Blog
It’s not a lack of ambition, investment, or innovation. The problem is simpler: most AI being used in merchandising today wasn’t built for merchandising decisions.
For example, picture a merchandiser mid-season, staring at an AI-generated pricing recommendation. The logic isn’t clear. The data source isn’t traceable. They don’t know if it’s looking at their competitors, let alone the timeframe. So they close the tab and return to the spreadsheet that has underpinned their decision-making for years.
According to McKinsey’s Global Merchandising Survey, 71% of retail merchandisers say their AI tools have had limited or no impact on business performance. More importantly, 61% say their organizations are only slightly prepared or not prepared at all to scale AI across merchandising.
These aren’t failure statistics. They’re a diagnosis. Most retail teams are deploying the wrong kind of AI – bolting generic tools onto fragmented data and calling it a strategy.
Most retail AI strategies follow the same pattern: take existing processes, layer generic AI on top, and expect transformational results. This is the “+AI” model – and it’s why so many projects fail.
Generic AI is only as strong as the context it is given. These tools were trained on broad internet data, not the operational realities of retail. They can summarise information and generate ideas, but they lack the structured retail context needed for confident merchandising decisions.
Merchandising decisions depend on something far more difficult: clean competitor matching, structured assortment hierarchies, historical pricing normalisation, product attribution, and commercially relevant context.
Without that foundation, AI outputs become directional rather than dependable. Teams find themselves spending hours manually pulling data to verify a recommendation that was supposed to save them time.
And once they stop trusting the outputs, adoption quietly disappears.
The AI+ model inverts this entirely. It means AI built around proprietary, structured data, with retail expertise embedded into the intelligence itself, not bolted on as an afterthought.
McKinsey found that merchants still spend 40% of their time on low-value tasks like system consolidation, repetitive spreadsheet work, and performance reporting. Agentic AI- when built on clean, structured foundations – could automate up to 60% of those manual tasks, returning meaningful time to strategy, vendor negotiations, and customer insight.
But here’s where most vendors miss the point entirely: you cannot automate your way out of bad data.
The merchandiser who receives a recommendation they cannot verify will reject it. An organisation that deploys AI without validated, structured data underneath hasn’t built an intelligence function – it has built a fast way to be confidently wrong.
Unlike tools used in isolation, true AI+ doesn’t clock off, surfacing signals around the clock so your team arrives Monday morning already ahead of the market, not scrambling to catch up.
What separates AI that looks impressive in a demo from AI that actually gets adopted?
It comes down to three non-negotiable dimensions.
These aren’t aspirational goals. They’re baseline requirements for any tool expected to change how a team actually works.
The difference between a failed AI experiment and a transformed merchandising team isn’t the algorithm. It’s the foundation.
EDITED has been AI-first since 2009, combining three differentiated capabilities that generic AI cannot replicate on its own:
That’s not a feature list. That’s the foundation that turns AI from a tab you close into a tool teams trust.
The next two years will separate merchandising teams that adopted AI from those that adopted the right AI. Generic tools will get smarter, but they’ll still be starting from zero every time a team asks a category-specific question. They weren’t built for this. EDITED was.
Built on proprietary, structured retail data. Grounded in how merchandising actually works.
Book a demo today to learn more.
