Archive for November, 2016

Photo Credit: https://images.unsplash.com/photo-1468309218808-b673d545d957?ixlib=rb-0.3.5&q=80&fm=jpg&crop=entropy&cs=tinysrgb&s=7f613564f8e45eb116209f29ed92b808

This is Edition 9 of Overdraft. Signup here if you feel like it.

Edition 9: Demand Generation Funnel

Hey friends and family,

I’ve read a series of articles on Medium about lead scoring, Marketing Qualified Leads (MQL) & Product Qualified Lead (PQL).

My head is buzzing so I will try to keep this coherent.

First of all, this article talks about building a predictive lead scoring model. It’s quite the buzz these days in B2B Marketing circles, with different vendors with hefty price tags and ominous black boxes selling their wares. However let’s break it down; B2B technology vendors are notorious for vague copy on their website, without actually explaining what or how they do it. I guess that helps them fuel the ‘demo request’ forms.

Traditional lead scoring is based on 2 dimensions:
– Straight up things like job title, company size (a proxy for revenue which means ‘disposable money to spend on technologies’) and mostly country and other such things.
– Behaviour, which means if they’ve actually been to your website/blog a couple of times and read or downloaded some eBooks and such.

Note: Remember to have negative scoring for things like email bounced, blank title, unsubscribed from marketing emails and such. Otherwise you’re lead score will, in theory, trend towards infinity.

Where lead scoring falls short? A couple of things, the world isn’t perfect and data isn’t clean. Someone might self identify as a VP of Marketing but in reality they might be a Marketing Manager. It’s all user generated anyway. Secondly if you use a third party data service to augment your lead records, you have a single point of failure. If the third party returns null or wrong data, your whole model is thrown off. 

So what is predictive lead scoring? It’s the idea (data science anyone?) that you look at a sample size of all the customers you got and look for similarities. If you have a large enough sample size and enough data points, there might be some commonalities like certain industries, behaviours (attended your webinar twice around ‘Lead Management for High Performing Marketing Teams’), spent 3 minutes look around your pricing page, had job titles within a certain area or other overlaps. These themes give you data points on what your ideal customer profile looks like and instead of having a linear lead scoring model, you have a weighted one. After all, not all behaviours are created equal, some are more equal then others. The challenge of-course that the process from visit to customer isn’t linear or simple. But a large enough sample size offers insights.
Should you then drop a hefty sum on a predictive lead scoring tool? Nope – Hubspot already offers it in built into their tool and you can build one yourself using excel by using regression analysis. What is regression analysis is a project for me for next week.

The second article talked about how traditional excel sales/revenue growth models often are based on historical funnel conversion rates which leads to total lead/MQL target madness. Yes, I’ve been there. But historical conversion rates aren’t predictive of the future, they’re an opportunity of improvement. If you MQL targets or lead targets are 8000 a month, then there’s definitely something leaky in your funnel and it’s better to fix that. What is the bottle neck? is it the Sales Qualified stage? Is that because of head count? Or is there a broken process? For example your funnel looks like this:

Lead -> MQL = 90%
MQL -> SAL = 40%
SAL -> SQL = 8%
SQL -> Opp = 50%
Opp -> Customer = 60%

It’s clear that there’s a 8% conversion rate that is causing the MQL/Lead number to inflate beyond control. What is causing that? What can be improved? If only 8% of the leads are being qualified by sales, what is happening to the rest of the 92%? Are they thrown out and replaces by new MQLs? Why not recycle them? Or is it simply that the MQL’s are not true MQL’s and hence not replying or engaging with sales? Can the channels/messaging and positioning be shuffled to better explain what the product does for whom so those who do fill out a form actually know what they’re getting into? One too many B2B SaaS products have vague ambiguous copy on their website, but interesting blog/ebooks that someone will download with no intention of buying the product or rather no idea what the product does. 

Process is essential but hard to get right but essential to the demand generation funnel. Hard to get right because there’s two ‘departments’ on a constant head butt with each other: sales & marketing. Traditionally, marketing is responsible for Leads -> MQL and Sales is SQL -> Customer. But both have varying incentives in places. I am all for SDR/BDR teams being under the Demand Generation/Marketing teams, to help simplify the process changes and align goals. 

Last thing: Product Qualified leads. This is a concept on freemium products, where a user has a free trial or free version and hits a ‘wall’. The wall can be a premium feature leading to a landing page or hitting the max quota on their plan or something of the sort. However I find this only works in the case of freemium products, if your product has no entry point to a account and usage, this model would not work.

Here are all the links:
Stop the Lead Scoring Madness.
Lead Scoring Models [Slide Share]
Product Qualified Lead
Before You Widen The Top of the Funnel

Before I go, here’s a fantastic read by Erin from Fortune on Cruise Automation that sold to GM Motors for a a mind boggling amount of money and how GM and Cruise is keeping things alive.

It’s been a while. I hope you’ve all been well. 


This is Edition 9 of Overdraft. Signup here if you feel like it.

Read Full Post »

%d bloggers like this: