During my tenure as a COO at several companies (most recently Marketo, and previously with Business Objects, and Crystal Decisions), marketing and sales teams shared the goals of top of funnel growth.
A recurring theme in those efforts was the ongoing debate (and sometimes more than debate) regarding lead quality and lead quantity.
I have seen (and have personally been guilty as charged) excellent go-to-market executives put hours of work into rationalizing results and writing/rewriting rules and definitions regarding lead quality, lead flow, and other top of funnel metrics.
This entire lead quality quagmire occurred largely due to the fact that human hypothesis and retrospective viewpoints on what constituted a “quality lead” was all leadership could rely on to create the rules/definitions
Bias, ego, compensation, and other external dimensions would virtually always cloud the efforts to reach clarity.
I personally found the experience to be unproductive, energy sucking, and sometimes divisive (both as a participant and an observer).
All in all, clearly there was and still is opportunity for improvements that would impact execution, resource management and morale when it comes to leads.
Obviously, everyone wants both higher quality and a higher quantity of leads. Sales teams are measured on close rates (perhaps more impacted by quality) and marketing teams measured more heavily on quantity(where quality measures can by default act as a governor).
You can see how this creates problems - as well intended as each group is, each team is motivated by a slightly different but highly dependent target.
I believe through the use of quality data and the elimination of human hypothesis, this problem and the related incongruence can be minimized if not eliminated.
The opportunity to introduce machine learning to this mix is obvious and the possibilities are exciting!
The Problem: Subjective Human Bias
There are many models, methods, and techniques for determining the quality of a lead.
Most recently, marketing and sales teams have relied on software to determine the quality of their leads using scores from 0-100, or stages from cold to hot.
Put simply, once the prospect reaches a certain score threshold, they are passed from marketing over to sales.
Here’s the limitation with most lead qualification and scoring software: they are based on subjective, human-generated assumptions.
- Lead Scoring: If a visitor goes to the pricing page, give that visitor +20 to their lead score.
- In-market Intent: If a visitor has searched for a relevant keyword to your products/services in Google, give them +10 to their lead score.
- Account-based Fit: If ID reveal shows a visitor is from a company that fits your Ideal Customer Profile, give them +50 to their lead score.
Each of these methods rely on human-generated scores. Those scores might be based on historical data or industry benchmarks, but they are still a best guess using cause and effect thinking.
This thinking is then turned into simple algorithms (which are step-by-step instructions for computer software to follow). “If the user does X, then Y will likely result.”
The reality is that visitors are not so one-dimensional. They are complex. A given visitor could have thousands of possible journeys resulting in an outcome.
Looking at just a handful of data points within the hundreds or thousands of possible permutations can’t be counted up together for an accurate representation of lead quality.
So, who or what can?
The Solution: Machine Learning for Buyer Intent
The only way to process hundreds or thousands of nuanced datapoints including historical and real-time data, then determine an accurate and repeatable outcome that is not subject to human bias, is through machine learning. To learn a little more I recommend you read this article which explains why machine learning trumps human-based algorithms for marketing software. I believe this application of ML is a hand in glove fit to complement your MarTech infrastructure
If we harness this breakthrough technology, we can settle the debate over quality by removing human bias.
However, what does that look like?
We need to stop pinning lead quality on only WHO is on your website, and start to include WHAT they are doing in REAL-TIME and WHY those actions are signaling intent to buy or convert when compared to millions of previously processed visitors in the model.
Simplistic measures such as “the visitor was on your pricing page” are not enough In fact, on average, 64% of your most promising web visitors do not go to a pricing page at all.
Which leads us to the exact solution for measuring the what, the why, and the real-time - which is buyer intent. Stay tuned next week for the next chapter in the value of Intelligent Intent Discovery.