Agentic shopping needs agentic merchandising 

Get the merchandising right. Agentic AI turns it into margin

The agentic shopper has arrived. AI-driven ecommerce traffic rose 693% over the 2025 holiday season, according to Adobe. Shoppers now ask an assistant for “a lightweight rain jacket, under £60, in stock near me by Saturday” and let it do the comparing. 

Most retailers are answering at the front end. They’re cleaning product feeds, launching shopping assistants and rewriting descriptions for machines to read. All of that matters. None of it decides who wins. 

The agent doesn’t browse. It shortlists. And it shortlists on the things merchants argue about every week: the right product, in the right place, at the right price, with enough stock to sell. That shortlist was decided weeks ago, in the range review, the buy and the allocation. 

Agentic shopping doesn’t need a cleverer storefront first. It needs agentic merchandising. 

What agentic shopping does to your merchandising 

The shopper’s agent checks product, price, size availability and delivery promise across every retailer it can reach, returns a shortlist, and increasingly completes the purchase. 

A human shopper forgives weak merchandising. They take the navy instead of the white, wait a day for their size, or stay loyal out of habit. An agent forgives nothing: 

• It’s binary. You’re on the shortlist or you’re not. There’s no “lower conversion”, just no sale. 

• It sees everything at once. Price, stock and delivery are compared across every competitor in seconds. Brand and store layout can’t hide a weak offer. 

• It moves demand faster. Spikes arrive sooner and sharper. A weekly report finds them after the agent has already sent shoppers elsewhere. 

• It has no habit. The loyalty that used to cover for a missing size or a late buy disappears. 

Every one of these failures was decided in merchandising, weeks before the shopper asked. And putting another agent at the front end won’t fix it. An agent can only shortlist the stock you bought, in the places you put it, at the price you set. 

Why agentic AI isn’t paying off for merchants yet 

Merchants aren’t short of AI. They’re short of AI that changes the numbers. In McKinsey’s January 2026 research, 71% of merchants said AI merchandising tools have had limited or no effect on their business so far. They still spend about 40% of their time on low-value work such as consolidating data and fixing spreadsheets. And 61% said their organization is not at all, or only slightly, prepared to scale AI across merchandising. 

The reasons are familiar to anyone who has sat in a range review: 

• Agents without context. A generic model doesn’t know your hierarchy, your margin rules, your open-to-buy or why a line is on the floor. It makes confident recommendations that a buyer dismisses in seconds. 

• Agents bolted onto silos. One tool for the range, another for the buy, another for allocation, and finance in a spreadsheet. Each agent optimises its own corner, and nobody sees the whole decision. 

• Recommendations with no price tag. “Chase this line” means little until you know what it does to margin, markdown risk and cash. Merchants won’t act on advice they can’t cost. 

• Pilots that never scale. A clever proof of concept on one category, then back to the weekly report. 

None of these is an AI problem. They’re merchandising problems that AI has made more visible. 

Merchandising is getting the basics right 

Great merchandising has always been a handshake between three things: the market, the customer and the discipline of the merchant. The market tells you what’s changing. The customer tells you what they want. The discipline turns both into product, placement, price and stock that make money. 

The agentic shopper doesn’t change that handshake. It tests every part of it, on every request.

The basic	What the shopper’s agent checks	What merchandising has to get right
Product	Is this what was asked for?	A range built around customer missions and local demand, not last year’s sales
Placement	Is it available here, by Saturday?	Stock allocated to the stores, regions and channels where demand is
Price	Is it the best value on the list?	Price and promotion decided with their margin impact in view
Stock	Is the size I want in stock?	Buy depth and open-to-buy kept back for what’s actually selling

Every row is a decision a merchant made weeks before the agent arrived. Get the basics wrong and no amount of AI at the front end will save the sale, or the margin. 

Agentic AI releases the value 

Once the basics are connected, agents start to pay their way. Their job is to give back the 40% of the week merchants spend gathering data, so they can spend it deciding. 

In practice, a merchandising agent should do five things, in order: 

1. Analyse: spot the change in sales, margin, markdown or average selling price as it happens. 

2. Explain: show why, by category, region, channel or store. 

3. Simulate: model the options and their impact on margin, open-to-buy and stock. 

4. Align: put the trade-offs in front of merchandising and finance together. 

5. Act: execute within agreed limits, once the merchant approves. 

Here’s what that’s worth on a single line. Two retailers each buy 40,000 lightweight jackets at a £16 landed cost, to sell at £55. A wet, mild spring arrives early and searches spike in week two. 

• Retailer A sees it in next week’s report. Core sizes sell out, the shopper’s agent sends people elsewhere, and 12,000 leftover units clear at half price. 

• Retailer B sees it within days. The agent recommends reordering 8,000 units of core sizes and rebalancing stock to the regions that are selling, and shows the margin and cash impact before the merchant commits. 

The illustrative result: £535,000 more gross margin from one line in one season, and 87.5% full-price sell-through against 70%. 

(Workings: A sells 28,000 at full price and clears 12,000 at 50% off, for £1.23m gross margin. B chases 8,000 units at £17 landed, sells 42,000 of 48,000 at full price and clears 6,000 at 30% off, for £1.765m.) 

Same merchant, same judgement. The difference was speed, and knowing what the call was worth before making it. 

What agentic merchandising takes 

Agents are only as good as the merchandising they sit on. That’s why Board starts with the discipline, not the AI. 

• Grounded in how you trade. Board holds your merchandise hierarchy, margin rules, constraints and trading calendar in one model, alongside market and customer signals from Board Signals and Board Foresight. AI recommendations start from your business, not a generic one. 

• One plan that never stops. Merchandise financial planning, open-to-buy, assortment, and allocation and replenishment work as one continuous plan. A demand shift reaches the range, the buy and the stock plan without waiting for the next review. 

• Profit protected and profit pursued. Board was built for finance, so every chase, hold, buy and markdown is shown in sales, margin, stock and cash first. Merchandising, supply chain and finance stay in lockstep, from SKU, store and week up to the general ledger. That covers both profit motions: protecting margin when demand turns, and going after it when demand surges. 

• People in control. The Board Merchandiser Agent analyses, explains, simulates and recommends, then acts only within agreed limits once the merchant approves. Every recommendation is traceable. 

The result is confident merchandising, winning economics and AI at every level. 

Start with merchandising 

There’s a temptation to answer the agentic shopper with more AI at the edge. Resist it. The fastest way to get profit under control is the decision that sets it: what you range, where you place it, how you price it and how deep you buy. 

Merchandising is your first port of call. Get it connected, and every other planning decision has something solid to connect to, from supply chain to the finance plan. 

Shopping has gone agentic. Make sure your merchandising has too. 

Sources 

• Microsoft propels retail forward with agentic AI capabilities, Microsoft, 8 January 2026, citing Adobe holiday traffic data 

• Merchants unleashed: How agentic AI transforms retail merchandising, McKinsey & Company, January 2026 

• Board Merchandiser Agent, Board, June 2026