September 4, 2013 ↘︎

One really useful thing that we can do with all the data that streams past us on a daily basis is create two (or three) dimensional scoring models.

The purpose of a scoring model is to better understand where visitors are in their lifecycle.  By assigning values to different visitors, as they come across the digital channels over multiple visits, you can do things like rank them in order of importance. Or, you can look for common behaviours amongst common scores (or different scores).

One dimensional scoring

One dimensional scoring is quite straight-forward.  You’re assigning points to visitors as they do certain things, like view a product page, or interact with a calculator etc.  The more points they have, the more engaged they generally are. Your digital analytics platform is then able to report back on each visitor and their score, this is also useful when you want to target them with something.

Ideally though, you also want to ensure your analytics platform can “bucket” the scores, such as 0, 1-25, 26-50, 51-75 etc, because with that you’ll be able to see scores based on different segments such as, traffic sources and campaigns.

So now you have your one-dimensional scoring; and it’s a good place to start. It should immediately begin to provide some insight to you around what customers vs. prospects are doing differently, and from it, you’ll be able to produce a histogram of scores.

Below illustrates the scores by various metrics: Visits, Visitors and Revenue.  We can see that the majority of Visits and Visitors have scores of between 41 and 320, and that purchasing tends to occur when their scores are between 161 and 2560.  In fact, the highest revenue is attributed to scores in the 641-1280 range.

So, one dimensional scoring is useful to start the ball rolling, and helps us to determine whether the customer or prospect is ready to begin a conversation with us.  The higher the score, the more ready they are.  Notice that there are a bunch of prospects in the ‘None’ and ‘<10’ buckets, these are likely to be brand new visitors who haven’t really done anything on the site.  The guys between 40 and 160 are the prospects, it is these ones that should most definitely be engaged through some type of targeting capability, such as Adobe Target.

Two dimensional scoring

This is, quite simply, a second factor to the scoring.  If we think of the above as activity-based scoring, now we add in demographic scoring – that is, assigning points for different attributes of the individual.  This helps us to determine whether they’re the ideal customer/prospect to sell to.

Many of these can be gathered through forms they may complete, or through 3rd party lists, or through their actual activity on the site.

There are lots of different types of demographic scores that could/should be considered, including:

• Role
• Social network participation
• Social connections
• Personal interests
• Location
• Visitor type (prospect, customer, vip)
• Products purchased
• Revenue
• Email subscriber

Once again you attribute points for different demographics.  You can also have negative points – points you remove from their score if they do certain things, like unsubscribing.

Now you end up with a second dimension to your scores, customers and prospects neatly fit into each “bucket” with the more active and sales-ready customers generally being top-right hand side of the model.

The following is an extract over a few more months of data, showing visitor scores across both behavioural activities (the x-axis) and demographics (the y-axis).

Once again you can see the bulk of customers that have purchased (in the red square), and the bulk of the interested opportunity is in the blue square.  Over to the far left with the very low scores on both dimensions are individuals who are a long way from purchasing anything.

So, armed with this information you can now target individuals who have a particular behavioural score, and a particular demographic score, with relevant content to push them more towards purchase.  This might be via email or social channels, or even through onsite behavioural targeting. Retailers can further optimise onsite search to restructure offers based on this.

So, are we there yet?

Well, you’re getting there; but as you discover more opportunities from your data, you’ll want to extend your digital experience even further and make it more relevant to your customers.