Feel like the world’s coming at you from all directions? It is. It’s the new retail world of omni-channel, where customer demand pops up in brick-and-mortar stores, on your and reseller websites, via mobile, and even from David Chang’s Fuku restaurant in New York City—where this past June users of Nike’s SNKRS app could purchase a limited edition sneaker, designed in collaboration with the restauranteur, by taking a photo of the menu within the app, which unlocks a sales portal.

Once the consumer places an order through one of these channels, the pressure is on the supply chain to complete that customer experience. Companies need an omni-pronged strategy to satisfy this omni-channel demand: smart forecasting, predictive analytics, data collection, and Inventory buffers.

Smart forecasting begins with the recognition that there is no way you are going to forecast demand for a product accurately: you’ll never come up with one exact number—for either sales in aggregate, by channel, or by individual item—across an increasingly convoluted omni-channel landscape. The best you can hope for is a consensus forecast based on the unbiased input of all stakeholders, supplemented by crunching sales histories of previous products that share attributes with the offering.

That’s where predictive analytics comes in. These analytics—often powered by machine learning—identify trends and patterns that reveal what’s selling, in what quantities, and where the demand originates. Data collection from all the channels feeds that analytics pipeline—the evidence-based input you need to understand sales velocity, product volumes, channel distribution, and the effect of promotions, to predict the sales flow and update inventory buffering and replenishment.

These inventory buffers, staged to meet service-level targets for each channel POS, give you the ability to confidently promise delivery when a customer makes a purchase. While it’s axiomatic in business to minimize inventory to free working capital and minimize obsolescence, you can get too “lean.” There are few things more frustrating to manufacturers and retailers than a product that garners customer acceptance, yet stumbles because of stock-outs and an inability to promptly fulfill orders.

This head-scratching omni-channel puzzle is why supply chain planning and optimization software vendors—like Quintiq, part of Dassault Systèmes—rely on advanced algorithmic solutions based on technologies like mathematical, constraint, and graphical programming. There is no way, in today’s omni-channel environment, for even an army of supply-chain planners to keep pace with everything that is coming at them from all sides. On the supply-chain front, the solution is automated software to quickly process and analyze the data, then take autonomous action or make recommendations that planners can evaluate through the lens of their domain acumen and knowledge of business strategy.