Enterprise Intelligence Group
AI maturity is not how much AI you own. Six dimensions that show where AI can create practical value in retail — and what to do next.
A category manager opens a report on an item that, by every measure available to her, should come out of the range: thin margin, middling rate of sale, duplicated by two better performers. Before she signs the delist, she wants to know one thing — why was it ranged in the first place? The assortment system can rank the product. It cannot answer the question.
That gap is the most useful place to start a conversation about AI in retail, because it has almost nothing to do with AI. The organization already owns the assortment platform. It has the transaction history, the performance reporting, the category strategy deck from the last review. What it cannot retrieve, in the moment the decision is being made, is the reasoning: the supplier commitment, the basket argument, the exception someone approved two resets ago and never wrote down.
Retailers, CPG manufacturers and distributors do not have a shortage of AI tools. ChatGPT, Claude, Copilot, Gemini and increasingly capable agent platforms are all available. What is scarce is practical clarity about where AI can help, and what has to be true first.
Every scenario in this article is an illustrative example rather than a named client result.
Owning the technology was never the constraint
A retailer can run sophisticated forecasting, assortment optimization and space-planning platforms and still have category managers rebuilding the same analysis in spreadsheets, planners investigating exception queues by hand, and commercial reasoning disappearing into meetings, inboxes and individual memory.
This is not a failure of those platforms. Specialist systems calculate, optimize, transact and hold the authoritative record — and they should keep doing exactly that. The work that stays manual is the work around them: retrieving context, assembling evidence, reconciling sources, explaining exceptions, deciding, and capturing why. That is where AI belongs first, and it is why we do not think the credible enterprise proposition is replacing a forecasting engine or a planogram application. It is making the people and workflows operating those investments substantially more effective.
Which reframes the maturity question. AI maturity is not how much AI you own. It is how effectively your organization turns trusted data, retained knowledge, defined decisions, governed workflows and human judgment into better outcomes.
Intelligence before automation
We work to a simple sequence, and we call it Intelligence Before Automation : understand the decision; capture the knowledge, evidence, rules and exceptions behind it; govern what can be trusted; then automate the smallest useful part of the work, with the right human accountability and measures.
The middle step is the one that is easiest to skip. The tool gets chosen before the decision is mapped, and it is pointed at sources nobody owns. Governance is what closes that gap, and it is not a compliance review bolted on at the end — data quality, access, source traceability, decision ownership, approval and monitoring belong inside each workflow. That is why governance runs across everything below rather than sitting in a box of its own.
The Retail AI Maturity Assessment examines six dimensions, in three tiers — Emerging, Developing and Ready. None of them is pass or fail. The point of the model is to find the condition that is limiting progress, and the next sensible move.
1. Retail Intelligence and Enterprise Knowledge
Does the organization retain the reasoning behind decisions — or only the resulting data and outputs?
Back to the delist. Data records what happened; Retail Intelligence makes the context available — why the category was built that way, which supplier commitment matters, what exception was approved, what the team learned last time. Much of it exists only as the Unwritten Knowns held by experienced people.
The EIG Library · Insights · AI Maturity Assessment · Contact