Lokad’s forecasting technology is uniquely geared towards the demand patterns observed in fashion. In this industry, collections, which primarily include short-lived products, are first-class citizens from the forecasting perspective. Lokad’s forecasting engine is capable of forecasting the demand of a new product, or in another words, a product that has never been sold before, purely based on historical sales data observed for former collections.
Unlike traditional approaches, this forecasting capability does not rely on manually linking older and newer products by telling the system which older product should be deemed as the most relevant for forecasting the new product. Instead, our forecasting engine relies exclusively on advanced machine learning algorithms to automatically detect the similarities that may exists between products, and to identify, by itself, which specific products are relevant for forecasting a new product in the collection.
This automatic detection of similarities relies on the numerous product attributes that are typically present in fashion: product type, product family, size, color, fabric, style, price point, brand, etc. While one might be worried about the amount of data required, our experience at Lokad indicates that catalog data, as it exists for operating the facets of an e-commerce front-end for example, is usually enough for achieving good results.
Traditional forecasting solutions that are based on manual pairing between products are too time-consuming to be effective - there are too many pairs to consider - since it is precisely the pairing that happens to be the main ingredient of the forecasts. Due to the ineffectiveness of this method, companies tend to revert back to their spreadsheets as the manual pairing forecasting solution fails to deliver the necessary value. Lokad tackles the challenge head-on, focusing on the core difficulty of the challenge, instead of passing the burden on to the users.
At Lokad, while we can refer to this forecasting process as product pairing, we do not assume that there is a 1-to-1 mapping between products from an old and a new collection. For example, one product may be split into multiple variants, which may generate cannibalizations. Then, another product might be truly “new”, with no close-matching past products. In such case, the forecasting engine falls back on broader considerations, such as the product category, family, brand or price point.
Advanced forecasting models that leverage correlations between articles and collections are a must-have for fashion companies, precisely because they have so much data to correlate in the first place.
Fashion retailers have gone global: the country where articles are produced is typically not the same country where articles get sold. Lead times are typically long and erratic and Lokad’s forecasting engine contains a native support for lead time forecasts. Just like collection forecasts, lead time forecasts are also first-class citizens at Lokad and come with their own specific patterns. For example, every year, the Chinese New Year tends to add 2 to 4 weeks of manufacturing delays for China; and this pattern, like many others, is taken care of by Lokad’s forecasting engine.
In addition, fashion supply chains must face many other numerical constraints: MOQs (minimum order quantities) and container batches probably being the most commonly found constraints in fashion. Forecasts which do not take any supply chain constraints into account are insufficient because if the suggested order quantities produced by the forecasts don’t comply with the purchasing constraints, then no purchase order (or at least no purchase order which makes sense) can be made. Lokad offers native support for an incredibly diverse range of purchasing constraints found in fashion, and we have developed a series of numerical solvers specifically geared towards the resolution of such constraints.
For example, our MOQ solver can tackle multiple overlapping MOQ constraints: there might be a MOQ at the product level (for example, a minimum of 100 units per product for every purchase order), another MOQ at the fabric level (for example, a minimum of 3000 meters of fabric per color), and a final MOQ at the supplier level (such as a minimum of $50,000 worth of merchandise purchased per order). Addressing all of these MOQs while keeping stock levels under control is a major hassle when MOQs are processed manually. Lokad streamlines the process entirely through numerical solvers that allow to identify the most profitable purchase order “envelop” that satisfies all given constraints.
Lokad’s platform offers programmatic capabilities that allow to integrate a company’s business drivers into the solution while also taking its supply chain constraints into account. The business drivers represent all the economic variables that may, positively or negatively, impact your business: gross margin, carrying cost, stock-out cost, etc. While the word programmatic might sound technical and scary, the reality is that fashion is subtle: selling one article may only make sense if a related accessory is also available. As a result, tackling domain knowledge requires a platform capable of dealing with practically any type of business insights. Buttons and drop-down menus systematically fail fashion companies when it comes to predictive supply chain optimization. Lokad, on the other hand, addresses the challenge through a domain-specific programming language.
If your forecasting solution does not have programmatic capabilities, your solution is less capable than Excel.
Can your business afford to have less than Excel?
Fashion at scale involves hundreds of stores, and tens of thousands of variants. Not only does the forecasting engine need to be able to scale up to millions of SKU positions, but it also needs to be fast enough so that calculations can be performed ten times a day if needed, as assumptions get adjusted, revised and corrected while preparing for the next collection. In fact, it is precisely because most fashion companies have only one shot per collection (since the initial purchase orders can only be adjusted so much) that an extreme agility is required on the forecasting side.
Our forecasting engine was natively designed for cloud computing. Unlike traditional solutions, the cloud is not an afterthought for Lokad: when data is sent to us, our system automatically allocates computing resources on the cloud, and gives back the results as soon as possible when the calculations are done. For a sizeable fashion company this may represent several hundreds of servers that are allocated within minutes. Auto-scaling, or in other words, the dynamic allocation of computing resources, is a critical ingredient of the forecasting engine. And this is precisely how Lokad can process terabytes of data in less than one hour.
A few years back, Lokad became one of the first companies to be charged by the minute by Microsoft for its usage of computing resources on Azure. Indeed, while the capacity to process data for hundreds of stores is very desirable, computing costs can be staggering too. Lokad keeps its own computing costs under control by aggressively de-allocating computing resources as soon as the calculations are done. For fashion companies, this feature of Lokad’s technology vastly reduces the operating costs, especially since forecasting tends to be a rather seasonal operation; precisely matching the fashion industry’s collection-driven approach.