Vision Express

Undoubtedly, the recent pandemic has configured a new world for all of us and the retail industry isn’t an exception. After going through a series of lockdowns, the current context still represents a big challenge for this sector in many senses. One of them, perhaps the most important for operations management, is labour.

Chart 1: Labour Cost in the Retail Industry

Cost of labour input (ULC) per unit of output, inflation-adjusted.

Source: Office for National Statistics.

In fact, staffing costs have increased sensibly in the recent years relative to both pre-covid levels and the whole UK economy (see chart 1). This is mainly due to a combination of low real gross value added (GVA) and high growth in employment costs, especially fuelled by the rise in national living wage and compensation of employees (including social contributions). Indeed, according to the Low Pay Commission, 45% of all minimum wage jobs last year were in just 3 occupation groups: retail, hospitality, and cleaning & maintenance.

Chart 2: Retail Productivity

Unit of real output per hour (OpH), chained volume measure.

Source: Office for National Statistics.

To simplify our point here, let’s assume for a moment that all employment costs are given (exogenous) to any retail business (price taker), then, the key driver to minimise unit labour cost (ULCs) is productivity, defined as the output per hour worked (e.g., number of dispenses completed by a sales assistant). So far, the retail industry has detracted sensibly from productivity growth over the last years, not only dropping sharply after the pandemic outbreak but also performing clearly below the average of the whole UK economy (see chart 2).

This is when labour planning comes on stage, as it can directly influence retail productivity. It’s worth mentioning that staff optimisation is not about reducing staff per se but having the right number of colleagues at the right time in the right place, which is much more complex and denotes several challenges and questions to respond to. Is it possible to forecast the exact number of customers that will visit our store in a particular day and hour? What is the ideal workforce to reach sales potential? What is the right mix between part-time and full-time workers? Which labour regulations or practical constraints affect our ideal planning? What is the impact of omnichannel on store labour?

Unfortunately, we won’t be able to find all the answers in this article, however the spirit of labour planning can be summarised in three major steps. First, we set store potential essentially as a function of labour and customer traffic (other things being equal). We should be aware of the “dual” effect of labour on store profit at this point. From a top line perspective, staff availability will improve sales in line with the queuing theory, which states that the more sales assistants the fewer customers will leave the store without being served. On the other hand, from a bottom-line view, labour is one of the most important costs in the retail structure. Hence, we have to find the right balance between labour availability and costs, adding labour as long as their contribution to store sales (output) exceeds their incremental in cost.

It’s worth mentioning that staff optimisation is not about reducing staff per se but having the right number of colleagues at the right time in the right place

The second step in this process is demand forecast to estimate expected sales. Inputs commonly used for this purpose are footfall, seasonality, calendar days (school holidays, bank holidays, religious celebrations, etc.), weather conditions and promotions. Some businesses may also have additional inputs, such as appointments booked, that could serve as a short-term proxy for customer traffic too. Then, different (supervised) machine learning techniques can be used to run our forecast (linear and nonlinear regression, autoregressive time-series, generalised additive models, artificial neural networks, gradient-boosted decisions tree or random forest, among others).

Thirdly, ideal staffing levels can be defined after incorporating labour impact on sales and forecasted customer traffic into our optimisation model. As a result, we will be able to identify stores that are either under or over staffed by day of the week and review labour planning accordingly. Yet to set a plausible working shift in a more granular detail (hour-by-hour), extra considerations must be taken, such as peak hours, part-time mix or employee skills.

To sum up, labour planning has become critical for every retail business to drive productivity, offsetting recent increments in employment costs. Good news is that the boom of big data combined with novel algorithms and ML tools can nowadays support retailers to reach best (feasible) rota planning.

What is more, the provision of schedules to colleagues ahead of time (including their personal preferences and skills) have proved to boost not only productivity but also store performance (top line sales) due to colleague retention and satisfaction (eNPS).

Last but not least, innovative staff scheduling methods also take store managers little time to implement since the bulk of the work is performed centrally by computers. This also allows them to spend more time with both store colleagues and customers, while improving the robustness of their rotas.

Finally, as always in retail, there are still many challenges ahead. Management of unexpected absence and sickness or usage of digital tools aside, flexibility is probably one of the main focuses; not only because of higher productivity (through better labour adequacy) or help to attract new talent to the business, but also because of the new market trends and regulations (the recent flexible working bill is a clear example). There are definitely challenging and interesting(?) times ahead for the retail industry… and labour planning!