An optimized and accurate assortment plan is the first step retailers take towards ensuring the sales season is profitable and not plagued by stock-outs, overstocks, customer churns, etc. During the present times of high inflation and supply chain glut, balanced assortment planning is all the more important for retailers to bring in the right product in the right quantity to the right place, and at the right price, to meet customer demand.
However, retailers continue to struggle to solve the mystery around an ideal assortment planning strategy and hence, today we will focus on five things that are preventing them from building a precise assortment plan:
- Not using the right (or any) system
Even in 2022, one of the most favored tools used by retail assortment planners to manage their buying process is MS Excel. It is an undeniable fact that MS Excel is a wonderful product with myriad features. However, it’s safe to say that it was not designed to manage an iterative process that involves teams of planners, buyers, merchandisers, and allocators working together and taking swift decisions to meet tight deadlines, while trying to ensure the most profitable assortments.
A major challenge with Excel is that the data it holds or processes needs to be entered manually. Each planner responsible for managing a part of the business will run queries on databases and enter those numbers in custom-created excel templates. This further complicates assortment decisions that are made across multiple channels (outlets, full-line stores, e-commerce and wholesale) each with different templates and formulas.With so many moving parts, there is a lot of room for manual errors like mistyped numbers, missing or additional zeros, and incorrect formulas. Any of these small mistakes can roll up to millions of dollars of misallocated receipt budget. Additionally, during the buy review meetings, as data is collated from multiple sources, reconciliation is a nightmare, and cross-channel/functional teams lose valuable days in data validation.
- Working with basic store clustering methodology
With the greater availability of socio-economic and demographic diversity data, making sure that the right products are sent to the right stores is critical. Store clustering helps get this done. However, most retailers still use simple clustering methods taking in basic parameters such as ‘Overall Sales $’, and ‘Store Area’, to determine which stores should be grouped together. The number of clusters remains constant over years, the cluster composition stays the same for all seasons, and cluster revisions are few and far between.
However, when stock-outs and overstocking of the same product appear across different stores, it is clear that the store clustering methodology should be further refined. Clustering needs to be at the product category level at the minimum, and preferably at the category cross product attribute level, which ensures the weightage is not just on how much each store sells but also what it sells. Two stores that have identical sales volume in a particular category may still have associated customer groups with preferences very distinct from one another. Additionally, multiple KPIs such as GM%, ST%, and APS should be included in the algorithm to make the clustering more robust. Moreover, clustering needs to be run as often as every season to capture recent trends and seasonalities. This not only helps in assortment planning, but also connects the buy with the in-season allocation.
- Using traditional forecasting algorithms
At its core, assortment planning is the process of forecasting customer demand and deciding how a retailer can manage that demand profitably. That makes accurate demand forecasting one of the most powerful tools for a retailer to improve their assortment planning. However, assortment decisions at most retailers are still taken based on a macro-level hindsight view of how the LY assortment performed, and at best by applying traditional forecasting algorithms on that historical data. But with rapid changes in product preferences and consumption patterns, especially during this post-COVID period, this just isn’t enough.
Inaccurate forecasting not only affects the buy for that season, but also impacts the accounting for Carryover or BOP of next season. As buy plans for one season are finalized long before the previous season has actualized, the BOP information is only as good as the forecasting against the inventory. If the forecast is too optimistic, one will buy too much and if the forecast is too low, one will not have enough BOP and will need a rebuy.
We typically find that using advanced forecasting models such as “Regime Change”, “Drift Detection” etc. along with recent data and dozens of macro exogenous inputs (i.e., inflation rate, consumer confidence index, Google mobility info) are essential to accurately predict demand. This is critical for fashion retailers, as trends change rapidly and what worked last year is no longer relevant. One needs to be ahead of the curve by looking into trends data available through multiple sources such as Instagram influencer pages and WGSN reports.
Retailers that utilize available internal and external data sources coupled with cutting-edge AI algorithms, indeed get far closer to generating accurate demand forecasts for all SKUs at the store, style or any hierarchy level across their entire lifecycle.
- Suboptimal receipt budget allocation
Typically retailers plan the allocation of their receipt budget based on standard KPIs such as ST%, APS, GM%, and determine product listings and delistings. However, a lot of opportunities get lost as the planners do not have the right tools or data to provide deeper insights into product performance.
With more granular data, planners can hone in on areas of opportunity created by local customer preferences that often go unnoticed. A particular product might be an under-performer when viewing overall numbers and delisted in the traditional approach. But that very product might be a clear winner in a certain group of stores, due to the socio-demographic profile of the customer base near those stores. In this scenario, rather than delisting the product, it would be better to continue the product, provided they can still meet the MOQ constraints in those stores.
For new product launches, planners tend to be extremely bullish and launch across all stores with a deep inventory. However, mapping of a new product with like products in history through a similarity mapping algorithm results in data-backed decisions to balance the depth of product across stores.
- Running the planning process in isolation
As with most business processes in any industry, retail assortment planning is intertwined with other upstream and downstream business processes such as financial planning, product development and lifecycle management, purchase order creation, space and allocation planning, and more. However, if one looks at how these are managed in retail organizations, these processes are often siloed and disconnected. Typically updates in one silo are only reflected in others by manual intervention. As most of these processes are iterative by design, this makes them very inefficient and error-prone.
Take for example assortment planning. This starts by seeding the financial budget from the Merchandise Financial Planning (MFP) system. However, it is extremely common that the MFP budget undergoes multiple updates while the assortment planning has already started. With every minor update in the MFP, the budget data in the corresponding assortment plan(s) must be manually updated and subsequently, the assortment plan(s) needs to be run again.
How can Impact Analytics help?
All of the challenges mentioned above can be addressed with the right toolset to execute the buying process. Teams should be equipped with an Assortment Planning Software that:
- Connects with their database to generate necessary hindsight reports automatically
- Integrates with their MFP system to fetch the category level budget and targets
- Allows dynamic store clustering while allowing the planners to choose from an exhaustive list of clustering parameters
- Optimizes budget allocation based on accurate forecasting across sub-categories, product attributes and store clusters
- Integrates with other systems such as MFP, Item Creation, Product Lifecycle Management and Space Planning Systems
- Generates a comprehensive buy roll-up report automatically for the buy review meetings
IA’s Assortment Planning Software AssortSmart delivers all of the above, while still providing planners the flexibility to make necessary adjustments at every step, based on their business understanding. Assortment planning, at its best, is an optimum amalgamation of science (the AI and the data analytics) and art (the business understanding the planners gather based on years of experience). This is what we help our clients achieve.
If you want to create optimized assortment plans that will improve your inventory turnover, reduce markdowns and as a result, improve your bottom line, Impact Analytics can provide the perfect solution.