Go figure workforce analytics
Lots has been written about Amazon Go and reinventing grocery, Wholefoods, and addressing perceived customer pain points. And yet if you talk to big retailers in the UK their #1 concern is labour cost, coming huge labour cost inflation, and poor labour productivity. And the difference in sales per employee in major retailers can be as much as 3-fold - this is astounding. With 2% profit margins, this is obviously unsustainable. This is against a backdrop of expected huge wage inflation (for lower paid workers) in the next 3 years and a potential post-Brexit labour shortage perhaps bidding wages up even more. So the real story of Amazon Go, and why it's electrifying UK retail, has nothing to do with customers...it's all about cutting labour costs.
We're looking at serious labour cost challenges in retail, coffee shops, forecourts, and QSR. And analytics can help figure out a plan! We see huge inefficiencies to how labour is deployed and applying analytics demonstrably can add a lot of value.
So can analytics be a solution to tomorrow’s workforce challenges? The long answer is expectedly complex, but still a yes. Analytics has evolved into an advanced science for deeper and better understanding of consumers. But its potential to enhance productivity, increase job satisfaction levels and manage workforce costs is under-leveraged. It is about time we use the power of analytics to optimise one of the crucial drivers of any economy – its workforce.
But why is this important?
First of all, economy has been changing and so has consumer demands and behaviours. Footfall to physical stores is declining fast. The variation and volatility of footfall as exploded. Many organisations also don’t have a clear view on proportion of in-store and on-line investment, resulting in sub-optimal store opening, closing and staffing. By time of day and day of week, labour demand in stores is often at odds with how it is staffed with temporal homogeneity. Staffing, rostering and roles (tills vs shelves vs warehouse) need to become much more flexible.
The concerns around Brexit squeezing both availability and quality of workforces are increasingly growing. The UK government is working on policies, which if implemented, will increase the minimum wage to £8.50 per hour in April 2019 (up from £7.20 per hour in April 2016). In the light of higher minimum wages promised in 2020 and possible Brexit labour shortage (sat writing this in a coffee shop in London), higher wages in return for greater flexibility, we know many retailers planning for a £10 per hour scenario.
Rostering is not an easy job even when it is “one size fits all”. It is no wonder that some organisations cringe at the thought of flexible rostering or on-demand deployment. Luckily, advance in technology and data analytics have enabled tailored workforce optimisation and unlocked huge potential. With their help, we can accurately forecast demand, simulate different scenarios, and search for the optimal staffing considering productivity, costs and employee preferences.
A critical role that analytics can play is understanding and predicting consumer demand. Consumer demand has a direct relationship with workforce planning, and more so with flexible working models. Efficient staffing levels are key to enhancing productivity and improving consumer satisfaction. For example, on a shop floor, staffing levels are determined by how busy the shop is. We encounter these models multiple times in our daily lives, but fail to recognise how inefficient most of them are. Busy, rushed, harried, frustrated and irritable staff in a restaurant, coffee shop, supermarket or a retail outlet is an example of a staffing model gone wrong. In one of the projects we did with a leisure client, we discovered that wallet conversion rate and spending are both strongly correlated with staffing level, but the client has never opened its tills to full capacity, resulting in large amounts of lost revenue.
Analytics can also help redeploy staff within a shift between tills, shelves, kitchen, warehouse etc. Based on past data, signals can predict where optimally we need to deploy fixed resources for greatest benefit minute by minute.
We have worked on examples creating staffing models for mobile workforces, e.g.: repairmen. Do you staff to peak levels of expended demand by region, daypart, day of week? Or staff to far lower levels and use contractor staff to fulfil higher demand? Or you can use analytic tools like a Monte Carlo simulation to exactly optimise the number of repairmen you need against a number of criteria in the model such as weather, location, costs, jeopardy of higher wait times versus agreed SLAs and so on.
The use of analytics to improve and strengthen staffing models should not be restricted to when we should be summoning extra workforce to meet sudden additional demand. Whether to over-staff or under-staff in different situations is a strategic decision. Analytics is able to model the loss in sales and customer satisfaction and the gain in cost savings at the same time. Then we can find the best, or most profitable staffing level an organisation should have at different times.
Staffing can hugely benefit from data rich, machine learning models that utilise huge volumes of historical shop data, different periods in the year, impact of broader economic factors and evolving category dynamics.
A simple question we've also answered with analytics is how much can we afford to pay our staff extra in return for flexible working? By accepting flexible working practices, and then using analytics to dramatically boost deployment productive, wage increases can be afforded. It's again another simulation of increasing productivity in different scenarios and consequently the higher labour rates than can be afforded in return for flexible staffing, shift lengths and patterns, locations, rostering and roles.
In quite a few of the engagements we have done recently with UK retailers, we have used analytics to measure and predict their staffing needs. These predictions have then been used to arrive at workforce models that combine both permanent and flexible arrangements. In one specific instance, our client offered £1 extra wage per hour to workers who were willing to work flexible hours. These workers were also given the freedom to choose these flexible hours for the job they loved doing the most. It was a win-win situation for both parties – Providing flexibility and added income for hours to employees and a happier workforce that benefited the business.
It’s worth mentioning that this is also not just a “tech job”. It is true that we find the optimal staffing level using advanced modelling and simulations, but ultimately we are dealing with human employees and human consumers. Uber was criticised heavily for surging price during terror attack in Sydney in 2014 and London in 2017. The algorithm was just purely following the higher demand, higher price staffing model, and it all went wrong. We need commercial sense and business experience to give our algorithms a steer and also to implement our results right.
The use of analytics to predict and plan flexible working needs is only scratching the surface. Scheduling of workforces in organisations is a much bigger and wider strategic challenge. We have discussed about how consumer demand influences staffing levels. There are other factors too, which include internal initiatives like restructuring, expanding manufacturing capabilities, broadening retail presence, integration of acquisitions, expansion into newer geographies and diversification.
The application of advanced analytics to create sophisticated, accurate and optimised workforce models is a hugely underdeveloped, but high potential area. As businesses increasingly get used to operating in uncertain and fluid environments, they would need workforces that enable them to ride these uncertainties. The role of analytics in creating, shaping and implementing these workforces will then assume critical importance.
Analytics can be a huge help with labour scheduling, rosters, and shift patterns in the face of labour cost pressure... go figure!
- James Walker, Partner and Global Head of Analytics
- Shane Mono, Associate Consultant