Leveraging Data: Why It’s Important for Enterprises to Separate Matching from Prediction

 In Data Science, Predictive Intelligence

Over the last six months, the predictive analytics space for B2B marketing and sales has seen a real positive spike in momentum from a vendor advancement and customer awareness perspective (see Google Trends graphic above). The discussion with business-to-business CMOs and their teams is starting to change from “Is it even possible?” to “How can I get this into my sales and marketing architecture?” But along with every emerging SEO and buzzworthy term (yes, we’ve all experienced it with ‘Cloud’, ‘Analytics’, ‘Data Science’, etc. over the years), companies flock to position and align their wares under that buzzworthy category. As the leading player in the Big Data-driven prediction world, and focused from our inception on predictive intelligence and buyer intent, we’re excited that our space is getting so much attention.

Of course, this is also why we find that there is an increasing amount of confusion among enterprise marketing leaders who are trying to sort out what capabilities the various vendors are lumping under the buzzword “predictive.” No doubt, there is a place for each vendor’s solution depending upon the customer use case. However, when talking about predictive, it’s just not an apples-to-apples comparison although many vendors try to argue this fact. Check out our previously published 14 Questions to Ask Before Investing in a Predictive Vendor (under Buying Guides) to help you open-up the kimono a bit when having such conversations with vendors.

Separately, here is a perspective from 6sense product management on how to think through what does ‘predicting’ really mean from a technology perspective:

(a) Matching Is NOT Predicting.

Most successful athletes train a lot. Hence if you train a lot, you will likely be a successful athlete.

Sounds like a logical statement, yes? But is it really a causal “prediction?”

The approach of utilizing firmographic characteristics – company type, size, industry, purchased technology, location, job role, job function, etc. – of a past customer win, and finding prospects that match similar firmographic characteristics as high potential leads and accounts is exactly that: a match. Not a prediction. We typically refer to this as “buyer profile fit” which indicates how closely a prospect matches to an enterprise’s typical customer profile.

Buyer profile fit and matching techniques have their place, and for many companies, this step is achieved at some level by looking at basic trends within their CRM data sitting in Salesforce, SAP, and other such tools. Some more advanced marketing organizations can even choose to think through much more fine-grained matching algorithms that consider a larger number of characteristic variables like we do in the 6sense product offering.

However, we think a majority of marketing organizations are going to skip ahead to seek out the joint power of matching and insight from predictive analytics, and thus many purely ‘matching’ focused solutions that claim predictive capabilities will remain unattractive to competitive enterprises.

(b) Predicting Involves Matching.

Most successful college athletes typically practice 4 days a week, maintain a balanced diet, work out their core muscles every other day, and wake up early. Hence, if you do the same while you are in high school, there is a 75% likelihood you will be prepared to be a college athlete hopeful. If instead you follow all these rules, except for waking up early, you have 60% likelihood.

Does this statement seem different than the matching example above? It should, because now we have some concrete concepts of time, behaviors, and likelihood of outcome, factors that were not considered in the matching technique.

Predictions involve data analyses that look for behavioral characteristics that depend on time and significant (but transient) event types. Predictions look at the strength of these various behavioral signal matches and are able to discern just how much something is a contributing factor to a successful probabilistic outcome and a high predictor of success at a specific point in time.

In the sales opportunity prediction scenario, behavioral characteristic inputs can range in complexity – from marketing activity and campaign participation, information which is typically available within marketing automation systems like Marketo, Eloqua, and others, to web and search driven demonstrations of intent, which are more advanced and require additional data. These input differences and modeling approaches lead to various flavors of truly predictive solutions even among existing vendors who are doing more than matching. This leads us to the third point:

(c) Not all predictions are created equal.

This seems like a rather obvious assertion, but not for the reasons you might think. Yes, there are major differences in the types of data inputs, volumes of data consumed, and even data science modeling techniques.

However, the truly impactful reason that drives differences is how the various vendors manage the data pipeline. This is crucial for any solution claiming to leverage Big Data or to offer a solution that depends on it. . Having inherent DNA within the company to build such a data platform has to be present from the start, and natively architected into the solution.

To migrate from matching techniques to Big Data prediction techniques is possible, but not without significant investment, retrofitting tradeoffs and skill changes. In fact, not having the right talent from the get-go focusing on this kind of scale is a huge challenge. The know-how that is required to ingest massive volumes of data daily, clean it to remove the noise and garbage, process it to make it useable in algorithms, and ultimately study its value for solving the unique customer’s business problem is more fiction than fact for many vendors transitioning from matching into predictive vendors. This is probably the biggest reason why startups like 6sense, across various predictive intelligence applications, are highly disruptive to incumbents in their respective categories, uniquely positioned to provide value, and more in-line to meet what customers expect to get out of predictive analytics.

Ultimately, the ability to make the distinction between matching strategies and predictive intelligence solutions is crucial for business success. To return to the athlete analogy, there is no replacement for talent, grit and a bit of luck for an athlete to thrive. However, it helps to have the right coach on your side and the right tools in your training kit to increase your chances for success.

With this perspective, enterprise marketing leaders can – and should – choose the right matching or prediction coach with extreme caution and based upon specific marketing and sales goals. As the product owner for 6sense’s intent-driven predictive solution, my team is continuously focused on becoming the central part of every B2B enterprises’ marketing athletics.

Recent Posts

Start typing and press Enter to search