At today’s Pardot User Conference Elevate 2011, Scott Voigt did a great job talking about a number of topics, one of which was marketing measurement. One example that he presented, and I thought was especially valuable, was that of the waterfall model, also known as a cohort or vintage model. The idea is simple but has several parts:
- Time is a challenge for marketing data as it is hard to attribute marketing-contributed value in an exact manner
- One of marketing’s most important KPIs is the number of opportunities created from marketing-generated leads
- Marketing-generated leads often don’t turn into opportunities immediately so it’s important to look at time series analytics
- A waterfall model takes the number of leads generated in a specific month, then tracks the number of opportunities generated from those leads in that month as well as subsequent months on a monthly basis
Here’s an example waterfall/cohort model for tracking marketing-generated leads that turn into pipeline opportunities:
| Jan | Feb | Mar | Apr | |
| Leads | 50 | 52 | 49 | 55 |
| Opportunities | ||||
| Jan | 2 | |||
| Feb | 3 | 2 | ||
| Mar | 2 | 2 | 1 | |
| Apr | 1 | 3 | 3 | 3 |
In the table above, note how leads generated in one month turn into opportunities over the course of time. Without this type of analysis it’s difficult to understand marketing’s on going effectiveness.
Scott did a great job with his presentation and I enjoyed hearing his theories on marketing.
What else? What are your thoughts on the waterfall method for marketing measurement?
David Cummings on Startups 
Interesting concept, time-series modeling opportunities derived from marketing efforts/campaigns is notoriously difficult and I would argue near-impossible to quantify the exact impact (but through rigorous design of experiment we can achieve statements like “this effort lifted sales x% +/- a standard deviation”). Couple of thoughts:
1) If you have a significant sample population of historical leads and closed sales (that you can link to an original lead) you can calculate your average lead-to-close time. Often this “lag-time” is unique to each B2B business/model and is where most people give up measuring their marketing effort impacts, and in my opinion one of the first things to understand about your sales channel as you mature past your “first sales” cohort.
2) Once you have calculated lead-to-close time then this waterfall/cohort plotting becomes really insightful and helps isolate/parse areas of opportunity. Example, if your lead-to-close time is ~60 days, your close rate is ~10%, and the current lead pipeline spins off 50 leads a month, then in 4 months time you yield roughly 10 closed sales (0-0-5-5). Then you start a new marketing effort (and only one new marketing campaign) that helps lift the lead pipeline to 60 leads/month, but closed sales remain flat at 10 closed sales (should have risen to 12 if we held the 10% close rate), then we have reason to believe that there is a scaling issue(s) in the sales channel. This helps the company keep the successful marketing campaign, and helps focus on either adding sales FTE resources or increasing the productivity of existing sales resources.
3) One noted risk here is the risk of false-positives in the lead population. A new marketing effort always presents the risk of adding interested parties to the population of leads, but due to the nature of the campaign were never really opportunities. Always hard to parse out this risk, and give me reason to believe that we should iterate closely to previous marketing efforts rather than taking large leaps into the marketing frontier…
Appreciate the daily posts, short enough to read, long enough to be interesting. Hopefully this comment is not too wordy…
Thanks for the great insight and feedback! This is really helpful and corroborates the challenges marketers face measuring their efforts.