The 30-day AI readiness assessment: what it covers, what it costs, what you walk away with

A specific scope-and-deliverables breakdown for the most-asked-about engagement we run. What gets measured, who gets interviewed, and the four findings that determine whether AI projects ship at your company.

What does the 30-day AI readiness assessment actually cover?

Three things. One: an honest read on your data layer (entities, metrics, lineage, freshness) against the use cases you want to ship. Two: a sequenced remediation plan with effort, owner, and expected outcome per item. Three: a go or no-go on the AI use cases you are considering with a defensible reason for each.

Every AI consultant offers an "AI readiness assessment" right now. Most are a two-week glance with a 12-page PowerPoint that says "improve your data foundation." That's not an assessment. That's a sales meeting.

This is the actual scope and deliverables of the 30-day AI readiness assessment we run. It maps to the work we've done across 47 enterprise engagements and the patterns we've found in the 2026 Enterprise Data Readiness Report.

What this is

A specific, hour-by-hour breakdown of a 30-day AI readiness assessment. Who we talk to, what we measure, what report you get, and what the findings typically look like.

What this is not

A free sales pitch. We charge for the assessment and the deliverable is a 30-page report that's useful even if you never engage us for the implementation work. Treat it as paid consulting, because it is.

The four-week structure

Week 1: Discovery

Days 1 to 5. This is the listening week.

Stakeholder interviews, 6 to 10 people across executive, engineering, data, and end-user roles. Each is a 60-minute structured conversation against a fixed question set so we can compare answers across roles. We're listening for:

• Where the friction is between data team and business
• Which AI projects have been attempted and stalled, and why
• What "production AI" means to each stakeholder (the answers vary by 80%, which is itself a finding)
• What success looks like 12 months out

Artifact review, we ask for read-only access to your data warehouse, BI tool, ticketing system, and any existing AI project repos. We're scanning for:

• Naming conventions (or absence of them)
• Schema documentation coverage
• Test coverage
• Model cards (almost never exist)
• Existing data contracts

We don't change anything. We just look.

Week 2: Measurement

Days 6 to 10. This is the quantitative week.

Metric definition audit, we pull the top 20 metrics from your BI tool and trace each one back to the source. We're checking whether the same metric name produces the same number across systems. The answer is almost always "no." The size of the discrepancies is the finding.

Data quality sampling, we run statistical tests against the most-queried tables: completeness, uniqueness, freshness, value-distribution shifts. We score each table on a 0-to-100 scale.

Pipeline reliability, we look at the last 90 days of pipeline run history. Mean time between failures, recovery time, alert noise. This tells us whether the data team is firefighting or operating.

Compute and cost, we pull warehouse spend, BI tool spend, and any existing AI/ML tooling spend. We map it against utilization. Most clients are paying 20 to 40% more than they need to.

Week 3: Diagnosis

Days 11 to 18. This is the synthesis week.

We look across everything from weeks 1 and 2 and write up the findings. Every assessment we've run produces findings in five categories:

1. Identity resolution gaps, places where the same entity (customer, patient, transaction) shows up under multiple keys with no reconciliation. This is the silent killer of AI projects.

2. Metric definition mismatches, places where business stakeholders mean different things by the same word. "Revenue" is the most common offender.

3. Pipeline fragility, pipelines with high failure rates, high mean-time-to-recovery, or high alert noise. AI on top of fragile pipelines fails silently.

4. Governance vacuum, no data-product ownership model, no versioning of schemas, no tests, no SLAs. AI projects can't ship without these.

5. Tooling waste, overspending on platforms whose capabilities you're not using, or running 3 tools that overlap when 2 would do.

Week 4: Recommendations

Days 19 to 30. This is the deliverable week.

You get a 30-page report (we measure pages because some buyers want the artifact for their own reporting; the actual document is 25 to 35 pages of substance, no filler).

The report has four sections:

• Executive summary, 2 pages. Five biggest findings, scored by impact and effort.
• Findings, 10 to 15 pages. Every issue from week 3, with evidence (screenshots, query results, interview quotes), severity, and root cause.
• Recommendation roadmap, 8 to 12 pages. A 90-day, 6-month, 12-month sequenced plan. Each step has scope, expected effort, and dependencies.
• Appendix, your data quality scorecard, your pipeline-reliability numbers, your compute-cost analysis. The raw data so you can re-run the analysis yourselves later.

We also do a 90-minute readout call with the executive team. This matters more than the report itself because the report tends to confirm what people already suspected, and the conversation is where the agreement on what to do next gets built.

What it costs

The 30-day AI readiness assessment runs $35K to $60K depending on company size and the number of stakeholders we interview. We typically come in mid-range for a mid-market company (200 to 1,000 employees) and higher for enterprise environments with multiple business units.

The deliverable is yours. You can use it to drive an internal initiative, hire someone else to do the implementation, or hire us. We've had clients use the report to get budget approval and then choose a different implementation partner. That's fine.

What buyers get out of it

Across the 47 engagements behind the 2026 Enterprise Data Readiness Report, the assessment outputs map to three buyer outcomes:

Outcome 1: "We were ready, the assessment confirmed it" (about 15% of clients). This client moves directly into AI implementation. The assessment de-risked the executive-team confidence that the data foundation could handle production AI workloads. The cost of the assessment is rounding error compared to the cost of a stalled $4M AI pilot.

Outcome 2: "We had foundational gaps, the assessment found them" (about 65% of clients). This client takes the recommendation roadmap and works the 90-day section first. AI is paused or scoped down to a much narrower pilot until the foundation is in place. The assessment saved a misallocated AI budget.

Outcome 3: "We weren't sure why our previous AI project failed, the assessment told us" (about 20% of clients). This client had already tried AI, hit the data layer, and stalled. The assessment is the postmortem. The findings explain why the previous project couldn't ship and what would have to change for the next one to.

Common questions

Why 30 days?

Two weeks is too short to do real measurement. Eight weeks is too long and the urgency degrades. Thirty days is the smallest window that fits discovery + measurement + synthesis + writing. Anyone selling a 5-day or 10-day "assessment" is doing a sales meeting.

Do we have to give you access to production data?

Read-only. We do the work in your tooling using your accounts. We don't extract data, we don't connect external tools to your warehouse, and we don't keep credentials past the engagement.

What if you find something embarrassing?

We have. The point of the assessment is to surface what's actually there, not to make you look good. The report goes to whoever you decide it goes to. We've never had a client share the full findings beyond the executive team, and that's reasonable.

Can we run this internally instead of hiring you?

Yes, in theory. Most internal teams can't because (a) they're too close to the system to see the patterns, (b) they don't have the comparative dataset across 47 other companies, and (c) the political dynamics inside the company prevent honest answers in stakeholder interviews. External consultants are paid in part for the freedom to ask uncomfortable questions and write down the answers.

How is this different from the assessment my SI already did?

Most system-integrator assessments are scoped around the SI's services. They find what they sell. We don't sell platform implementation, so we can recommend "stay where you are and fix governance" or "switch SIs". That changes which findings show up.

What does the engagement look like after the assessment if we hire you?

We typically do the 90-day plan from the recommendation roadmap as a follow-on engagement. That's a separate scope and proposal. Most clients sign within 30 days of the readout call. Some take 90 days. About 25% never engage us beyond the assessment, which is by design.

If you want to see what the deliverable actually looks like before buying, we send sample reports under NDA on request. Reach out via AI Readiness Assessment or read the full pattern-set in the 2026 Enterprise Data Readiness Report. The report is the macro view; the assessment is the micro version applied to your environment.

The clearest case study on this is Express Scripts' AI readiness engagement, we ran the 30-day version, found three critical foundation gaps, and they used the roadmap to prioritize the next two quarters of data work before resuming the AI initiative.

Frequently asked questions

What does the 30-day AI readiness assessment actually cover?

Three things. One: an honest read on your data layer (entities, metrics, lineage, freshness) against the use cases you want to ship. Two: a sequenced remediation plan with effort, owner, and expected outcome per item. Three: a go or no-go on the AI use cases you are considering with a defensible reason for each.

Who runs the 30-day AI readiness assessment at Thinklytics?

A senior practitioner runs each engagement end to end. Same person who reads your tables runs the readout. We don't ramp junior analysts on your data. The team is small on purpose so the work is consistent.

What is the deliverable at the end of 30 days?

A 14 to 22 page report with the data-layer findings, the use-case go/no-go list, the remediation plan, and the budget envelope. The report goes to whoever sponsored the engagement plus an executive readout call. No slideware. The document is the deliverable.

How much does a 30-day AI readiness assessment cost?

Most engagements land at $35,000 to $55,000 depending on environment complexity (number of source systems, number of in-scope use cases, regulatory overhead). We send a fixed-fee quote after a 30 minute discovery call. No retainer creep.

Will the assessment recommend a specific AI vendor?

No. The assessment evaluates which vendor categories your environment is ready for. Vendor selection happens after the data layer is sequenced because the right vendor depends on what you have, not on what is trending. Our AI workflow automation vendor evaluation 14 questions walks through the selection logic.

What happens after the 30-day assessment?

Most clients pick one to three remediation items from the plan and engage Thinklytics or their internal team to execute. We don't require a follow-on contract. If the plan says fix it yourself, the plan says fix it yourself.

Who reviews the assessment output internally?

Whoever sponsored it plus the operating leaders for each in-scope use case. The assessment is not meant to be IT-only; the business owners need to see the remediation plan because their workflows are downstream of the data layer. We always do an executive readout call to make that explicit.

How long after the readout can we start remediation?

Most clients start the first remediation item within 2 to 4 weeks of the readout. Faster is fine when the sponsor is decisive; slower is fine when the budget cycle gates the start. We don't push to start before the sponsor is ready.