Being Local

The Science of Clustering and Range Assortment Planning

Introduction

As a traditional retailer it’s time for change but this article is NOT going to be spouting the negatives about the high street instead, we are going to talk about what you can do to change and even better, to be Local.

We all know that online retail is massively data rich, however so is traditional retail, every retailer has it, the question is are you using it correctly?

Online retailers use data like their lives depend on it and to a certain extent, it does. Consumer specific advertising floods our social media pages and specific products that we want are there in front of us to buy.

“Data is King”

While a traditional retailer cannot update their POS (Point Of Sale) every time a consumer walks in the front door as a website can (although I don’t think this is far away with smart phone tracking) what we can do is replicate consumers specific local needs on a shelf.

To achieve this goal a retailer needs to undertake a clustering exercise and build their range based upon this.

What are the basics of Clustering?

To understand clustering, we need to talk about “Local”. Local is and should be the most discussed thing in retail right now (other than the B word). It is concentrating on the core needs of a particular customer group allowing retailers to stock appropriate ranges rather than having to stock everything.

The most important part of the cluster analyses is being able to understand consumer behaviour with the ability to determine the over/under indexing of sub categories, sectors and product assortments. Once this can be established it is very easy to ensure that you can identify the correct range for each store cluster.

The theory behind Clustering:

Top Down:

Traditionally “Top Down” attributes have been used to determine store clusters. This may be by Geographical Regions, Store Size, Sales Value, Store Locality, Demographics or Supply Chain Depots.

An example of this would be in a central London supermarket’s washing detergent department.

Consumers in this department would have quite specific buying habits driven by demographics, transport and size of dwellings.

Typically, central London residents have:

1. Flats in high rise buildings

2. Low car ownership and high public transport use e.g. the Underground

3. High Yearly income

In this scenario high end branded detergents in low volume would be driving the largest proportion of sales. Driven by the high income and public transport i.e. they would not want to drag 10kg of detergent on the Tube and upstairs to their flat.

There is always an exception to the rule however circa 80% of the sales will be driven through high end and low volume products. It would therefore be wasteful and costly for the supermarket to be grouped in with similar size stores from a more rural or different demographic which could arguably have the reverse buying habits.

The above works upon an assumption of a demographic and would probably suffice a simple category such as washing detergents. However if you take a soft drinks category, it is nearly impossible to determine what drinks will be sold from top down attributes, especially with more bespoke products such as Irn Bru. This is where Bottom up Clustering comes in.

Bottom Up:

“Bottom Up” attributes look directly at the product sales mix through Category, Sub Category and Sector attribute levels, plus many others. This identifies product trends, consumer behaviour, shopping missions and localised/regional assortments.

Not only will this achieve operational savings but will also satisfy bespoke local consumer demands.

An example which we mention above is Irn Bru. Irn Bru is the top selling soft drink in Scotland, beating Coca Cola and Pepsi and therefore without question is ranged in large quantities across Scottish supermarkets.

However in parts of England and Wales where it is far from the top, based on a top down attribute you would assume that it would be stocked in smaller quantities if at all.

However this would be an assumption that would steer you down the wrong path.

In England there are pockets within regions, where local demand for Irn Bru is high. If we looked into the detail behind this you could find that there is a high Scottish resident population or maybe in fact the local area just likes Irn Bru?

Whatever the reason to be truly local and to fully understand your local demand, Bottom Up attributes must be used as well as applying top down to give you an understanding of your clusters.

So how does it work?

Here at OSP Retail we use RELEX clustering to identify similar purchasing patterns. It uses Cohesion, Separation and the number of Stores in a Cluster to create up to 25 cluster schemes, ranking the most accurate number of clusters. The solution will complete a hundred iterations to ensure the final ranked number one cluster scheme is the most accurate.

Once completed, different versions of this can be added and the solution will compare up to five different schemes that use the same Store, Product and Performance data. From this the most accurate cluster scheme and clusters can be identified.

What insight will Clustering deliver?

Once the solution has determined the most accurate cluster scheme, analysis can take place.

This analysis will identify what type of clusters have been created, an example of these could be:

Cluster 1 – Large stores, High Spend, Northern

Cluster 2 – Large Stores, Low Spend, Southern, Food to Go

Cluster 3 – Small Stores, Low Spend, Impulse

 

Overlaying product and demographic data, further analysis can take place, this will allow additional information to determine the type of clusters.

These could include –

Cluster 1 – Weekly Shop, Higher Price, Larger Sizes, Premium Products

Cluster 2 – Food to Go, Shopping Mission, Category Flow

Cluster 3 – Daily Shop, Lower Price, Smaller Sizes, Private Brand

 

Finally using demographic data by post code we can show –

 

Cluster 1 – Premium, High income, Large Houses, Older Age

Cluster 2 – Mainstream, Medium Income, Families with Children

Cluster 3 – Economy, High Ethnicity, Large Families, Low Education

 

Armed with all this data it is possible to “name” the type of clusters that will be used in determining the assortment and space. This data allows stores to really focus on local demand.

Other considerations:

There are many considerations to make in using the clustering analysis to determine the assortment, these may include – Core Range, Choice, Regional & Local Products, Media Activity, Fresh Food Wastage, Availability, Logistical Efficiency & Restraints.

In addition, working with many stakeholders in the Category Management process, retailers may want to consider a more streamlined Pricing, Promotion and Space Strategy using different options for Premium versus Economy Clusters.

Recognise the oppertunity:

Technology is here to help retail work much leaner, allow for targeted trading and fundamentally a huge step change from a ‘one size fits all’ approach. Together this concentrates on driving local demand to keep up with the digital world.

If clustering is not already part of your category management process then utilising OSP Retail to do this is the simplest and quickest way of making this happen now. With significant experience from different sectors of the retail market, we can deliver you a very different view on your retail assortments.

Here at OSP Retail we are passionate retailers and are here to help. Stop letting disrupters, disrupt, make a change, you have the power to be Local!

Information is useless if it is not applied to something important or you will forget it before you have a chance to apply it. – Tim Ferris.

“OSP’s unique approach and delivery have provided Wolseley with the foundation, tools, and ability to build on and deliver continued success. Their skill to quickly build the necessary foundations based on business requirements, develop processes and coach individuals have catapulted Wolseley into a new era of true Customer Centric Merchandising. OSP’s ability knows no bounds.

Charles Milgate, Wolseley UK

“OSP has been providing essential ongoing support to Wyevale as it makes the significant transition to planogram all of its retail space, as a result, we have started to maximise our retail space. Their abilities, business understanding, and expertise across different platforms have supported the planogram capture of circa 70% of total SKU count so far”

Peter Aylott, Visual Merchandising Manager

“OSP’s merchandising solution revolutionised our space planning facilities. A thorough day’s training gave us all the tools to get started creating effective planograms. If we ever have any questions the support is always quick, detailed and friendly.”

Sarah Dixon, International Marketing Executive

“OSP stepped into the breach at short notice, quickly acclimatised to the business needs and delivered a superb merchandising outcome which is now being rolled out”

David Weaver, Retail Consulting & Interim Management

We guarantee to give you the tools and information your business needs to make smarter decisions, function more efficiently, and save money

Ready to find out how we can help you?

OSP RETAIL LTD

Bakers Loft
Quay Street
Lymington
SO41 3AS
UK

SOCIAL MEDIA

SUBSCRIBE TO OUR NEWSLETTER

OSP RETAIL LTD

Bakers Loft
Quay Street
Lymington
SO41 3AS
UK

SOCIAL MEDIA

SUBSCRIBE TO OUR NEWSLETTER