Part 2: Welcome to the Real World
In my previous blog in this two-part series, I provided a whistlestop tour through the past thirty years of tech innovation in retail. In this second and final installment, I examine what retail companies can learn from that journey – and where they should be focusing their efforts from now on.
Retail companies today aren’t asking “should we use AI?” Instead, they want to know “where will it actually move the P&L, and where is it just glorified window-dressing?” One thing is abundantly clear. Where AI’s involved, the real opportunity is less about shiny prototypes and glossy PR, and more about wiring the technology into creaking estates – ERP, OMS, store systems, the lot – so that predictions turn into real-world decisions, orders, lorry movements, rotas and price changes.
Companies know, often from hard-won experience, that the real constraint on value is not the AI models themselves. It’s data quality, change, governance and a retail culture that still carries scars from earlier “next big things” that went nowhere.
AI: from buzzword to retail plumbing
Since it first emerged in the 1950s, AI has been through plenty of winters and springs. But the current wave – consisting initially of classic machine learning at scale and now generative models – is the first to permeate almost every layer of retail. That’s why AI should now be seen as plumbing, not magic: hard-working prediction and optimization engines powering everything from supply chains to pricing, content and service.
The non-theatrical applications are the ones that matter most. A few examples? Better demand forecasting and replenishment to cut stockouts and overstock while tightening working capital. Personalized recommendations, offers and content that shift basket size, frequency and loyalty, rather than creeping people out. And dynamic pricing and promotion optimization that lives in the trade-off between volume, margin and competitor activity.
However, snake oil in AI is still all too commonplace. It’s in the deck that promises “end-to-end autonomous retail” without mentioning data lineage, bias, governance, or how you persuade a merchandiser to trust a model over their gut instinct. Sadly, a number of UK retailers have fallen for that pitch. They now risk millions in failed AI investments, with chatbots and basic tools underdelivering – leaving sales on the table, customers frustrated and boards disillusioned.
As more and more companies have discovered, the durable value comes when AI is paired with process redesign and change management so that insights actually flow through to different decisions in merchandising, supply chain and store operations.
Snake oil, fakery and the things that stick
Looking back over thirty years of retail tech, the hype-cycle pattern is depressingly familiar. From Web1.0, Web2.0 and Second Life through to blockchain, NFTs and AI, a new technology appears, reference cases get inflated into narratives, capital and careers pile in, and metrics drift from reality. Then eventually the tide goes out and we discover who was swimming without a viable business model.
The silver lining? Every boom-and-bust leaves something useful behind. Web1.0 gave us the basic rails of eCommerce and online payments. Web2.0 normalized social proof, user reviews and marketplace dynamics. Second Life stress-tested ideas of virtual identity and digital economies. Blockchain made immutable, programmable ledgers part of the mainstream vocabulary (even if implementations often lagged). And now AI is busy binding all of this into continuous, data-driven optimization across retail organizations.
The sanity check that would have saved a lot of time and money in every era involves asking some boring but effective questions. How will customer behavior change because of this latest innovation? How will that improve revenue, margin, risk or cash? What has to be the case – across adoption, cost and regulation – for this to scale beyond a case study and a press release?
If you can’t answer those questions without handwaving or jargon, you’re probably looking at snake oil.
Getting ready for the next wave of retail…
So, what do retailers need to navigate this next phase, avoid the mistakes of the past, and harness real value? It sounds simple, but it’s really anything but: 20/20 insight so they can separate foundational capabilities – such as data, AI, automation and secure platforms – from fads dressed up as revolutions.
That level of insight can only be built up through hard-won, first-hand experience. Which is why companies need partners who can bring a retail go-to-market playbook that’s been honed through many years of stitching digital transformation into Fortune 500 and FTSE 200 businesses.
That doesn’t just mean deploying new AI models. Experience earned through rewiring ERP, OMS and store systems for end-to-end visibility is just as important. To put it another way, the ideal partner excels at the unglamorous bits as well – from data-cleansing, governance frameworks and change management to proving ROI through pilot-to-scale transitions.
Over the next few years, the best success stories will appear almost dull on the surface. But whether the results are better demand plans, fewer write-offs, lower shrink or happier store staff, the net impact will be far from dull, such as increased ROCE and cash generation.
…by laying down solid foundations
The bottom line: never bet your brand on worlds your customers don’t actually inhabit yet. Use each technology wave as an excuse to quietly refactor your retail plumbing so, when the noise dies down, you’re left with something much better than screenshots – a business that runs measurably better every single day.
