Lead scoring blends itself well into customer journeys or contact journeys. These models are designed to drive intelligent customer journeys. Getting it right can be a difficult process. The presenter at IBM Amplify walked us through basics on content scoring and how to then make those models more cognitive (e.g. how a machine can give you better insights and thus better target a lead or contact.)
Understanding Lead Scoring
Scoring is assigning value to a data point. You take the numeric values of events, demographic information, etc. and assign the aggregate value. You then take these and put people into groups. Most systems allow for multiple models.
Note: the numbers are very arbitrary. But you do want to have at least three to five groups in any given model.
Fact: Lead Scoring is nothing but a souped up version of dynamic segmentation. The main difference is that the lead score is being fed in real time with dynamic data. Activity then drives a significant component of the score. This is about scoring an individual so you need unique identifiers to this users like email address, cookie, etc.
Myths and Facts
- Lead scoring is only applicable to B2B.
- Lead scoring models are difficult to create.
- They are really simple to build…just as easy as building a segmentation query.
- One lead scoring model is sufficient.
- If you are SMB then sure, it’s true but if you are larger with a wider range of customers then you need more.
Part of the myth is perpetuated by the fact that you call it a lead which suggests B2B. Some are starting to use terms like contact scoring instead.
Lead Scoring Data
You can use three types of data:
- This is the typical data that you get appended or acquired from cart data (address for example)
- Behavioral data
- Includes explicit like direct email opens and form completions
- Also includes implicit like web browsing and referring links.
- This is one of the main reasons people use a campaign management system to track this sort of thing.
- It’s about likes and shares, which conveys some information.
Is one model enough?
The short answer is that it depends. In some B2B settings, yes, it’s enough, but you assume a single product or service. The number of products and business units will drive additional.
A consumer model will contain the following points:
- email open – 1 point
- email clieck – 5 points
- Browse site – 5 point
- Product click – 5 points
- Site visit over 5 minutes
- Complete a form
Setting up a Cognitive Lead Scoring Model
Dynamic segments are based on scoring models. You build the segment based on the score value. The contact associates with the segment and the segment then drives entry into a journey. Order:
- Score puts you into a segment
- Segment drives you into a journey
- reaction or lack thereof, drives the ongoing communication
This would put you into a low touch, medium touch, and high touch programs. Scoring will accumulate over multiple touch points.
What Does Watson Bring?
Notice how Watson is being embedded into all the elements of this three key components.
When connecting scoring to Watson Marketing Insights you would:
- Watson adds a cognitive element to traditional scoring to improve the model.
- improved models – more accurate models
- Accurate models facilitate real-time personalization.
Watson Marketing Insights is a dashboard tool. It cherry picks out the stuff you need to take a look at. It’s a red, green, yellow approach. It gives you a deeper understanding of the customer behaviors. It starts with some common KPIs.
How to implement this model:
- Get the right data into the right place. You need to feed the data into the core Campaign Management database
- Build the segment
- Now think about your “programs.”
- These are highly personalized, segment based, and the fact that relevant content is based on behavior.
- Now you can have Watson Marketing Insights start to look at this data, segments, and programs.
- Note: this will not come as a core part of IBM Campaign Management.
- Scoring is never “done.” It becomes a big chunk of a person’s time to work on this.
- Improved targeting
- Higher campaign response rates
- Do more with less
- But yes, it’s going to be more work in the beginning to setup the models and programs.