Closing the Say-Do Gap: How Behavioral Data Makes Insight Communities Smarter

Rick

Before I jump in too much, I want to talk about the ability to connect data and insights into a system of action. I remember building a slide deck and taking it around to a bunch of customers in 2016 to talk about this. In 2016, it was a little early for the concept.

Ten years later, I think this is the wave of the future: building a continuous customer intelligence engine, and connecting business systems to your insights to bring that to life.

Rick

The first key challenge researchers face is data integrity. We talk about the say-do gap all the time in research. It's challenging because in many cases we're asking respondents to recall behavior: to remember what they did, what they bought, how they acted, and report it back to us.

Well-meaning respondents end up providing incorrect data, not through any fault of their own. We have a real challenge collecting actual behavioral data unless we're witnessing it or can observe it somewhere else. This is well documented. Some of the research citations we're pulling in here are nearly 50 years old, because this is a long-standing issue. People are almost always going to misremember what they've done in the past, and that affects decision making. Brands often make incorrect or fuzzy decisions, not because of data quality or respondent quality, but because consumer recall isn't as accurate as it ought to be.

Fuel Cycle is primarily a community provider. We build insights communities, which let us measure respondents and consumers longitudinally. The traditional way of doing research is episodic: we go out and ask consumers to share behaviors from yesterday, a week ago, or a month ago, and those may not be accurately represented in our data. Think about the decisions people make about restaurants, about which deodorant to pick, about banking. There are going to be faulty memories, or motivations that aren't accurately represented in our data set.

The say-do gap is fairly well covered in our industry, so I won't belabor it. We have a challenge connecting behavior to intention, and it often results in faulty decision making. That's not the researcher's fault, and it's not our research process's fault. It's the system we have in many cases.

Rick

Another challenge for many research practitioners, and you can tell me if I'm wrong, is translating insights into action and bringing research to life.

You run discrete, episodic projects and bring back a beautiful deliverable: a PowerPoint deck, a Word document. That artifact ends up in a corporate SharePoint or Google Drive and goes there to die. You put a lot of resources and effort into it, and then it's on you to push it through the business and make sure the right decisions are acted on.

This is a well-documented issue. If you look at the last 10 years of conference presentations, setting the AI conversation aside, there's a common through line: storytelling, getting a seat at the table, and acting on insights to affect business decision making. That can be a real challenge.

We think that connecting real-world behavioral and attitudinal data with your insights data leads to a better understanding of your consumers. First, it's very high fidelity. Second, you get better activation, because it's connected to CRM systems and your data lake, so you can act on it. Third, you can connect ROI and measure the impact of your decisions on the business and on revenue.

Rick

I want to draw a line between traditional research and research done through a platform like a research community. In many cases, traditional research is discrete projects done in silos: a survey, a focus group, a set of user interviews. You recruit a group of respondents, and then you never see them again.

In a community, we have a longitudinal record of respondents and we engage with them over time. This is a privacy-centric, enterprise-grade way of connecting data. We can connect respondent behavior with your CRM systems and data lakes. We have APIs that connect to Salesforce, Snowflake, Microsoft Dynamics, or whatever your system of record is for respondents, and you can start incorporating that data into Fuel Cycle.

We can also push the data and insights collected in Fuel Cycle back out to the business: into the CRM, the data lake, the marketing automation platform. That's incredibly useful because we build a longitudinal record and can watch behaviors change over time. We close the say-do gap by tying actual purchase history to your research project, so you make better decisions. And you can take automated actions downstream in marketing, product improvement, and innovation.

Those are three core benefits. When you change how you collect data and use a Fuel Cycle community, you compound the data, you look at respondent behavior longitudinally, you improve decisions, and you can act on the data much more simply than if you're sharing a PowerPoint report around the business.

Rick

Within Fuel Cycle, we have a concept called progressive profiling. It lets us combine attitudinal, behavioral, and experiential data with your research data.

Let's take a silly example: respondents' favorite color. I have an eight-year-old whose favorite color changes anywhere from every two days to every year. Say you have favorite color data from three surveys: one you ran three months ago, one you ran yesterday, and one you ran two years ago. Call them surveys A, B, and C. For the same respondents, you have data in your CRM system, and for some others, data you collected offline in a spreadsheet somewhere. That's five competing data sources, and pulling them together into a persistent customer profile is what progressive profiling solves for.

With integrations to your CRM, to whichever survey platform you use, and to other sources as simple as a spreadsheet, we build a persistent profile that tracks respondent behavior and preferences over time. You decide which source to trust first. For example: trust the CRM first, then the spreadsheet, then surveys A, B, and C in order. If a respondent's data is missing from the CRM, we fall back to the next available source. When you run a future survey, or a respondent's preference changes in your CRM, the profile updates automatically and ties to all the research they do.

Now apply that to something more complex than color: what someone purchased in the past six months, their phone plan, how often they log into a mobile app. Those behaviors are really challenging to track. We can consolidate them into a record that follows the respondent. That improves decision making because we're closing the say-do gap, and we can push that record back into the business to accelerate decisions downstream of your research projects.

Rick

We're taking existing systems, combining them with research data, and pushing it back out to your operational systems. Fuel Cycle started sharing this idea about 10 years ago, and I can promise you there was zero uptake. It sounded conceptually great, but it sounded really challenging to do. Today, I'd estimate about 40% of Fuel Cycle customers are tying this data together, connected to anything from Salesforce to Qualtrics to Snowflake, and pushing it into their insights community and then out to solutions customized to the business.

Getting started is relatively easy. It comes down to business definitions: what you want to track and measure, and where you want your insights data to end up. Those are complex decisions. When I say it's easy, I mean it's technically easy. If you wanted to set up an integration this week, you could. The harder part is knowing what data you want to bring in and what data you want to push out, and then socializing the integrated data and pushing it back out to the business.

Rick

For many telecom providers, one of the biggest challenges is customer retention, because consumers can generally jump from contract to contract. The first 90 days of onboarding a new customer are especially hard. If they have a poor experience, they're likely to churn later. They're making plans to churn in those first 90 days, even if they're locked into an annual contract.

We pulled together an ecosystem for a telecom provider that incorporates data from their CRM system: longitudinal behavioral records like which plan respondents are on, which devices, how many people are on the plan, and location. Those are things many respondents struggle to remember and report accurately. At the same time, customers go through an onboarding experience that is measured by a separate system, such as Qualtrics or Medallia, where they provide NPS responses. Those NPS responses are pushed into Fuel Cycle and combined with the CRM data.

Because we have a longitudinal, non-transactional relationship with these customers, we can target them for follow-on research to redesign the onboarding experience. For example, we can see that people on plan A with an iPhone 16 Pro tend to have a poor onboarding experience, then bring them into a focus group or a set of user interviews to find out why they're promoters or detractors.

The brand can continuously refine its onboarding experience, because all the data flows are automated. It isn't a one-time project. It's a living, breathing system. They can find issues, then find those respondents and people like them again through their marketing efforts to promote retention. It runs continuously and supports a strong customer retention effort.

Rick

In retail, there are lots of loyalty programs, and we can integrate with them to see what people are purchasing, what they're ordering, and which channel they're ordering on, whether that's online or in store. We tie all of that data together to improve promotion timing and how new products are promoted.

We set up a program for a brand where they tie behavioral data from their loyalty systems and purchase behavior, from online or offline sources, to what they learn from respondents. That lets them identify low-frequency purchasers and repeat fans who come into the stores or shop online all the time, and dive deeper with those people about their actual behavior.

Rick

The third example is an organization we work with in AAA gaming. Gaming lives on hits and retention, so it faces challenges similar to the telecom provider: retention and gameplay are critical. If you think about the say-do gap in banking or telecom, it's even more challenging in gaming. Sometimes people pick up a game casually, and sometimes they sit down and play for hours. Do they remember when they ran into a bug? Do they remember who they were playing against? It's very complex.

We can take telemetry data, which is actual in-game behavioral data like sessions played, how often players engaged, how frequently they came back, and whether they encountered a bug. That data is piped into Fuel Cycle and used to conduct research on gameplay, from surveys to focus groups to user interviews. Respondents can also take part in what we call Delta Testing, where they submit ideas for gameplay or report bugs. This gaming company runs a beta testing program on Fuel Cycle. They recruit and identify players to expose to new versions of the game before they're released to the broader public. Those respondents, tied to their telemetry data, submit feedback to the brand through Fuel Cycle.

That feedback is organized and pushed into product management tools like Jira, a software development platform. Rather than pushing insights into a PowerPoint report or an Excel spreadsheet that goes out to stakeholders, they push the data to where their product managers, QA managers, and engineers actually live. The feedback gets acted on, instead of sitting in a static report that might be outdated by the time it reaches their screens.

Rick

Across the three examples, the common thread is moving insights to action and closing the say-do gap. We do that by incorporating behavioral and attitudinal data, whether it comes from a CRM where we're looking at purchase behavior, loyalty data, transactional NPS-type work, or reams of in-game telemetry. We can push that back into the business: using marketing automation to target lookalike respondents, and sending data straight to product managers and engineers so they can improve digital products at a faster cadence than they otherwise could.

We believe AI is going to have a huge impact on this. We already see the ability to look across broad swaths of data that we otherwise wouldn't have been able to. Think about historical analysis and crosstabs: human eyes start to bleed past about 30 banner points. But whose eyes don't bleed when you're looking at 200 banner points and cutting the data multiple ways? AI agents. They can detect changes, slice data across multiple sources, bring a more holistic picture back to the brand, and speed up the process.

It's not only about speed, because we're going to see a lot more of that. The game changer is the ability to incorporate multiple data sources, bring them together, and act on them downstream. We think that's what the combination of AI and a continuous intelligence system makes possible.

Rick

The call to action is to not think about research as a discrete project. Think about how it fits into a continuous customer lifecycle, a continuous customer intelligence engine, where we combine data from multiple sources and improve reporting overall.

To summarize: connecting different sources of data, behavioral and attitudinal alongside stated preference data, is the wave of the future.

Rick

It's a lot of industries. Anything that's high frequency and low touch, where there isn't a long-term relationship, tends to be one of the biggest challenges. Unfortunately, it's also one of the hardest to get at. CPG is one example, and loyalty systems have helped there. Overall, it's a pretty universal challenge.

Rick

Wherever you have the ability to match respondents through a single identifier. If you have an email in one system and the same email in another, you can connect them. From a technical integration perspective, CRM systems are well suited. We do this through API calls, those APIs are usually pretty well defined, and we have a lot of experience doing these integrations.

Rick

I think it's an education process: socialization, and telling them how you got the behavioral data. In many cases it's an upgrade to what they've seen in the past, so it tends to be well received.

Rick

Research is still research, and in many cases the processes look the same. What these teams tend to have is a seat at the table, so they influence organizational decision making more than they otherwise would.