Behavioral audits for better outcomes

Great products are built on effectively diagnosing behavioral challenges. We know that some of the most successful innovations are the ones that help solve users’ problems and get them from point A to point B. This journey might be from where they are to where they’d like to be or who they are to who they’d like to become.

When we think about the different products that we use often in our daily lives—the ones we love—what are some common themes? They’re easy, timely, relevant, and actionable. To the extent that we can continue to deliver these kinds of experiences for users, users will continue to hire us. To the extent that we fail to do so, well, they’ll go elsewhere.

However, the good news is that we have some choice in the matter. We all happen to sit closest to the data and insights, which means we can come up with effective solutions that lift barriers and make it easy for users.

Getting to know Christelle

I’ve spent the better part of the last 15 years leveraging behavioral science and behavioral science methods to improve people’s health and wealth outcomes.

I currently lead behavioral research at Voya Financial—a health, wealth, and investment management company. We leverage product analytics, insights and behavioral science methods—think frameworks and experimentation—to optimize people’s decisions, particularly concerning things like savings and benefit elections.

I’m really excited to talk to you about how we can leverage some of those types of methods when we’re looking at our qualitative and quantitative data to do things like better understand customer behaviors and the obstacles that get in people’s way. This enables us to lift those barriers and make things easy, timely, and relevant.

It’s really about thinking about the data we already have and how to apply frameworks to it.

Behavioral audits (the diagnostic)

I want us to approach thinking about behavioral audits as a diagnostic. Human behavior doesn’t exist in a vacuum, but the value of behavioral audits is that they really allow us to start contextualizing it.

We do that by pulling from across different data sources and different data streams so that if and when we start to observe that, for example, our product analytics are showing that people are getting stuck or dropping off, we can have a little bit more confidence around what’s happening and why.

So, I’m going to talk about a process that we can all start to use: leveraging behavioral science.

Your diagnostic process will have a set of inputs and outputs.

• Inputs: Product analytics, voice of the customer, market research, customer support, communications, secondary research

• Outputs: Barriers, facilitators

Depending on where you sit, you may be used to leveraging a few of these inputs. Or maybe you’re actually leveraging a lot. We’re going to talk about how we start to triangulate across a lot more of these types of data inputs and sources.

Let’s say you’re used to trying to understand people’s different types of behavioral interactions through your product analytics. As part of this process, for example, you’ll also want to marry that with the user feedback you’re getting through your voice of customer research, the behavioral sentiments and attitudes you’re getting through your market research, or the insights on when customers need help from interactions with customer support. Maybe you also want to understand how people are engaging with marketing content and outreach.

Secondary research is another huge source of wealth that we often overlook. Academics are doing really valuable research—I used to be one of them! Do you have pre-existing research available that you can leverage? That will save you costs and time.

In terms of your outputs, you’re trying to understand where people are getting stuck. What makes it hard for customers to get from A to B? What makes it hard for them to make decisions? Where are the moments of uncertainty? Where are the barriers?

When it comes to starting to make, or think, about solutions, what facilitates your customers’ ability to take action and make decisions? Sometimes this is about looking at other segments beyond your core users—maybe your power users, for example—as it could show you what to augment to make it easier for someone else to take action.

Introducing the COM-B model

So now, we want to start thinking about the barriers and facilitators in our output—the known drivers of behavior. This is where we start to pull from behavioral theory and where frameworks can be helpful.

Capability

Imagine your product analytics indicate that people are starting to get stuck at a particular point, or there’s a drop-off moment. That means there’s an action you want to drive. To what extent are people capable of actually doing this action? In other words, to what extent do they have the information, the necessary skills, or the education they need to actually do this?

Opportunity

To what extent do people have the opportunity to do this action? Are there opportunities provided through the digital or non-digital environment for engaging in the action? How can the UI/UX be changed to facilitate the action? What triggers can be used to prompt the action?

Motivation

And then, finally, to what extent are people even motivated to do the action you’re trying to drive? Motivation is huge, and it’s very multifaceted. There are many explicit aspects of motivation. There are the beliefs and intentions around motivation and the more automatic and implicit aspects of motivation—emotional processes or biases that may be at play.

So, as you start to pull across your different data sources that may be coinciding with where people are getting stuck or the drop off moments, to what extent are you finding that you maybe have some data or evidence that can speak to any of these drivers of behavior—capability, opportunity, and motivation?

The COM-B model

The COM-B model isn’t something that I’ve come up with. It’s actually an evidence-based behavioral framework. You can start to layer this onto your approach to looking at your behavioral data.

As the acronym suggests, the COM-B model posits that behavior—the B in this case—is a function of three core factors, or drivers: capability, opportunity, and motivation.

The COM-B model helps identify gaps in an experience and what needs to change for an experience to be even more effective at driving a behavior. It can help you identify what’s missing in your experience and then what you can introduce to help drive a particular action where you notice that people are stuck.

What’s also nice about leveraging this type of framework is that you can start to solution with more confidence. I agree with what Eniola Abioye said in her session, “From the Anatomy of an Insight to The Business of Research.” As researchers, we should feel confident and empowered to make strong recommendations.

Let’s look at an example. Back to the work we do at Voya, we are very hyper-focused on trying to get people from “zero” to feeling financially independent. I’m referring to someone who has never thought about saving or has never really started to save starting to engage in the behavior and adding it to their behavioral repertoire.

In the last 45 years or so, at least within the US, there has been more of a shift away from retirement plans and retirement savings being more managed and controlled by employers. Pension plans, particularly in the private sector, are moving toward being much more controlled by individuals—think defined contribution plans. So the responsibility—the onus—of thinking about setting themselves up for retirement now falls more on the individual.

And so, the moment we introduce the individual into the equation, we have to think about things like obstacles and barriers that can come in the way of decision-making. That’s a lot of what our team has to focus on—how do we get someone from A to B, recognizing that there are different types of barriers and obstacles that come up for people when they have to engage in a new behavior.

This isn’t just the case for savings. Have you ever tried to start an exercise regimen? Meditation? As humans, we are present-biased, meaning it’s really hard for us to forgo immediate rewards. It’s also really hard for us to envision the future. We have a really hard time with uncertainty. A lot of these things tend to lead to inaction.

So, as we’re building these products, taking into account these natural human tendencies and predispositions is important if we want to enable people to take the next step or lift these barriers to make things easy.

For people who do and are trying to start saving, we’ll employ a behavioral audit process. So, what do you think we learn? What do you think we find out?

Well, we learn a lot of things that you would probably learn too if you started to adopt this process with your products! People get stuck, and they get stuck in a lot of different ways. People get stuck when they have to start something and when it comes to contributing to their future. When we do an even deeper dive, pulling across these different inputs and applying that framework, we see there are a lot of different obstacles that relate to those drivers of behavior—capability, opportunity, and motivation.

Thinking about solutions that lift barriers and facilitate things for people

This puts us in a position to think about solutions that facilitate things for people.

So, how do we make that transition from understanding the obstacles that people face to coming up with solutions that we can have confidence in that will lift those barriers to enable people to engage?

Going back to the COM-B model, another really powerful thing about it is that it has “intervention functions.” When you want to introduce something, like a feature, nudge, or communication, and you consider the degree to which you leverage some of these at a functional level, you can have confidence that they’ll be relevant in terms of helping to address barriers.

So, back to our example from Voya with people who are trying to start saving.

We know that people get stuck when it comes to getting started and making a contribution to their future. We noticed that a lot of the barriers were opportunity-related, so we started to explore solutions that involved restructuring the environment.

In terms of science methodologies, we leverage a lot of experimentation. Think commercial randomized control trials. These can give us confidence as we are testing these different hypotheses.

Choice architecture

In terms of some of the pain points, we saw that some people said they couldn’t understand the information when trying to go through the enrollment process. We even saw that they were overlooking important plan information that they would actually need to make the right, or optimal, decisions.

Through testing, we found that just moving information to more closely align it with where customers have to make decisions promoted engagement and savings behavior, as did simplifying the language.

Color choice was also important, particularly leveraging colors that have strong behavioral associations.

Defaults

We also know that defaults, or preset courses of actions, can be really strong and effective interventions in situations where there’s inertia, or where people face ambiguity or uncertainty. So, when people are unsure about how much to contribute or not wanting to contribute to their pension, this can be pretty effective as a way to guide or nudge people to do so.

Through a randomized control trial, we varied the contribution amounts to see how much we could increase contributions. But we didn’t want it to backfire by being so high that someone actually opted into inaction. We found that we could probably get as high as 10% without it backfiring, which helped increase savings.

Framing effects

As humans, our choices and behaviors are influenced by the way things are presented or framed. When it comes to contributions, just slightly tweaking how percentage by paycheck is presented to people can have a huge impact in terms of how people engage and how much they save.

In this particular case, presenting the percent per paycheck in pennies actually had a huge impact on people who struggle more with numbers and percentages.

So, not only did it help with increasing engagement, but also with decreasing the savings gap. So, from an equity perspective, that was a great intervention for us to introduce.

Attribute partitioning

In the context of rolling over accounts, it’s actually a pretty complex process. Some people back out because they realize there are lots of steps and components.

With attribute partitioning, you’re essentially “chunking” information by what’s conceptually related. Then, you sequence these chunks in a way that reduces complexity and makes them more digestible. You also combine things according to the relationship among attributes.

By applying attribute partitioning to this part of the process, we enabled people to engage more and continue with their goal: rolling over their accounts. It was a win-win for customers and the business as it allowed people to resume their relationship with us.

So, there are lots of different ways you can go from your behavioral diagnosis to thinking about solutions, testing them out, leveraging behavioral audits, and looking at the impact on outcomes.

Considerations when applying behavioral science

There are a couple of things to keep in mind.

1. Build with measurability in mind

How did we even get all that data? How did that process work? Well, unless you already have that infrastructure in place, initially, it’s a very cross-functional process. You have to figure out what the teams are and which cross-functional relationships make sense.

You and your team always have to build with measurability in mind so that you can measure downstream impacts.

2. Just because people are motivated doesn’t mean they’ll take action

This sounds a bit counterintuitive, but just because someone has all the information they need doesn’t mean they’ll take action. If you think about the COM-B model, there are different drivers of motivation. So when we’re thinking about these experiences or about our product experiences, we have to think more holistically and really consider the different drivers of action. To what extent are we helping facilitate a person’s ability to take action?

3. A product is only as successful as its ability to change and influence behavior

This is why it’s really important that we generally and holistically apply more behavioral science thinking to our product development processes, particularly the more strategic ones. We know that it’s the products that take this much more behavioral and accurate view of humans and human decision-making that tend to be more successful and resonant.

These are the products that deliver experiences that are easy, timely, and relevant.

Editor’s note: This article is a condensed overview of Christelle Ngnoumen, PhD, session at Insight Out 2024.