with monitoring and governance baked in.02 Archetypes
The four score patterns we see most often
The total score hides the shape of the problem. Two companies with a 55 can need completely different next steps. Here are the patterns that come up again and again across 100+ assessments.
- 01Exec team is sold on AI. Roadmap mentions AI in three places. Then you go to pull the data and it's in 8 systems: a 12-year-old CRM, Stripe, two warehouses, a spreadsheet the ops lead maintains by hand. Nobody owns the canonical version. The score gives you cover to slow down and fix the foundation before someone hires a vendor to glue it together with prompts.High strategy, low data
- 02Cloud-native team. Modern data stack. A platform engineer who can ship anything. The board put AI on the strategy deck and now leadership is chasing a use case to justify the line item. The score is fine but the brief isn't. The right next step is a use-case sprint, not a build.Strong infrastructure, unclear goal
- 03A mid-level analyst or eng manager ran the assessment because they care. Score looks decent. But there's no VP who'll defend the project budget when the first quarter doesn't show an obvious win. AI projects without exec air cover get cut at the first re-prioritisation. Fix the sponsorship before the build.Skilled team, no sponsor
- 04Executive wants AI. Budget is signed. Hiring is taking 9 months because the market for senior ML engineers is brutal. Score is held back by the team dimension only. This is the classic outsource pattern: bring in a build partner for the first 12 weeks while you hire the long-term owner in parallel.Sponsor and budget, no team
Honest read
This assessment doesn't tell you whether AI will work for your business. It tells you whether your organisation can carry the project to production right now. A 35 today can be a 75 in six months. A 75 doesn't guarantee the use case is the right one. We treat the number as a starting point for a 30-minute call, not an answer.
How it works
Evaluates your organization across 10 dimensions: data quality, infrastructure maturity, team expertise, process readiness, leadership alignment, and more.
Related tools
How this was built
The 10 questions and scoring weights were built from 100+ project intakes since 2022, across AI consulting, AI development, and data engineering engagements. We update the question set as foundation-model maturity shifts. The last revision moved weight away from in-house ML headcount and toward data quality and governance, because pre-trained models pushed the bottleneck upstream. Last updated 2026.
Frequently asked questions
- You need three things: clean, accessible data; technical infrastructure that can support model deployment; and at least one executive sponsor who understands the investment timeline. This assessment evaluates all three plus four additional dimensions.
- Data quality. Most organizations have data spread across siloed systems in inconsistent formats. Before any AI initiative, invest in consolidating and cleaning your data. This single step can cut AI project costs by 30–40%.
- Yes, especially with pre-trained models and APIs. Small companies don’t need massive datasets or ML teams. Start with off-the-shelf AI services (ChatGPT API, cloud vision APIs, etc.) and focus on a single high-impact use case.
- No. A low score means the foundation isn't there yet, not that AI is the wrong direction. The fix is usually a data audit and one scoped use-case sprint, not a 12-month transformation. Most teams move from a 35 to a 65 in 4–6 months when they pick one workflow and clean the data behind it.
- A high score means your organisation can carry the project to production. It does not validate the use case. Plenty of 80-score teams have shipped models that nobody used. Pair the readiness check with a use-case scoring exercise: business impact, data availability, and time-to-value should all rate as high before you build.
- A paid audit goes deeper: interviews with 8–15 people across the org, system-level data quality scoring, vendor and tooling review, and a written 30–60 page report. Cost is usually $30K–$80K. This free tool gives you the directional read in 4 minutes so you can decide whether the deeper audit is worth it. Most teams in the 50–70 band don't need the deeper audit; teams in 20–40 usually benefit.
- Whoever has the clearest view across the six dimensions. In smaller companies that's usually the founder or COO. In mid-market it's typically a VP of Engineering or VP of Operations. The CTO often over-rates infrastructure; the CEO often over-rates strategy. The most accurate scores come from the person closest to where the data actually lives.
Got your score? Let's talk about the next 90 days.
30-minute call. We walk through your score, point at the biggest gap, and tell you whether a pilot, a data project, or a use-case sprint is the right next move. No deck.