Retail runs on a combination of repetitive work and moments that make the difference with the customer. Inventory management, pricing, staff scheduling, order processing and part of customer service take up a large share of the hours worked. In addition, there is work that revolves around judgment on the spot: how a shelf is arranged, how a complaint is resolved, how a salesperson responds to what a customer has not yet said. The outcome of a sales moment is shaped by circumstances that differ per store and per chain: assortment breadth, the degree of physical versus online sales, the margin per item and the extent to which customers compare prices before walking into a store.
As long as inventory management and pricing were mainly driven by experience and manual checks, their quality was a result of staffing levels and employee experience. Whoever had more people on the schedule generally had fewer mistakes. The moment systems largely calculate inventory and price themselves and only bring exceptions to an employee, what wins there changes. It is then no longer about how many people are planning, but about the quality of the data the system runs on and about how quickly deviations are recognised and corrected.
This is already happening today at some retailers, and not at all of them. The difference lies in the extent to which a company has its inventory and sales data in order, in the willingness to leave decisions to a system within set margins, and in the size of the assortment: with a limited assortment, automation is easier to validate than with a broad and rapidly changing offering. Chains with many locations and a standardised assortment often lead here, while specialist or highly seasonal stores continue to rely more on human judgment.
Within the same department, the three categories overlap. Order processing and basic inventory replenishment are tasks that a system can increasingly handle independently. Price adjustments and the compilation of staff schedules more often fall into the intermediate category: the system makes a proposal, a manager approves or rejects it, with reason. Building a customer relationship on the shop floor, handling an unusual complaint and assessing local preferences remain, for now, human work. Exactly where the boundary between these three categories lies differs per store format and per role, and this is also why generic statements about 'AI in retail' offer little grip for a concrete comparison with competitors.
If two retailers claim the same delivery time, the same service level or the same pricing, the question is no longer only who executes that best, but also who has structured that process in such a way that the system can steer it independently. A chain that largely automates inventory decisions can respond faster to demand fluctuations than a chain that still makes the same decisions through a weekly meeting. That difference is measurable in turnaround times and error margins, not in intentions or ambitions.
This is precisely where a peer group matters: only by comparing your store format with retailers that are comparable in size, assortment and sales channel does it become visible whether a difference in delivery time or pricing stems from a different approach to AI or from something else. A white space analysis then shows on which dimension none of the comparable parties score strongly, and where there is therefore room that has not yet been claimed.
The underlying question of which work in your own company can actually be taken over by AI is answered per task with the work scan from FTE TO AI. If that question touches on personnel decisions, separate legal requirements apply; the work scan describes the work, not the consequences for people.
The way AI changes the comparison between competitors differs greatly per sector. In hospitality, the shift mainly plays out around scheduling and staffing, in the IT sector around delivery speed and code quality, and in financial services around risk assessment and acceptance speed. The agricultural sector also shows how seasonal work and automation intersect, which shows parallels with seasonal assortments in retail.
You can start by naming the points on which you believe you win against your competitors, and testing those claims against what can actually be substantiated with evidence. The free dimension check from FTE TO AI is the starting point for that: you indicate where you believe you are winning, and see which of those claims hold up and which still lack substantiation. The full benchmark, with the peer group and the evidence matrix for retail, is under construction.