After four articles on what AI-ready commerce actually requires, the question we get most often from enterprise commerce leaders is the same one: "How do I tell where my organization stands without spending six months on an outside assessment?"
The honest answer is that an outside assessment will produce a more rigorous picture than a self-diagnostic. But a self-diagnostic done in an afternoon, with the right questions, will tell you 80 percent of what you need to know to prioritize the work. The other 20 percent is where the consultants earn their fee.
Here are the twelve questions we use to anchor every conversation with a new enterprise commerce client. They are grouped into the four pillars of AI-ready commerce: data, integration, organization, and measurement. Answer them honestly and the priorities will be visible by the end.
Data: four questions
1. What percentage of your active SKUs have complete, consistent attribute coverage across the catalog? AI features need this number to be above 85 percent. Most enterprise catalogs come in at 55 to 70. If your answer is "I do not know," that is also an answer, and it is the same priority as scoring below 70. We have written about why this is the most underestimated gap in enterprise commerce data.
2. Where does the system of record for product data actually live? If the answer requires more than one sentence, or if multiple systems claim authoritative status over different attributes, the architecture is the bottleneck. PIM as a single source of truth is the target state. Catalog data scattered across PIM, ERP, custom databases, and merchandiser spreadsheets is the common state, and the most expensive to leave unresolved.
3. How is product content structured for retrieval? If specifications live in downloadable PDFs, image-only assets, or unstructured marketing prose, AI search and recommendation engines cannot access them. Structured HTML with schema markup is the requirement. PDF-first product pages are a citation killer.
4. What is the cadence at which product data updates flow from source to commerce platform? Real-time or near real-time means AI features see a live catalog. Hourly is acceptable for most attributes. Nightly batch means AI features are reasoning over yesterday’s data, which compounds into degraded recommendations and inaccurate buyer-facing answers.
Integration: three questions
5. Is your ERP-to-commerce integration event-driven or batch-based? Event-driven means the commerce platform reflects changes in seconds. Batch means it reflects changes in minutes to hours. For AI features that depend on real-time inventory, pricing, or order status, the cadence of integration is the cadence at which the AI can deliver value.
6. Who owns the system of record when ERP and commerce data conflict? There has to be a defined answer for each data domain. Products, pricing, inventory, customer records, orders, returns. If conflict resolution is improvised ("we will work it out when it comes up"), the integration is structurally fragile. We have written in detail about the architecture this requires.
7. What happens when the ERP schema changes? Good answer: regression-tested integration with documented change management. Acceptable answer: ad hoc but with a runbook. Concerning answer: silence. Schema drift is one of the leading causes of integrations that pass UAT and fail in production six months later.
Organization: three questions
8. Who is the single accountable executive for the AI commerce outcome? Not "we collaborate across teams." A name. A title. A P&L. If the answer is not crisp, the ownership is not crisp, and the initiative is structurally underwritten to stall. We have covered this org-chart problem as a category of failure on its own.
9. Who has the authority to make cross-functional trade-off decisions when AI initiatives require resource shifts in IT, Data, Marketing, and Commerce? Distributed authority is functionally no authority. Trade-offs that get escalated to "the leadership team" never resolve at the speed AI projects require.
10. Does commerce have a dedicated data and integration governance function, or does it share the enterprise data team’s priority queue with every other initiative? Commerce data has commerce-specific requirements that get systematically deprioritized when they compete with finance reporting, HR analytics, and enterprise BI. Dedicated commerce data governance is the pattern that works.
Measurement: two questions
11. Do you measure AI commerce features against a holdout control group, or against all transactions? If the answer is "all transactions," your conversion lift number is conflated with selection bias and will not hold up to CFO scrutiny. Clean measurement is the difference between a defensible AI investment story and one that quietly loses funding in year two.
12. Is substrate health a tracked metric, with quarterly scoring against the data and integration conditions AI depends on? If not, the AI features will appear to suddenly stop working when in fact the underlying data quality slipped silently over the previous two quarters. Substrate health is the leading indicator that prevents painful surprises.
Scoring honestly
There is no point system here. The diagnostic works because the answers reveal priority, not because they sum to a maturity grade.
A team that answers all twelve questions confidently is ready to scale AI commerce investments. A team that hesitates on three or four has a defined work plan for the next two quarters. A team that hesitates on eight or more should hold AI investments at pilot scale until the foundations are in better shape.
The most common pattern, in our experience across enterprise B2B commerce organizations, is hesitation on six to nine questions. Data and integration gaps usually cluster together. Organizational ownership is unclear in roughly half of the engagements we run. Measurement is the most consistently under-built pillar, often because it gets attention only after AI features have launched, when retrofitting the measurement is significantly harder than building it in from the start.
What to do with the results
Three actions are worth taking inside two weeks of completing the diagnostic.
First, share the answers with the cross-functional leaders who need to align around them. The diagnostic is more useful as a conversation starter than as a private scorecard. Leaders who see the same set of gaps tend to align faster than ones who hear about them secondhand.
Second, identify the two or three gaps that are blocking the most other work. Data and integration gaps tend to be foundational. Fixing them unblocks AI features, AI search, measurement integrity, and self-service capability simultaneously. Organizational gaps tend to be next. Measurement gaps are typically the fastest to close once the priority is set.
Third, decide whether the work ahead is one your internal team can run on its own, or whether an outside partner accelerates it meaningfully. The honest answer is sometimes the former. The honest answer is more often that an outside eCommerce technology assessment will identify gaps the internal team has institutional reasons to under-report, and that the diagnostic is a productive precursor to that conversation rather than a substitute for it.
The teams that handle AI commerce well in 2026 are not the ones with the biggest budgets. They are the ones that know what they have, what they do not, and what to fund next. The twelve questions above are a faster path to that clarity than most internal conversations produce on their own.
FAQs
Q: What is an AI readiness assessment for B2B eCommerce?
A: An AI readiness assessment is a structured diagnostic that evaluates whether an organization’s data, integration, organizational accountability, and measurement infrastructure can support AI commerce initiatives. A complete assessment covers four pillars: product data quality and structure, integration architecture and cadence, single-executive accountability for the AI outcome, and clean measurement frameworks. A self-diagnostic using the twelve questions in this article will surface most of the priority gaps in an afternoon. An outside assessment adds rigor, benchmarking against peer organizations, and identification of gaps internal teams may under-report.
Q: How long does an AI readiness assessment take?
A: A self-diagnostic using the twelve questions in this article takes a couple of hours of focused leadership time and a few additional hours to source the underlying data. A structured outside assessment typically runs three to six weeks, including stakeholder interviews across IT, Data, Marketing, and Commerce, a data and integration audit, and a benchmarking review against peer enterprise B2B organizations. Both produce a prioritized action plan. The outside version produces a more defensible one for board-level conversations.
Q: What is the most common gap an AI readiness assessment surfaces?
A: Across the enterprise B2B commerce organizations we have worked with most recently, the most common gap is product data structure. Specifically, inconsistent attribute coverage across the catalog, specifications buried in PDFs rather than structured HTML, and competing taxonomies inherited from acquisitions or organizational silos. The second most common gap is unclear executive accountability for the AI commerce outcome. The third is measurement infrastructure that was not built before AI features launched, making ROI questions hard to answer credibly in year two.