As a firm specializing in analytics, and doing a lot of it for Retail, we have seen a lot of good work around customer, product, promo, pricing and store happening. The best of it, of course, is at the confluence of them all, but if one were to take up the Store as the focal point, there are a few areas analytics can help with that are impactful – adding 10 per cent to the revenue in incremental sales, or 15 per cent improvement in sell through rates, to better deployment of LSM budgets.
You would already have a store classification. However, store clustering can be a very different take on it. A fashion retailer with 300+ stores wanted a classification on the basis of one single variable of the AGR (Average Garment Rate) for the store. We felt this would lead to a lot of misclassification (where the wrong inferences are made just because you called a certain store an “A” store basis AGR), and there was room to look at many more variables.
So we went completely ballistic and threw 68 variables into the mix. Clearly overkill – I mean, how do you even assimilate 68 variables to cluster stores into A/B/C stores? A lot of these were customer centric variables (this brand has a loyalty program contributing over 70 per cent of its revenue) – variables like: Womenswear contribution, per cent customers who are over 40 years of age, Ability of the store to drive a repeat transaction, per cent Contribution of “Premium” products in each category and so on… for around 65 more variables.
When finally we did the classification, we reduced the 68 variables down to a much smaller set by eliminating non-contributing variables, but even then the big discoveries were that there is so much more to classifying your stores right – and things like per cent womenswear sold, or per cent product sold with a discount can have a large say!
Tracking Sell-Throughs/ Managing In Season Markdowns
Once the season has started, it’s about closely tracking sell-throughs over time. The time frame could be days or weeks depending on how fast your fashion is, but the logic is the same – If it ain’t selling quick enough what can you do, and when?
The starting point for this is straightforward, set up the baselines for what should be the sell through by the week, track the current season, and have a clear action plan if a product/ category dips down into a Red Zone. Can you push out a message to customers who bought similar products in the past with an attractive price point? Can you change the display? Can you run a store incentive? Or do you need to run a markdown?
Store managers can be equipped with 1) Reports and leaderboards and 2) Levers on what to do for their store.
Placing the right bets
Given the store classification, another thing you can do to really help improve margins is help the team place the right bets on what will sell. If you’re in the fashion industry and need to place bets on what to buy, and how much, here’s what can help:
Look at sell through trends of the last few seasons. There will be patterns that repeat, products that are missed opportunities because you ran out of sizes too soon, or, at the other end, products that you had to liquidate a lot of unsold stock of in the Sale. The buy decisions really are risky ones – you may err on the side of caution and have missed opportunities, or you may take a punt that goes wrong and damages margins. Here’s where analytics should be able to help you understand what has consistently been selling well that you can stick your neck out for.
Of course, all of this needs to be done at the store level. Some stores are just more likely to sell, for instance, large sizes than others – you cannot have the same formula for everyone!
Tracking and Driving L2L Growth
Like-to-Like or Same-Store growth is a key health metric. However, there are stores that taper off in sales after the initial excitement. The important thing here is to have a full breakdown of 1) What differentiates a good-performing store from the universe and 2) What needs to be done to help a poor performing store. This kind of Root Cause Analysis can identify the cause of the poor growth and the action quickly and accurately.
For instance one may realize that for some stores the problem is excessive discounted selling, for another it is lack of enough new footfalls while for a third it is lack of customer repeat. For each store the action to be taken varies – increased local store marketing push, increased CRM spends, more relevance and price point led communication.
Driving Decentralized CRM
At the store level, driving decentralized CRM can be a big one. The store manager is equipped with a set of tools – best customers, lapsed customers, best offers etc and can, in a contained fashion, drive his own CRM program with his best customers. This is straight out of the luxury hotels book where the GM of a property is expected to know his best guests and engage with them personally.
The store has to take on the job as part of its belief system, not an added chore. The people have to believe that this helps them, they have to demand it if a report comes to them late. They have to agonize over why they’re lower down in the leaderboards as compared to their peers. Which bring me to the last point:
Eventually, the only way to get your stores to get excited about what analytics can do for them is to go about it on two fronts:
Show them quick wins – build a case of how a store that was used as a pilot for analytics led interventions saw (for instance) an improvement in sell throughs from 55 per cent to 60 per cent as against other stores. Or saw an incremental 500 units being sold in a month etc.
Build the faith over time. It has to be consistently built and supported from the top. The practice of using data to drive decisions, of looking at and using reports, of being hungry for more data and insight to push your sales up.