Trend Forecasting Meets Planning Reality: Bridging the Gap Between External Data and Buying Decisions

A merchandising team has to commit to fabrics, volumes and price points months before a garment reaches a store. Last season's sales help with core items. They say little about a silhouette that has never existed. When the bet is wrong, the bill arrives as excess stock and early markdowns. External trend data is meant to close that gap. Gartner expects 70% of large organizations to adopt AI-based forecasting by 2030. The firm also ties slow uptake to gaps in data completeness, availability and accessibility. The bottleneck is often the plumbing and not the model.

The Bridge Merchants Still Build by Hand

Heuritech forecasts consumer demand from social media imagery up to 24 months ahead. Its feed sits upstream of planning platforms and does not replace them. Responding to an inquiry from The Supply Chainer, Léa Gossein, Chief Marketing Officer at Heuritech, argued that ERP integration is often the wrong first problem.

Her explanation is operational. Sales performance sits in one system and range planning in another. Predictive logic often sits nowhere. Merchants rebuild the bridge in spreadsheets every season. A new external feed, she said, becomes a fourth place to look. The vocabulary also clashes. An ERP speaks SKU and past sales. Trend data speaks shape, colour, fabric and print.

Gossein said the company narrows the scope on purpose: “Our view is that if the data makes sense to the teams, 80% of the job is done. That shapes how we work: we start with one category and one buying decision, we agree with the merchandising and planning teams on which attributes matter, and we deliver in their vocabulary and on their calendar. Our fashion experts are hybrid profiles who speak both the technical and the industry language, and they stay involved season after season. System integration comes later, once the teams have seen it work.”

Where Data Belongs in the Range

Gossein's view is that trend data has a narrow job. Carry-over and core items need little of it. Statement pieces need none. The pressure point is the ratio of carry-over to novelty, which is where overproduction tends to start. A trend adopted by edgy consumers in one market is a small bet. The same trend reaching the mainstream across several regions is a different production call.

Design teams keep their judgement under this model. The data supplies evidence. A brand can follow a rising trend or go against it on purpose. Accuracy claims deserve a check. In its 2025 review, Heuritech says it forecast boat shoes at +39% and the category grew +42%. The figures come from the company's own retrospective. Buyers can fairly ask any vendor for the same test.

Autonomy Only Where Speed Matters

The adoption debate reaches well beyond fashion. In an earlier written reply to The Supply Chainer, Aera Technology described where automated decisions land first. Gonzalo Benedit, Chief Revenue Officer at Aera Technology, said: “We’re seeing the fastest impact of autonomous decision-making in areas where decisions are high-volume, time-sensitive, and constrained by cost, service levels, or availability. In these environments, even small delays or misalignment quickly translate into lost revenue, excess inventory, or waste. Instead of generating plans that must then be manually interpreted, coordinated, and executed, AI-powered decision intelligence continuously evaluates changing conditions, assesses trade-offs across functions, recommends the best actions, and executes decisions within defined governance. The goal is not to remove humans entirely, but to apply autonomy where speed, scale, and repeatability matter most.”

The two vendors sit at different points in the chain. One supplies the signal. The other executes the decision. Both keep people at the edges. Gossein warned that adoption stalls when AI arrives as a mandate and not as an answer to a defined need. Change management, she said, deserves as much budget as licences.

For planners, the lesson is practical. Trend feeds are getting easier to buy. The harder task is delivering them in the language and calendar of the teams that must act. Vendors that manage this will own a real control layer. The rest will add another dashboard.