The UK heatwave, which boosted retail sales volumes – rising 0.7% month-on-month in August according to Reuters* – is becoming a distant memory, and we’ll soon all be racing into autumn. The latest data shows that approximately 80% of the weather impacts in fashion retail sales are observed in weather transition periods, such as Autumn or Spring.
For the consumer, the colder weather we’ve been experiencing prompts a change in shopping tactics, and for the retailer, attention turns to the all-important drive to clear unwanted spring/summer stock and fill its aisles and warehouses with the autumn/winter must-haves.
A look in some high street retail chains, as well as a glance at their websites and promotional emails, may have you thinking the industry is in a constant state of markdowns – there have, of course, been some high-profile cases of this low-price consistency having a major impact on bottom lines. However, the best operators out there are not being forced into heavy discounting – they are setting their prices at an optimum level thanks to good management and sophisticated forecasting techniques.
Online fashion house, Asos, is a shining example of this in action, as it continues to grow sales at home and abroad, as well as explore new ways to automate many of its merchandising and markdown decisions. But, on the high street where there have been well-documented challenging times, Fat Face is bucking the trend of slowing sales and store closures thanks in part to a steadfast full-price strategy and what CEO Anthony Thompson describes as the organisation’s “price integrity”.
The need for multidimensional forecasting
For more businesses to elevate themselves into the Asos, Fat Face school of happy retailers with happy customers, as opposed to the joining the East, Henri Lloyd, House of Fraser group of administration victims and uncertainty, multidimensional forecasting is required.
For a promotion to be effective, dynamic store-level demand forecasting is a key factor. Unlike using a traditional grade-based structure, as in normally the case for allocation and replenishment activities, promotional and clearance should be driven by understanding demand at store level.
Each shop has its own individual profile, factors and response rates, so forecasting should be tailored to each outlet, rather than decided centrally. Factors that influence demand by store and the need for stock to be recalibrated include short-term weather forecasts, where a store is located – i.e on the seaside or in an out-of-town retail park – local events/national events, competitor sales, and population demographics.
Retailers that are unable to factor these kind of items into their demand forecasts are likely to have stock in the wrong places, failing to achieve their promotional goals as a result, or disappointing customers with stockouts. Indeed, our data shows that using forecast-based and simulation-based automatic stock level management systems can decrease stockouts by up to 20% and increase service levels by up to 10%.
Margin management and dynamic forecasting
Ensuring it offers customers the right price to achieve maximum margin is something that Asos is starting to do at the highest level, using a blend of data science and skilled people power. But the wider retailer industry must start following suit because the constant low-price strategy and guesswork around markdowns can be best described as a race to the bottom.
Every retailer knows that a lower price can increase sales but may not increase margin. However, if clearance is the goal, margin may not be the driving factor. It’s important to understand the KPIs and make these the targets for the promotions, and the key is to strike the right balance between increasing sales, while protecting as much margin as possible.
Demand forecasting therefore needs to be dynamic, not simply done once prior to the promotion, and it has to be related to local stores and their individual influencing factors. As the promotion proceeds, the weather changes, local events come to an end, and the competition change their strategy, so dynamic demand forecasting is vital.
Recall and reallocation
At the end of a retail season, stock will be fragmented, resulting in poor assorted ranges across stores – there are ways of limiting this, but it’s hard to eliminate altogether. Poorly assorted ranges typically lead to low sales, as the assortment simply doesn’t look right, and will result in goods having to be sold at even more reduced-price levels.
An approach often taken is to recall items back to a distribution point and to re-allocate them to a more limited range of stores, and it’s of the utmost importance to forecast which stores will be those that the product should be re-allocated to. It’s crucial to make demand forecasts on an ongoing basis, because it’s such a fast-moving landscape.
If recall and re-allocation is not possible, then using demand forecasting to dynamically alter store minimum/maximum levels is important. This will ensure that any remaining stock is sent to the stores that need it the most.
There’s a real science to successful retailing today – especially because the old-style warehouse-to-store-to-customer movement of goods has been discombobulated by the rise of online retailing, the upturn in returned goods, and the need to fulfil customer orders in many new ways. Adopting a data science approach to forecasting can ease the burden, especially around the crucial changing of the seasons, and the progressive retailers in the UK have already realised it.
Robin Coles is Director at Inovretail, a retail data intelligence company.