Data science for eCommerce Retailers

A series of posts on approaches to customer analytics

Sometimes the hardest thing about doing data science on retail ecommerce sites is deciding ‘where to start?’ 

 

Well, pick a comfortable chair, pull out your Brian Eno ‘Oblique Strategies’ deck of cards and kick off with this one. Your CFO will thank you, when you explain that you’ve boosted profitability with it.

 

After importing your eCommerce transaction data, it’s possible to detect changes in customer buying behaviour such as reduced shopping frequency. 

 

Selecting on intra-order time and then clustering often reveals interesting new segments and also some category-specific behaviours.

 

Customers can be dynamically allocated to a tiered retention campaign based on prior value and product choices. This does naturally assume you have these capabilities in your marketing team already. If not – jump to it!

 

For more advanced eCommerce operations, custom retention campaigns could involve personalised offers, or early access to new lines.

 

This data can then be combined with a data enrichment exercise, to ensure that your most valuable customers receive the personalised, attentive  experience they have come to expect when shopping online.

 

In the example below, a customer has switched to a different product type, and then significantly reduced their regular monthly spend, eventually leaving completely.

 

The AI can identify common behaviour patterns like this in your customer transaction histories, which might affect 100’s of customers in a similar way. It selects and presents the most important behaviour patterns for your attention and possibly intervention. 

 

Once you have more than a couple of thousand customers, it’s impossible to eyeball this kind of trend emerging in your data, unless you’re specially trained. This is where a data scientist, fully prepped and familiar with your data & customers, will come in very handy. 

 

Bonus tip: Try some time series decomposition – one hidden factor that’s often important for retail ecommerce managers is understanding the impact of Payday timing on certain customer personas. This can often be an important factor in timing your email offers to them. At the very least, it’s a good starting point. And don’t forget that US and UK payday patterns are very different beasts altogether!

 

 

Download Form Title

You're nearly there! Just enter your contact details and you will get instance access to the whitepaper