Introduction and Step 1: Brainstorm metrics
The task below will be automatically checked off once you complete watching the video.
“If you need to use statistics to understand your experiment, you ought to have done a better experiment.” — Earnest Rutherford, Nobel Prize Winning Chemist
You don’t need an advanced set of statistical methods to have an actionable strategy. The data you already have plus your understanding of your organization and how your members work is likely all you need to act in preventing a member’s exit.
Click each tab to learn more.
5 steps to predictive analytic success
Having a simple framework to follow has a stronger effect on business outcomes than complex business models because, as rules and models become more complicated, people are less likely to use them. Simple becomes actionable, so focus on using your intuition and impulses to create a much more effective strategy.
Click each tab to learn more before diving in deeper to each step below.
Step 1: Brainstorm predictive metrics
Using your knowledge and intuition of how your organization works and how members interact with your organization, compile five to 10 specific metrics that you believe will predict whether or not a member will leave. They need to be some version of a summary of the time before that person is going to leave; for example, the total number of logins to a website, the total amount spent on non-dues purchases, the number of continuing education units (CEUs) they’ve attained in a quarter, the average number of comments they write in a forum per month, etc.
Nominate people from different parts of your organization to help brainstorm around what they think can predict whether somebody might leave the association. Add these five to 10 data points to your spreadsheet.
Recency, Frequency, Monetary (RFM)
Think of metrics in RFM terms. For instance, days since last login, total number of logins over three months or total amount spent in the store. Metrics should be a summary of behavior over time including the count, the total and the average.
Your predictive metrics starter pack
Here are some potential metrics to use as you begin to identify the five to 10 that your organization will use to predict if a member is going to leave.
- Tenure (years/months of continuous membership)
- Days since last purchase
- # of non-dues purchases last 3, 6, 12 months
- $ of non-dues purchases last 3, 6, 12 months
- Change in # of purchases last month compared to this month
- # of logins/page views of website last month
- Days since profile update
- Credit card on file is expired
- # of cancelled/returned purchases last 3, 6, 12 months
- Days since last contact with member
- CEU/CPE awarded last 3, 6, 12 months
- Ratio of # unused seats to membership cost
- # of webinars or events attended last 3, 6, 12 months
- Ratio of usage / membership price
A note about tenure
When considering tenure as a metric, keep in mind that it will usually display as a u-shaped pattern. The newest members with the less time with your association are more likely to leave and appear at the top of the curve. Members with an average tenure of 5 to 6 years are less likely to leave and appear toward the bottom of the curve. Members closer to 10/15/20 years have a higher likelihood of churning, and that increases over time; they display at the top of the u-shape, too.
Download and save the Metric Brainstorming Worksheet to document your ideas.
[accessally_course_navigation prev_button='Previous' next_button='Next']
[progressally_note note_id='1' allow_attachment='no']
Share your progress!
[progressally_social_share type='facebook' share_id='1' post_id='42'][/progressally_social_share][progressally_social_share type='twitter' share_id='1' post_id='42'][/progressally_social_share][progressally_social_share type='email' share_id='1' post_id='42'][/progressally_social_share]
[progressally_progress_bar post_id="6437" width="100%" height="25"]