top of page

Optimizing a Merchandise Hierarchy for Customer-Centric Fashion Retailing

A common question from fashion retailers and department stores is ‘What is the optimal way to define the merchandise hierarchy so that we can best leverage it for customer-centric purposes?’

Specifically the interest is – given the confounding factors in fashion of colours and sizes – at what level of granularity should the actual product be defined at, and what should the hierarchical roll up look like? At various times fashion retailers are interested in colours, or sizes, at other times it completely splinters the results of the analysis and a consolidated view is much more valuable. While this post reads in the context of RIC the principals should be interesting and useful to anyone thinking about how they can get a more customer-centric view on their data.

The fundamental question is easily answered – you should generally define the product (description) at the lowest level of granularity, as we will discuss below.

The second part of the question related to roll up is a bit trickier, and to some extent will depend upon how you are most interested in analyzing your data. However we will give some discussion in the context of some of the newer customer-centric techniques of analyzing data as are exposed in some of the modules in RIC. In RIC there is the flexibility to structure the data in several ways. The good news also is that if you decide you would like to structure in an alternate way all you need to do is replace the Products table with your new preferred hierarchy and everything will work under the new structure immediately.

We will illustrate with a hypothetical product the ‘Elle Tee Shirt’; and for simplicity assume this tee shirt comes in two colours: Blue and Red; and it comes in two sizes 10 and 11. It is manufactured by the fictional fashion designer Gloria Lauren (Brand).

The question we are trying answer in this case is ‘should I be defining the bottom level product ( ‘description’ field in the Product table) as:

1. ‘Elle Tee Shirt Blue Size 10’ (YES)

2. ‘Elle Tee Shirt Blue’ (NO)

3. ‘Elle Tee Shirt Size 10’ (NO)

4. ‘Elle Tee Shirt’. (NO)

The answer to this first question is relatively straightforward, we believe that in other than exceptional cases you should ensure that the bottom level of granularity is ‘Elle Tee Shirt Blue Size 10’. If you define it at any higher level than this you will lose the ability to add attributes such as colour and size, which even if you are not always interested in knowing, you will certainly want to use sometimes. Because RAP’s filters and product selectors allow the user to cut in at any level in the merchandize hierarchy, or on any attribute if a user wants to look at ‘Elle Tee Shirt’ in aggregate, they just select it at that level, so in defining in that way you will have the best of both worlds (assuming that the hierarchy conforms broadly with either of the options outlined below).

The next question to consider then is ‘which dimension do I want the roll up above the bottom level product to work on (the colour or the size)?’

Keep in mind there are two elements to defining the products in the Product table.

1. Product Hierarchy

RIC will allow you to build a single merchandize hierarchy (e.g. Department -> Category -> Sub Category -> Description) . These are defined by a ‘Level’ prefix when the data is prepared, and to reflect the nomenclature of Department, Category, Sub Category you would set them up like this: Level1_Department, Level2_Category, Level3_Sub Category, Description. (The names Department, Category, Sub Category are completely flexible, whatever you add after the ‘Level1_‘ will be displayed, if you call it Level1_World, the word ‘World’ will appear in your filter menus, etc.) Keep in mind description is the product description – always the lowest level of the tree.

2. Product Attributes

The second tool is that you have the ability to add numerous (it can be hundreds, though seldom would be) product attributes to each product. Product attributes are defined by custom labels. In this case we are only concerned with two attributes: colour and size, so they would be expressed as follows: customlabel_colour and customlabel_size.

TWO DATA PREPARATION OPTIONS TO CONSIDER AND DISCUSS

Now we will show two potential ways of setting up the data and outline pros and cons of either way, in general the restrictions will only apply to a small subset of the available modules, however this will help you understand where.

Option 1 – Colour Centric Setup: You would select this option if users are likely to be interested to drill down on the colour dimension more so than on the size dimension. Keeping in mind you can still filter by either or both dimension.

Option 1 – Product Table Schema for Colour Centric

Pros of Colour Centric Setup

– In the Category Drill Down Module, Category Helicopter View Module, Growth Opportunity Module and Product Penetration by Store Module you will be able to drill down by colour. – In the Product Customer Profile Module you will be able to see the profile of shoppers who buy Elle Tee Shirts by colour. – In the Product Substitution Module, Basket Contents Module, and Customer Related Purchases Module you will be able to show substitution/cross-shop colour against colour for a given SKU (e.g. how many people bought both red and blue of the same tee shirt).

Cons of Colour Centric Setup

– In the Category Drill Down Module, Category Helicopter View Module, Growth Opportunity Module and Product Penetration by Store Module you will not be able to drill down by size. – In the Product Customer Profile Moduleyou will not be able to see the profile of shoppers who buy Elle Tee Shirts by size. – In the Product Substitution Module, Basket Contents Module, and Customer Related Purchases Module you will not be able to show substitution/cross-shop size against size for a given SKU (e.g. how many people bought both 10 and 11 of the same tee shirt).

Option 2 – Size Centric Setup: You would select this option if you are likely to be interested to drill down on the size dimension more so than on the colour dimension. Keeping in mind you can still filter by either or both dimension.

Option 2 – Product Table Schema for Size Centric

Pros of Size Centric Setup

– In the Category Drill Down Module, Category Helicopter View Module, Growth Opportunity Module and Product Penetration by Store Module you will be able to drill down by size. – In the Product Customer Profile Module you will be able to see the profile of shoppers who buy Elle Tee Shirts by size. – In the Product Substitution Module, Basket Contents Module, and Customer Related Purchases Module you will be able to show substitution/cross-shop size against size for a given SKU (e.g. how many people bought both size 10 and 11 of the same tee shirt).

Cons of Size Centric Setup

– In the Category Drill Down Module, Category Helicopter View Module, Growth Opportunity Module and Product Penetration by Store Module you will not be able to drill down by colour. – In the Product Customer Profile Module you will not be able to see the profile of shoppers who buy Elle Tee Shirts by colour. – In the Product Substitution Module, Basket Contents Module, and Customer Related Purchases Module you will not be able to show substitution/cross-shop colour against colour for a given SKU (e.g. how many people bought both 10 and 11 of the same tee shirt).

A COMMENT ON THE CUSTOM LABELS FOR SIZE AND COLOUR

You will note that both configurations proposed above have two identical custom labels at the end: customlabel_colour and customlabel_size. This is very important, as it allows you to filter on either of those, regardless of which axis you would prefer your hierarchy slides along. This also underscores the point of why it is importantto define the product at the most granular level, if you did not, you would be unable to attach both of these attributes to it.

CONCLUSION

In just about all cases you will want to define your product at the very most granular level.

Relative to the most appropriate roll-up of the hierarchy, our belief is that for most fashion retailers the colour-centric option is likely to be the better as the actions and insights in the Category Drill Down Module, Category Helicopter View Module, Growth Opportunity Module and Product Penetration by Store Module and Product Customer Profile Module are probably more likely to be performed on colour type analysis than size analysis. However the post outlines the context of this, so that you are able to make your own decision.

In closing, two final things to keep in mind:

1. For most RIC modules it won’t actually matter which of the two options you go with, however you should definitely give thought to the above for those modules where it will matter so you can maximize value from RAP.

2. Just because you configure your data in the colour centric setup, you will still have full ability to FILTER by size, colour, or both and indeed observe volumes globally etc by colour or size – provided that you include the customlabel_colour and customlabel_size custom labels in the product table as described above.

If you are interested in obtaining an 11Ants Modules Library which outlines the 30 customer science modules which help make customer-centric retailing a reality in department stores and fashion retailers – contact us today.

17 views0 comments

Commenti


bottom of page