Training a workforce that changes every season

Ask three sources how fast retail workers in Mexico turn over and you get three different answers. A national supply-chain study by Logística de México and EGADE Business School, cited by Expansión in September 2025, puts retail at 23 percent a year. In the same article, Puma México's head of people says turnover in its stores was close to 95 percent at the start of 2025 and fell to 75 percent within the year. ANTAD's 2025 annual report, which tracks monthly turnover across dozens of member chains, lists 98.1 percent for supermarkets in Baja California Sur over January to August. The United States is no tidier: a 2025 UKG survey of 541 retail managers and executives puts typical frontline retention at 46 to 60 percent.

These figures count different populations, over different periods, with different formulas. They cannot be lined up against each other. But the disagreement tells you something on its own. In large retail operations, the people on the floor change faster than the industry can agree on how to count them.

That is the condition neo.education was built for. Soriana needed to train a national workforce that changes constantly, and at WAOF, a technology, product and creative studio that started in Monterrey, we built its corporate learning system from inside the operation. It has been running across the country since 2023 and became neo.education. The public numbers are these: 16,084 employees trained in 12 months, 91.8 percent of assigned classes completed, more than 800 stores, and 58,146 registered users since 2023. What follows is less about those numbers than about the design choices that a constantly changing workforce forces on anyone who has to train it.

A catalog assumes someone who stays

Most corporate learning platforms are organized as a catalog. There is a library of courses, a person browses it, enrolls, and works through a path over months. That model quietly assumes three things: the learner has a desk, the learner has time, and the learner will still be there when the path ends.

On a store floor, none of the three holds reliably. Emergence Capital estimated in 2018 that 80 percent of the world's workers are deskless. When it surveyed about 1,500 of them in 2020, over 80 percent said the devices they were given for work were desktop PCs or laptops, and 60 percent were unsatisfied with the technology provided. A 2024 Lighthouse Research & Advisory study for Schoox found that 91 percent of frontline workers want training on a mobile device, and that people with less than six months in the job are about 50 percent more likely to need to learn something new every week or every day.

Then there is time. In a 2025 survey Fountain ran with Lighthouse, covering nearly 2,000 frontline workers, Fountain reports that 43 percent leave within their first 90 days. That figure comes from a vendor's summary and deserves some caution. Even so, the direction is clear. If a large part of a store's new hires are gone within a quarter, a learning path that takes a quarter to reach them reaches almost no one.

This is why the neo.education design starts from the hire instead of the catalog. Training is assigned at hire based on role, with no catalog to browse. The content is five-minute videos made for people working on the floor. The design choice is simple: the system decides what a new cashier or stocker sees first, so the store manager does not have to. Puma México's people director made a related point in the Expansión piece: a store manager cannot also be an HR director, because their attention is already split between inventory, sales and people. A catalog quietly hands the curation job back to that same manager.

Proof is a requirement, not paperwork

The second force is legal. Mexico's Federal Labor Law obliges employers to train their workers. The 2013 STPS rules that implement it require a certificate of competencies, the DC-3, to be issued within 20 business days after a worker passes a course. Employers must report the list of certificates to the ministry within 60 business days after each plan year, electronically if they have more than 50 workers, and keep the last year's certificates available for inspection.

For a single office, that is a filing exercise. For a chain with hundreds of stores and a workforce that keeps turning over, it is a volume problem. Every week, people who must be trained and documented arrive, and people whose records must survive their departure leave. Spreadsheets kept store by store do not scale to that.

Two of neo.education's four parts address this directly. One is verifiable training records for operations that need proof. The other is compliance tracked by operational unit, not just by individual user. The second matters more than it sounds. When the individual is the unit, a high-turnover store looks like a constant stream of strangers, and accountability leaves with each one. When the store, district or region is the unit, there is a stable thing to hold accountable: this location, this month, this share of its people trained for their roles. Whether a given record meets a specific legal format is a question for each employer's compliance team. The design point is that the evidence exists, store by store, when someone asks for it.

What completion proves, and what it does not

A completion rate of 91.8 percent is the kind of number that invites questions, so it is worth being precise about what it measures. It counts assigned classes completed. Like any completion rate, it depends on who is counted in the denominator, and in an operation with high turnover that choice matters.

Completion is not competence. Watching a five-minute video about receiving a delivery is not the same as receiving one correctly at six in the morning. The Lighthouse study names the biggest constraint for workers new to their jobs: limited opportunities to practice and apply new skills. No platform solves that on its own. That is why the thesis behind neo.education separates two jobs: the platform handles distribution and evidence, while the content is what gets people through it. Short video helps people finish. Whether they then do the work better has to be measured on the floor, not in the learning system.

The record outlives the worker

There is a less obvious consequence of building training this way. When every hire is assigned training by role, on day one, in a specific store, the learning system ends up keeping one of the more continuous records of who actually works in the operation: who arrived, for which job, where, and what they were shown. In an industry whose own turnover statistics range from 23 to 98 percent depending on who is counting, that kind of record is worth having. It does not explain why people leave. It does show, store by store, how much of the operation is being rebuilt at any given time.

That suggests a few working rules for any operator whose floor changes every season. Design for the person who started this week, not the one who might be there in a year. Hold the store accountable rather than the individual, because the store stays. And be exact about what a training metric proves, because completion is evidence of reach, and reach is only the start.