Guy White, Founder and CEO of Catalyx, on why asking consumers what they want is a waste of time, and what actually predicts whether a new product succeeds.
Between 50 and 75% of new consumer products fail within two years. That’s not a typo, and it’s not a niche industry problem. Guy White has spent 14 years trying to figure out why, and he’s landed on an uncomfortable answer: most companies are building products based on what people say, when they should be building based on what people actually do.
White is the founder and CEO of Catalyx, an innovation consultancy that’s worked with some of the biggest names in consumer packaged goods. He started his career as a financial analyst at Procter & Gamble before moving into brand management, and that analytical background shows up in everything he says about product development. He doesn’t talk about creativity and gut instinct. He talks about data, patterns, and process.
Why Can’t You Just Ask Consumers What They Want?
This was the most counterintuitive part of the conversation. You’d think the fastest way to build a better product is to ask your customers what they’d buy. White says that’s almost always the wrong move.
“Consumers are pretty terrible at telling you what the future product is that they want. You ask them a question, they’ll tell you of course, but it’s rarely right. But what they’re really good at telling you about is what they like and what they dislike. They’re amazingly good at reacting to something that you put in front of them.”
— Guy White, Founder and CEO, Catalyx
There’s real data behind this, not just intuition. White pointed to the standard market research question, “on a scale of one to five, how likely are you to buy this?”, and the research on how well it predicts actual behavior.
“The correlation between purchase intent and purchase is about 0.18, i.e., zero. When you show me a product on a piece of paper and it looks kind of interesting, I’m like, ‘yeah, maybe I’ll buy it.’ And then actually when you’re buying products, you won’t take a risk. You will buy the thing that you’ve always bought.”
— Guy White, Founder and CEO, Catalyx
That’s a genuinely useful number to have in your back pocket next time someone in a meeting says “the survey results looked great.” A 0.18 correlation is close enough to no correlation at all.
What’s the Actual Framework for Making Product Innovation Predictable?
White broke this down into three stages, which is the closest thing to a repeatable formula I’ve heard for something as fuzzy as “coming up with a good product idea.”
First, identify the platform, meaning the underlying consumer job to be done, before you think about your own product at all. White used the classic Clayton Christensen example: McDonald’s discovered a spike in milkshake sales during the morning commute, and the “job” the milkshake was actually being hired for wasn’t nutrition, it was keeping a bored commuter occupied for the drive to work.
“People don’t want a screwdriver or a drill. They want a hole.”
— Guy White, Founder and CEO, Catalyx
Second, build the “recipe”: the 10 to 15 concrete product attributes, name, claim, design, format, packaging, that actually deliver on that job. Third, execute and test it in an iterative loop of diverging into wild ideas and converging back based on real consumer reaction.
“The uncomfortable truth of the situation is that between 50 and 75% of products fail within two years, certainly in the consumer goods sector. I think that really there are three things that you need to do with data at the core that helps you craft predictably more successful products.”
— Guy White, Founder and CEO, Catalyx
How Do You Know If the Data You’re Working With Is Actually Reliable?
This question came from a viewer during the live show, and White’s answer should worry anyone buying third-party research data. The rise of cheap global consumer panels has created an incentive problem: pay someone fifty cents to a dollar per response, and you attract people optimizing for speed over honesty.
“There is huge fraud in our industry. The entire system has been gamified and botified, and you’ve found people that aren’t the people you’re meant to be talking to that have infiltrated that system because it’s a very quick way of churning out quick money.”
— Guy White, Founder and CEO, Catalyx
His practical advice: interrogate your data vendors on how they filter bots, work with large data sets rather than small ones so bad actors get diluted out, and treat anything suspiciously positive with skepticism. As he put it, “if it smells fishy, it’s probably a fish.”
What’s the Difference Between Data and Insight?
This is the section I’d bookmark if I were building out a research or analytics team. White draws a sharp line between the two, and it’s a distinction most companies blur.
“Data is increasingly easy and insight remains hard. Insight is the gap between the data, it’s the way that you look at the data or string the data together and make the leap and tell the story that the data alone is not getting you to.”
— Guy White, Founder and CEO, Catalyx
His example: research showing that mothers of young children feel guilty not because of what their kids eat today, but because of the eating habits and taste preferences they’re worried they’re failing to instill for adulthood. The data shows food choices. The insight is about identity and long-term guilt. That’s the gap White is talking about, and he says you’ll know you’ve found a real insight because “the hairs on the back of your head lift up” in the room.
On the role AI plays in getting there, White’s team uses it heavily for the tedious first pass, pattern recognition across large unstructured datasets, but draws a clear boundary around where it stops being useful.
“We talk about it like AI in the engine room, humans at the helm.”
— Guy White, Founder and CEO, Catalyx
His reasoning: AI is fundamentally backward-looking. It’s excellent at surfacing what a “job to be done” already looks like based on existing data, but genuinely novel tension points, the gaps that don’t exist yet in any dataset, still require a human strategist looking at real behavioral footage and making an intuitive leap.
The Bigger Picture
What stuck with me most is how universal this framework is. White’s clients are mostly consumer packaged goods companies selling on physical shelves, where a wrong product decision means wasted factory runs and blown listing fees. But the platform-recipe-execution structure, and the discipline of separating what customers say from what they actually do, applies just as directly to software, to marketing campaigns, to any decision where you’re tempted to skip straight to a survey instead of watching real behavior.
If you’re building something new and your validation process starts and ends with a survey, White’s data suggests you should be worried.
Connect with Guy White on LinkedIn or learn more about Catalyx.
At BDEX, we think about the data-versus-insight gap constantly, raw identity data only becomes valuable once it’s connected to real, verified consumer behavior. If you need that kind of foundation for smarter product and marketing decisions, visit bdex.com and click “Talk to an Expert.”
This article was adapted from an episode of Deconstructing Data, BDEX’s weekly podcast on data-driven marketing. Tune in live every Thursday at 4:15 PM Eastern on LinkedIn.
Watch the full episode: Predictable Innovation Through Smarter Consumer Data
About BDEX: For companies that need clean identity data to power their products, BDEX offers unmatched quality and execution. Visit bdex.com and click “Talk to an Expert” to get started.
