Chris_Willis1_0-1631996878816.png

What is up with Predictive Scoring?

Chris_Willis1
Level 6 - Champion Level 6 - Champion
Level 6 - Champion

What is hotter in the industry than lead scoring?  

 

You guessed it… predictive lead scoring!  

Sometimes you may hear the term “intent scoring” which refers to the same methodology.  

 

Every conversation that revolves around predictive scoring typically goes into one of two extremes, with levels of moderation between these extremes.  

 

On one side are your hard-core predictive scoring believers who will die on the hill of “AI and Machine learning will revolutionize marketing and why on earth would you do anything differently?”  Trust the algorithm, they would say.  We have data that you don’t have and you need!  Behavior-based scoring from your own resources is so antiquated!  

 

On the other side are people who believe that predictive scoring is conjured up from some form of magic trick.  Others consider it a form of technical Gnosticism, with the methodology of the vendor being opaque and requiring you to check your brain at the door if you gave up your tried-and-true behavior-based scoring you have built and maintain in Marketo Engage.  After all, Marketo behavior scoring is reliable, we control the environment, and we are assured that the score is valid because we can see the activities appear in the activity log.  

 

After a vendor call with a vendor proclaiming the advantages of their intent scoring methodology, I decided to poll the internet to see what many of you think.  Here are the results:

Chris_Willis1_0-1631996878816.png

These unscientific (I mean, there’s only 163 respondents) results do support my hypothesis, and my own belief about predictive/intent scoring, which is. 

 

Leverage predictive scoring as a demand tool, but don’t treat it as your only source of truth.  

 

Which means, I believe there is, and should be a place in our methodologies for traditional scoring.  As we know, scoring and the “MQL definition” is a methodology that is used to determine sales readiness, and sales readiness with your brand should be derived from engagement with it, so I don’t recommend that you uproot your tried-and-true Marketo lead scoring for a shiny new predictive model, regardless of how well-baked the methodology is.  

 

If you do choose to add a predictive tool to your stack, I recommend that you follow the advice from this Adobe Summit Presentation on building a stack that moves your business forward by Jessica Kao, along with other thought leaders like Josh Hill and Kelly Jo Horton who have some great materials available.  At the end of the day, a predictive tool is like any addition to your stack, and should be evaluated with due diligence and scrutiny.  

 

The reason why many people believe that predictive scoring is a “black box” is that there are black boxes out there, and you want to protect yourself against buying a tool that will not deliver value.  Additionally, if the vendor has sales enablement tools that can compliment Marketo Sales Insight to provide the “Reason to Believe” to your sellers when a lead reaches their inbox partly based on an intent score, they have all of the information they need for research.  

 

Delivering value with predictive scoring integrated into Marketo Engage 

 

I present to you three ideas.  

 

First, leverage predictive scoring to trigger persons into targeted engagement program streams.  I would call this the “trust but verify” approach.  Your well-designed intent score will naturally, if the score is valid, lead people to respond to marketing offers for relevant solutions.  Such a trigger may look like this.

 

First, trigger from a change in the Intent Score that would indicate readiness. 

Chris_Willis1_1-1631996878888.png

Then, leverage this trigger to add the person into your offer stream within your engagement programs, or your program of choice for converting high-intent predictions.

Chris_Willis1_2-1631996878902.png

In the subsequent High Intent Offers stream or campaign, you present your offers to convert this audience into opportunities to your sales team. 

 

Second, include a trigger in your Interesting Moments program to inform your sales team that “high intent” has been identified by your predictive scoring tool.  Sales can subscribe to these interesting moments and leverage them, for existing prospects in the database, as an opportunity for contact.  Additionally, this is an activity that you can execute alongside your engagement communication, and sales can be alerted to add people to their watch list in MSI to evaluate how they engage with the offers you send, and perhaps initiate a contact based on additional engagement they see in MSI. 

Chris_Willis1_3-1631996878841.png

 

Finally, incorporate the predictive score into your lifecycle’s lead scoring model.  

 

I would not recommend co-opting the behavior score or the demographic/firmographic score to include predictive scores, as this can create a confusing “red herring” scenario. To ensure that you are communicating clearly as to your scoring methodology, I would ensure that these scores continue to be used for first-party behavior tracking and for valuing the data profile you are collecting through your data collection and enrichment activities.  


I would recommend having “Intent” as a third score in your model that can be used in a similar way as the demographic score for accelerating a lead to the threshold.  Thus, creating an “Intent Score” (I like this label better than “predicted” as it is more communicable) that becomes a supplementary part of your model, and can be used in MSI for the “Stars” visual if your sales team values predictive scoring in their selling motion. However, if your sales team is skeptical of intent scores, it would be recommended to continue to use the “Stars” for ICP fitness as is the best practice. 

 

A sample scoring threshold with intent could look like this, based on a 100 point MQL threshold:

  • Demographic/Firmographic Score:  Max 40 points.
  • Intent Score:  0 - 20 points
  • Behavior Score:  Remainder of the threshold

In this model, a high-intent activity would reduce the threshold by 20 points and would accelerate offer acceptances to the sales team more quickly as MQLs, and get these high intent buyers warmly into conversations with your company in a smooth way.  You may model your approach differently than this.  The maximum score you will want to give to Intent will directly correlate with the confidence level that your sales organization has in the predictive scoring, which can change over time. 

 

One thing to note, however, is if you do use intent scores as an MQL accelerator, you want to degrade the intent score (either based on the predictive platform or rules set by marketing operations) as intent, like “lead hotness” does have a limited lifespan, which if ignored will put your investment into predictive scoring into question.  

 

To Close

 

While there are plenty of potential pitfalls with predictive scoring, with due diligence there is value that you can gleam from these tools.  A solid vendor acquisition strategy and integrating the methodologies of well-designed predictive tools into an integrated demand strategy in Marketo Engage can deliver great results.  Be sure to architect your solutions well and do not silo your predictive tool, and as many in my survey indicated, provide as much data as possible to the process to get the best outcomes. 

177
1
1 Comment
Vladislav_Vagn1
Level 4 - Community Advisor

I am in the "Useful but I want more data" camp. We are using intent scoring in our tool but we only add intent scoring after they reach a certain behavior point threshold. We do this to ensure we still pass quality leads to our reps (the leads still have to take actions with us) but also to prioritize which leads they should call (higher intent will get called first).