As associations continue investing time and resources into solving their member retention concerns, they may overlook one of their most easily accessible tools – member data. However, knowing what to look for and how can make all the difference.
There is often a misconception that data analysis requires massive amounts of information and technical training. The reality is, for many associations, this work can be done with your existing data and programs you likely already have.
We took some time to speak with data scientist and Tasio Co-Founder Thomas Altman to gain some insights into what a data-driven member retention strategy looks like and where to start on your first data project.
This interview has been edited for length and clarity.
Before I got my first real-world job, I went to grad school for applied statistics, applied analytics, machine learning and use cases for using machine learning in the wider for-profit world.
I got hired out of that job implementing AMS software. I never used a lot of my training, but I did see, as people were using AMS software, they were generating the same kinds of data the for-profit world would generate and use to solve things like churn.
None of that was really being done for the association space, and I wanted to start playing around with, ‘is there a reason for that? Is it an awareness problem? Or is there a reason that the data doesn't work?’
And I found it was an awareness problem. People didn't realize that these tools existed in the for-profit setting, and with some minor tweaks, they could be made to work for the association space.
The biggest thing I've noticed is when associations have a retention strategy. To them, that means a drip campaign towards the renewal period. So they'll have this 60-day, 30-day or maybe a couple of days out email campaign that's the same for everybody that says “hey, dues renewal period is here, pay your dues,” right?
One of the things I always highlight is that the retention strategy is happening too late. People who've already made up their minds are not going to change their minds. And really, you're just capturing people that are kind of lazy about going through renewals.
Instead, you need to bring that much closer to the initial membership process. So after somebody renews or if they joined for the first time, catch them because that's when the momentum is high.
Also, you want to get them engaged earlier in things that are not about retention. You don't want to talk about dues the first time you talk to them; you want to talk to them about the cool webinar that's coming up, the event they can get a lot of value out of, some networking opportunities, or volunteering for a committee — those are the types of things that then make the renewal process a lot easier.
Every association is unique, but I'd say there are two big pieces that apply to most, if not all. First, those first three years – they’re the risk period. If you get to your third, fourth and fifth year, you're pretty much on autopilot; you're gonna renew and that's across every association I've worked with.
Don't focus your retention strategies on people who've been members for ten years; focus on their first year. Focus on onboarding processes and re-onboarding processes in a second year. Make sure you're hand-holding people through those first three years. I would say 50 to 40% of people don't renew in the first year, and that's across the board. Even the most high-performing associations I work with, 50% don't renew.
The other thing is looking for changes in behavior, like something someone did one year but not the next. I took this acronym from the donor community, which is LYBUNT, last year, but unfortunately, not this.
What you want to do is look for some behavior that someone did the previous year, so let's say they attended an annual conference or they attended four webinars, and then this year, they either attended at a lower level or they didn't do it at all. Those are red flags. When someone disengages by not attending a conference, stops watching the webinars or maybe served on a committee and dropped, these are things that members are saying, “Hey, I've disengaged for some reason.” You would want to find those groups of people and treat them as at-risk.
One thing I found really effective is, before you do a campaign, to really talk to individuals. Let's take the member who attended last year's conference but not this year. Find everybody who meets that criteria, and then just at random, take three to five, and give them a call. Don't talk to them about why they didn't attend – you don't want to go that specific and transactional with it – but you want to talk to them about their experience.
Like, “Hey, this is Thomas Altman from Whatever Association. I really want to know if you’re getting the most out of your membership. Can I help you get more out of your membership? Where are you struggling with us? What's been good? What hasn’t worked?”
Just have that high-level qualitative conversation. And what you'll notice, even after like three or four of these, a theme will pop out. You'd say, okay, everybody's saying this year's conference sucked because I didn't do XYZ. So you address that and say, “Hey, we're working on this. Next year's conference is gonna be great, and we want to do something else for you in the meantime.”
The use of data here is really narrowing down to have those one-on-ones with the right people who give you relevant information that can then be scaled up to a larger campaign.
Yeah. I think a lot of people are surprised because I’m the ‘data guy,’ right? And people want it to be automated data. My whole concept is that data informs human action and human decision-making. (Data) makes it easier for you to be a human in a way that's relevant to the person on the other side of that call.
You actually don't want to automate this stuff. You want to filter and then talk one-on-one and then learn as a human being so the next time you run the data, you're actually smarter about what you're looking for.
I like to think of AI as not artificial intelligence but as augmented intelligence, and that's where we specialize. What we like to do is think of our AI as a research assistant. All we're doing is plugging into AMS data, doing a lot of that digging for you, finding the relevant behaviors that signal when someone's at risk and identifying who those people actually are. Then we use AI to feed all of that profile information into writing some targeted content.
What might take hundreds of hours, a computer can just sit there and crunch through and identify who these people are, what the risk profile is, and what is a likely good outreach method, and then present that back to you so that you can edit it and decide.
AI minimizes time to value. What would be like a year-long data project if you decided to look at every single body in your member database and write them a personal email, AI could do it in a day. And because you're minimizing time to value, you're also shortening the step to the next chunk of value.
I use Excel every day because to me, that's the easiest way to visualize what's going on. Excel is good for two reasons. One, people already know it – you don't have to download anything extra, and you can always learn more, but you probably already know enough. And two, it will, by its design, constrain the amount of data you bring in. You have to start at some smaller, reasonable level of data that won't overwhelm you.
The next thing to do would be to connect that to Microsoft Power BI or Tableau. They’re very good at connecting and allowing you to manipulate the data in a way that's really fast, allowing you to take that leap forward and do smarter visualizations or mix the data in a way that makes more sense to you.
I call them the two best friends of budding data scientists. First, you have something called pivot tables, and there's some good content on Sidecar about how to use those by my good friend Dave O'Connell. Understanding how to use pivot tables will help you slice and dice that data in a way that doesn't require any background in data science.
The second friend, this is the best friend of a pivot table, is something called V lookup. And that's a little bit more in-depth. But it's very simple to learn. Basically, what you can do is map the metrics that you would create off a pivot table back to an individual member and whether or not they've renewed.
So you can see, okay, this person had 13 transactions in the past year, and they renewed, this person had three and they didn't, and you can start; those trends will start to pop out once you map metric level data back to an individual's renewal process.
I would say it's a two-way street. A lot of times, by engaging with the churn in member data, you'll find out the characteristics of somebody who’s at risk but also find out the characteristics of somebody who’s highly engaged. If you start doing this type of analysis, you have a data-driven engagement score – like stuff that actually is relevant to whether or not someone renews.
Instead of sitting in a room, which was the common way of doing this with the membership person, the meeting person, the certification person, and everybody arguing over, ‘oh, no, my stuff's more important,’ you actually can show what matters. If someone got a certification, and the data tells us they’re going to renew, then let's prioritize that over if they downloaded a white paper.
You also start to learn about gaps in your member benefits. By doing some of this outreach, you'll start to identify what people want. They're clamoring for something we're not yet offering, and you can start creating better, more relevant member benefits because now you're asking the right question.
The easiest thing you can do is go look at years of membership and the retention rate for each. Out of 100 people who are in the first year, how many of those renewed, and people in their second year, take 100 of them and ask how many of them renewed? You create a simple graph, and it's going to look up and to the right a little bit, with the lowest point being the first year and the highest being around the fifth year.
Create that graph and then go to either your boss or your executive director and say, ‘Hey, this is what I learned in 30 minutes in Excel. Can I have some time to go and do some more metric analysis?’
Don't try to do a big project. Try to do a really simple graph in Excel that gives you some information that maybe your membership director or your executive director hasn't seen before. Get that quick win to start giving yourself permission to go a little bit deeper.