You can tell the school summer holidays are here, because stepping out of the office at lunchtime the shops are remarkably quiet. Meanwhile, over in more touristy parts of town the same chains are busier than ever. This Figure It Out looks at some of the ways the holidays affect retailers and some of the ways business analytics can be used to understand the issues.

Seasonality of demand has a big impact in retail, driving stock levels and staffing requirements. Last week’s FIO used predictive analytics to forecast cricket performance, similar methods can be used for demand forecasting in retail. More complex algorithms can forecast with multiple levels of seasonality (day-of-the-week, month-of-the-year, regular holidays and events) so retailers are prepared for the summer holidays.

Sometimes seasonal trends can be hidden in the big picture, but customer analytics techniques can reveal the detail underneath. For example, volumes of food shopping may not change much in the summer since people always need to eat, but the types of foods people want will change – more ice cream and less hot stew. Segmentation is a technique to split the customer base by characteristics that affect their purchasing habits and reveal underlying trends. Groupings can be defined by who the customers are (such as age, gender), what they want (the types of products they purchase, their needs) and how valuable they are (the revenue they generate, how recent their last purchase was, how loyal).

Once trends have been identified, retailers can use the information to prioritise marketing to the relevant customer segments. For example, an office supplies company may spot that their business custom will drop off during August, but it’s a great time to launch a back-to-school campaign for students and parents to stock up before the new term starts. This targets a different customer segment to pick up the shortfall of the core business.


Obviously geography is another category that will affect seasonal trends. Geo-demographic segmentation could show that our office supplies retailer would have more success with the back-to-school campaign in stores near schools than in central London stores catering fully to the business market. This could raise ideas such as redistributing staff seasonally or in extreme cases, why not take a break with your customers? If you have ever visited a Swedish city in July, you will find many shops and businesses closed all month as they and their customers flee the city for the beach.

Once you understand why the customers are buying seasonally, you can use live analytics to react to any changes this time round. From looking at historic data you might expect to see higher sales of barbeque meats in the summer, like burgers and sausages, but EBLEX, the organisation for the English beef and lamb industry, reports that it is actually roast and stewing cuts that are seeing the biggest increase in sales from a year ago – an increase of over 40% – so people are going for warm comfort food after all. Retailers with a live analytics capability and an eye on the weather might have spotted that July was a cool and unsettled month, the worst July since 2000, and adjusted accordingly for the poor barbeque weather.

Considering and reacting to the seasonal demand characteristics for a just a couple of meat products across a few stores might not prove too much of a challenge. However, when you consider some national supermarkets in the UK will have 12,000 stores often segmented into different store classifications and 40,000 product lines it starts to get complex. For such national retailers to meet their age old mantra of ‘right product, right place, right time’ they need to ensure each product variant is in each store at the right volume on a daily basis; that equals 12,000 x 40,000 = 48,000,000 decisions they need to make per day, 7 days a week. Add in the differing seasonal demand characteristics across different parts of town and across different parts of the country it becomes impossible to meet that mantra without the use of Retail Analytics.

The Operational Research team has experience in working with our retail experts to develop analytical solutions to problems such as these, including a tried-and-tested merchandising diagnostics to rapidly identify issues affecting a retailer.