Ask a customer what they bought last month and you’ll get an answer. It just won’t always be the right one.
It’s not that respondents are careless, but often, people don’t accurately remember their own behavior. Researchers have documented that for nearly 50 years. What surprises me is how little our research systems have done about it.
The usual response is better survey design. I think that treats the symptom. The say-do gap persists because the data that could resolve it, what customers actually did, lives in a different system from the research. Here’s how behavioral data integration closes that gap, what it looks like in practice, and where to start.
The say-do gap is a systems problem, not a respondent problem
Most research is episodic. You recruit a group of respondents, run a survey or a set of interviews, and then you never see those people again. Along the way, you ask them to recall what they did yesterday, last week, or last month. Memory fails, and brands end up making fuzzy decisions. The cause usually isn’t data quality or respondent quality. Consumer recall just isn’t as accurate as it needs to be.
Meanwhile, the behavior you’re asking about is often already recorded: purchase history in the CRM, orders in a loyalty system, sessions in product telemetry.
I first took this idea to customers in 2016, with a slide deck. It was early. It sounded conceptually great and practically hard, and almost nobody picked it up. Ten years later, the integrations are straightforward and the expectation has changed. Teams want research that connects to the rest of the business.
Better data still isn’t enough if the insight never leaves the deck
Closing the say-do gap fixes the input. It doesn’t fix what happens after fieldwork. Most research ends in a polished deliverable, a PowerPoint deck or a Word document, that lands in a corporate SharePoint or Google Drive and goes there to die. Then it’s on you to push it through the business and make sure the right decisions get made.
You can see this in a decade of conference programs. Set the AI conversation aside, and the same themes keep coming back: storytelling, earning a seat at the table, getting insights acted on. They’re all attempts to solve the same problem.
Connecting behavioral and attitudinal data to your research helps in three ways. The data is higher fidelity. Activation improves, because it’s tied to your CRM and data lake, so teams can act on it. And you can connect research to ROI, measuring what your decisions did for the business and for revenue.
A persistent profile replaces the one-off respondent
In a research community, you keep a longitudinal record of each member and engage with them over time. We can connect member behavior to Salesforce, Snowflake, Microsoft Dynamics, or whatever your system of record is, and push insights back out to the CRM, the data lake, or the marketing automation platform.
The mechanism that holds it together is progressive profiling. Take a deliberately silly example: favorite color. My eight-year-old changes hers somewhere between every two days and every year. Say you have that answer from three surveys run at different times, from your CRM, and from a spreadsheet someone kept offline. Progressive profiling lets you decide which source to trust first, falls back to the next one when a value is missing, and updates the profile automatically when a new survey comes in or the CRM changes.
Now swap color for what someone purchased in the past six months, their phone plan, or how often they open your app. In a Fuel Cycle community, those become one persistent record that follows the member, which closes the say-do gap and gives the business something it can act on.
Three examples show what changes when research connects to the business
A telecom provider worried about churn in the first 90 days of a customer’s life. We combined CRM data (plan, device, household size, location) with NPS scores from the provider’s experience-measurement platform. That showed which customer segments had a poor onboarding experience, and the team could bring those people into interviews to learn why. Because the data flows are automated, the team keeps refining onboarding and can find similar customers for retention marketing.
A gaming company pipes in-game telemetry, such as sessions played, return frequency, and bugs encountered, into its community. It runs a beta testing program there, so feedback comes from players whose behavior is already known. That feedback goes straight into Jira, where product managers, QA managers, and engineers work, instead of a PowerPoint that may be outdated by the time it’s read.
A retail brand did the same with loyalty and purchase data to improve promotion timing. The pattern is consistent. Insight goes where decision makers already work.
What this means for insights leaders
Start with one source. Technically, an integration can be running within a week. The harder work is deciding what you want to bring in, where insights should end up, and who needs to trust the result.
Match on a shared identifier, like an email address, and start with your CRM, where the APIs are well defined.
Expect an education process with sales and product teams. Most have never seen behavioral data come from the research function, so explain where it came from. In my experience it’s received as an upgrade over what they’ve seen before.
The payoff is influence. The research process itself looks much the same. What changes is that these teams earn a seat at the table, because their work ties to revenue and product decisions. It also prepares you for AI. A human analyst’s eyes start to glaze past about 30 banner points in a crosstab. An AI agent can cut across 200 and flag what changed. The more connected data it has, the more useful it becomes.
Surveys are still good at asking why. They’ve never been good at remembering what happened. Connecting behavioral data gives research the memory it was missing, and gives the business a way to act on what it learns.
If you’re deciding where to start, we’re happy to talk through which first integration makes sense for your team.
I recently walked through this topic in a webinar, including the telecom, retail, and gaming examples in more detail. Watch the recording.
