AI is reshaping how consumers shop. Most retailers still treat it as a marketing tool.
Personalisation, demand forecasting, pricing and customer service all run on AI now. The harder question is what happens to the decisions that actually move a consumer business: assortment, pricing strategy, where to play, the brand and channel bet.
Request a Strategic Briefing →Retailers and consumer brands have adopted AI fast, mostly as a layer on top of e-commerce and marketing: the recommendation engine, the forecasting model, the chatbot, the campaign generator. That work is real and worth doing. The trouble is that the decisions which actually shape a consumer business, what to stock, how to price, which markets and channels to bet on, get handed to a commodity AI that answers from the same public data every competitor feeds it. If the input is the market average, the strategy is the market average. Gildoni installs the Havruta Methodology (formerly the Think Partner Methodology) into how retail leadership teams reason with AI, so the big category and market bets are stress-tested rather than auto-completed. This is not a personalisation engine. It is the reasoning discipline at the altitude where the strategy is set.
AI has moved to the centre of how people shop
This is not a forecast. It is already the operating reality of the sector.
The shift is happening on both sides of the counter. Consumers now bring AI into the buying journey itself: nearly half use it to help them decide what to buy, ahead of the moment they ever reach a retailer's own channels. And the people running brands feel the pressure to keep up, while telling researchers their biggest blockers are joined-up systems and the in-house expertise to reason about any of it.
So the money follows, but cautiously, and the spend tells you where AI sits in most retailers' minds. It is treated as a slice of the technology budget, a tool to be bought, rather than a question the leadership team has to think through. The numbers below show a sector adopting at the surface and hesitating at the core.
Three readings of where the sector actually stands.
of retailers allocate 5% or less of their technology budget to AI today, and only 39% expect it to pass 10% within three years. AI is still framed as a line item, not a leadership question.
of brand executives report persistent challenges across channels and systems, and 51% name limited AI expertise as a barrier. The constraint is reasoning capacity, not access to tools.
of retail and consumer-products organisations are still in the initial stages of generative-AI integration, well short of the value they expect from it.
The adoption is real. The reasoning has not caught up.
AI is not a magic wand. You must test your solution to know whether it works.
The personalisation engine is winning the easy argument
Here is what happens in practice. AI lands in retail where it is easiest to point at: marketing, e-commerce, the customer-service queue. A recommendation engine lifts conversion. A forecasting model tightens inventory. A campaign tool ships in a tenth of the time. All good, all measurable, and all of it the same speed-of-output story playing out across every competitor at once.
That is High-Speed Waste with a personalisation engine bolted on. Faster outputs, more campaigns, more variants, none of it touching the decisions that actually set the trajectory of the business. The volume goes up. The judgement underneath it does not move.
Meanwhile the decisions that decide whether a consumer business wins get a different, quieter treatment. A merchant asks a commodity AI how to build next season's assortment, what the pricing architecture should be, which channel deserves the next round of investment. The machine answers, fluently and confidently, from the public data set that every rival is also drawing on. If the input is the market average, the strategy is the market average. The output reads polished, and it points everyone in the same direction.
It is worth being honest about why this happens. Personalisation and forecasting have clean metrics and obvious owners. The big bets, where to play, what to stand for, how to price against the category, are slower, contested, and harder to measure. So AI gets pointed at the easy wins and kept away from the place it could actually change the outcome.
Shipping more recommendations is the easy part
The part the martech story leaves out is this. Every vendor now sells the same promise: more personalisation, sharper forecasts, faster content. Almost none of them help a leadership team reason through the assortment call, the pricing strategy, the where-to-play bet, once that question is on the table.
That is the real gap, and it is not a tooling gap. Put a category-defining question to a commodity AI and it hands back the consensus, the option everyone with the same data would reach. At the altitude where the bets are largest, that is the Mirror Principle at its most expensive: if the reasoning going in was generic, the strategy coming out is generic, however confidently it reads. A recommendation engine can lift a conversion rate. It cannot tell you whether you are competing in the right place. That is a different discipline.
A commodity AI fed the public market data returns the answer everyone with that data would reach: the market-average strategy, where retailers converge. The same data run through a leadership team's reasoning produces a position that is theirs. The red is the distance between an auto-completed strategy and a defensible one.
What the Havruta Methodology installs at leadership level
The Havruta Methodology is that discipline. It changes the default behaviour of the machine from agreeing with you to reasoning with you, which is precisely what a category or pricing bet needs.
The Flip
The Flip puts the machine on the other side of the question. Instead of confirming the assortment plan or the pricing move, it argues against it: where is this just the consensus, what is the category not seeing, what would have to be true for this bet to fail. The leadership team gets challenged before the market does the challenging.
Ground Truth
Ground Truth anchors the reasoning in the business's own reality, its real customer base, margins, range and channel economics, rather than the generic market data an AI defaults to. A strategy built on the public average is the one every competitor is already building. Your own ground truth is what makes the answer yours.
Decision Velocity
And Decision Velocity lets the team decide at the pace consumer markets actually move, compressing the path from question to a position the leadership owns, without surrendering the judgement to the machine.
The fuller account of how all of this works is on the methodology page.
What this is not
This is not a personalisation engine and it is not a demand-forecasting tool. It is not a recommendation system, a pricing-optimisation platform, a martech stack, or a customer-data platform. It is not AI training or general AI literacy. Those tools are a separate market, and a useful one. This is the thinking underneath the bets they cannot make for you.
It changes how the leadership team reasons about the decisions it already owns: the assortment call, the pricing strategy, the where-to-play question, the brand and channel bet.
Where a leadership team starts
The methodology is installed along a ladder, and a leadership team enters at the rung that fits.
Most begin with the Eye-Opener Workshop, a half-day in which the team sees the shift on its own real work, a live pricing or assortment question, not a demo.
A leadership group embeds the practice through the Havruta programme, taking the discipline across the team and into its operating rhythm.
A single high-stakes question, a category strategy, a pricing architecture, a channel bet, can be worked through Advisory Havruta.
How a General Manager reasons with AI
For the leader who owns the P&L day to day, the role page takes the same discipline to the seat where the market and category bets actually land. A Strategic Briefing is how to decide where to begin.
Go to the general-manager pageStrategy questions about AI in retail
Is AI in retail a marketing tool or a leadership decision?
It is both, and the distinction matters. As a marketing and e-commerce tool, AI is already embedded in personalisation, forecasting and service, and it should be. As a leadership question, it is something else entirely: how a team reasons through assortment, pricing strategy and where to compete. Most retailers have the first part well in hand and have barely started the second. Treating AI as only a martech layer leaves the decisions that set the trajectory of the business running on a confident, generic answer.
Why does AI give retailers a generic strategy?
Because a commodity AI answers from public data, and that data is shared by every competitor in the category. Ask it how to build an assortment or where to price, and it returns the consensus: the move everyone working from the same inputs would make. That is the Mirror Principle. If the reasoning going in was generic, the strategy coming out is generic, however polished it reads. The fix is not a better model. It is anchoring the work in your own ground truth and making the AI argue against the plan rather than confirm it.
What is the risk of using AI mainly for personalisation and forecasting?
There is no risk in doing those well, and you should. The risk is stopping there. Personalisation and forecasting produce faster, more measurable output, which makes them feel like the whole AI story. But they do not touch the assortment, pricing and channel decisions that decide whether the business wins. Pouring AI into the easy wins while the big bets run on a generic answer is High-Speed Waste with a personalisation engine on top: more volume, the same judgement underneath.
Is this a personalisation or demand-forecasting platform?
No. It is not a recommendation engine, a forecasting model, a pricing tool, a customer-data platform or any other piece of martech, and it does not touch your stack. Those address the output: what to recommend, what to stock, what to charge. This addresses the thinking: how a leadership team reasons through a category or market bet so the answer is truly theirs, anchored in their own customers and economics, and stress-tested before the market tests it for them.
How should a retail leadership team use AI for strategy?
Start by treating the big bets as a reasoning question, not a prompt. Anchor the AI in your own ground truth, your real customer base, margins and channel economics, rather than the public market data it reaches for by default. Then make it argue against your plan: where is this just the consensus, what would have to be true for the bet to fail. Decide at the pace consumer markets move, but keep the judgement with the team. The forecasting tools tell you what is likely; this is how you decide what to do about it.
Where should we start?
With a Strategic Briefing, or with the Eye-Opener Workshop, where a leadership team sees the difference between instructing AI and reasoning with it on a live pricing or assortment question of its own. From there the path depends on whether you are setting strategy at the top, embedding the practice across a leadership group, or working a single high-stakes bet.