How can Nextail prevent unusual demand patterns from distorting our data?

Alba Vacas, Nextail Change Manager, provided insights to answer this Nextail FAQ on preventing data and forecast distortions.

Unusual demand can distort data & forecasts, but it doesn’t have to

Events like store closures, extreme shifts in demand, and changing channel preferences can skew and distort retail data, leading to repercussions in forecast calculations and inventory management decisions. But they don’t have to.

Below are four ways in which Nextail helps retailers mitigate unusual demand that threatens to distort data and forecasts and even contextualize data retroactively.

1. Preserving demand calculations when stores close

The issue: Store closures can create perceived drops in demand

Naturally, when a physical store is temporarily closed and no one is shopping in it, perceived demand at that store will drop to zero (in your data) and so will the demand for all of the products being sold there.

Closures per se aren’t usually a problem if a store is closed regularly according to a specific schedule (e.g. between certain hours, on national holidays), because you are likely uninterested in capturing demand during those periods.

However, if a store closes unexpectedly due to external factors (e.g. pandemic, extreme weather conditions, etc.), and especially if it will remain closed for an undetermined amount of time (e.g. store renovations), the perceived drop in sales in could distort data and future demand predictions. For example, when the store reopens, calculations might predict that you need fewer stock replenishments than you actually do.

The fix: Hit the switch and “turn off” your data

Nextail includes a “Store Closures” functionality that prevents calculations from considering these closed periods as days with zero sales. In this way, retailers can preserve demand calculations by leaving out periods of time when there was no demand due to store closures.

Nextail will take closure dates into account and essentially “ignore” them when calculating future SKU-store demand. And this functionality works for both times when store closure dates are predefined and when they will be closed for an undetermined amount of time.

If an estimated closing period was preconfigured but the actual dates differed, teams can adjust closure dates to reflect the reality of the closure the day before reopening.

2. Contextualizing data when demand increases drastically

The issue: Sudden spikes in activity can cause “false positives” in longer-term forecasts

Store closures aren’t the only factor that can potentially distort data and future forecasts. Retailers must also account for the opposite effect: unusual or uncommon increases in demand across channels.

For example, consider the uplift in activity on ecommerce channels during the pandemic, or moments of pent-up demand in some physical stores upon reopening. Many retailers perceived a rise in certain product categories as well, such loungewear or face masks.

While a pickup in demand can be positive on the bottomline, it can have the opposite effect of a store closure on future forecasts, causing demand to seem higher than it really is or will be and potentially triggering a bullwhip effect.

Take the face mask example. People likely aren’t buying as many as they were a year ago, and going forward, the downward trend will (hopefully) continue. The high demand perceived in 2020 should not, therefore, have been reflected in this year’s projections or those going forward.

The fix: Treat these periods like promotions

Luckily, uncommon periods of high demand are actually common retail phenomenon during promotional periods. That’s why retailers can use the Nextail promotions functionality to contextualize periods of unusual increases in demand, just like they would a promotional event, to further preserve their forecasts.

The promotions functionality allows you to “tag” exceptional periods of temporarily high demand so that your calculations will not consider this period of time when forecasting for both the short and long-term future.

In the short term, tagging a period of time as an “event” alerts the system to deseasonalize the sales. This means that the system knows that the sales performance of this period was exceptional because it was caused by an event, much like when there is an actual promotion. Therefore, retailers won’t run the risk of overstocking stores.

In the longer term, tagging a period of time as an “event” means that all the sales that “resulted” from the event will be filtered out when calculating the seasonality coefficients for the following year. In other words, it won’t be considered when calculating the general yearly sales trends used to forecast future demand.

For example, when a major UK retailer reopened stores after pandemic-related shutdowns, pent-up demand caused a spike in unseasonal demand. By tagging this period of distorted demand in Nextail as a “promotion”, the increase in activity will not bear any weight in future demand forecasts.

3. Filling in the gaps with comparable products

The issue: Gaps in data leave blindspots

In some cases, it’s not so much a matter of data reflecting too low or too high of demand, but rather periods of time for which data is simply not available.

This is something we often see when we begin working with customers that haven’t had a way to automate data collection to the degree to which Nextail does.

Other times, data is missing when we are dealing with new products being introduced that we have no previous sales performance for such as a new family or sub-family. How can we ensure the success of new product introductions when we have a limited or non-existent picture of demand?

The fix: Use comparable items to fill in the gaps

Nextail applies computer vision, a type of AI, as well as data from the master files to extract product attributes (e.g. family, price range, product description, material, color, print, etc.) for identifying comparable products and store performance. That way, if you’re about to introduce a product which doesn’t have any stock or sales history to draw from, Nextail will build a preliminary demand forecast based on similar products selling now or in the past.

Normally, Nextail customers use this First Allocation functionality when introducing new items for the first time. But it can also help you avoid certain data and forecast distortions when doing so.

The Comparables functionality allows you to draw from specific periods of time for identifying similar products. You can do this to avoid choosing products being sold when demand was distorted or to avoid a period of time entirely, choosing similar items only from outside of a period of distorted sales.

For example, imagine that you want Nextail to identify products that are comparable to a new dress you’re about to introduce (e.g. color, sleeve length, price point). However, you want to make sure that Nextail doesn’t look for comparables from SS20 due to pandemic-related distortions. You can instruct Nextail to search for comparables within, or avoiding, a certain time frame so that it won’t draw from periods with data distortions.

4. Working from the bottom up: Agile retailers are already at an advantage

While all retailers should be concerned with maintaining the integrity of their data, it’s worth mentioning that retailers that take a bottom-up approach to their merchandising (you can read a whole post on that here) have an advantage over retailers that take a more traditional route. This is because they can rely on much fresher, more granular data to get closer to real-time demand patterns and recalculate them dynamically as they change throughout the season.

In other words, traditional top-down approaches try to predict future demand through the lens of the past by basing inventory calculations solely on simplified past performance data. This makes them more susceptible to data distortions and they have less flexibility to adjust course whereas agile retailers can constantly anticipate demand and modify their strategy as necessary.

In the “Ask Nextail” FAQ series, we’ve compiled the most frequently asked questions that we receive. Team members from various teams across Nextail help to answer these questions, bringing expertise from their work helping retailers understand, implement, and be successful managing the radical change that comes with digital transformation in retail.

Want to learn more about you can avoid forecasting and data distortions with Nextail? Get in touch!

Nextail FAQ series artificial intelligence

How does Nextail use Artificial Intelligence and Machine ...

Luis Pérez Vázquez, Nextail Data Scientist, co-authored this frequently asked question. It’s not magic, it’s data sci...

Nextail FAQ series demand forecasts

What data does Nextail use for forecasting demand?

Giona Capurso, Product Manager, provided insights to answer this frequently asked question. In the face of uncertaint...

Nextail FAQ series identical products

How does Nextail manage multiple SKUs for identical produ...

David Vicente, Nextail Integration Project Manager, provided insights to answer this retail merchandising question. M...