Flowing Bee co-founders Sara Khorasani and Dr Michael Sankey sat down with hosts Clint McIntyre and Andrew J. Nash on Big Bets & Bad Calls this week for a wide-ranging conversation on behavioural science, AI, and the assumptions marketing leaders still make about their customers.
Listen to the episode
Flowing Bee on Big Bets & Bad Calls, hosted by Clint McIntyre and Andrew J. Nash.
Listen on SpotifyEarly in the episode, Sankey posed a question he uses to open conversations in boardrooms: if you give people financial education, what percentage of their actual downstream financial behaviour does that explain?
Most people guess somewhere around 80%. Teach people the facts, the thinking goes, and they'll act on them.
The real figure, drawn from meta-analysis, is under 1%.
It's a confronting number for an industry built on the idea that customers weigh up information and make rational choices. As Sankey put it on the podcast, marketing has largely operated on the assumption that customers "absorb all the information" and "make well informed decisions that are in their best interests." The reality, he said, is messier: "we are far from rational," and a tangle of emotional, cognitive, and social factors sits behind almost every purchase, hesitation, or walk-away.
That gap between the 1% marketing can explain and the 99% still hiding in customer psychology is the entire premise behind Flowing Bee, the Melbourne-based behavioural science and AI company Khorasani and Dr Sankey founded together.
The sausage sizzle test
Asked to explain Flowing Bee "like I'm a five year old," Sankey reached for a simpler picture: a sausage sizzle stall that customers keep walking past. The instinct is to put up a bigger sign or shout louder. Flowing Bee's approach is to work out why people are choosing you, or quietly choosing not to, before spending a dollar on the fix.
For marketing leaders, that diagnostic-first sequencing is the point. Rather than launching a campaign and reverse-engineering why it did or didn't land, Flowing Bee builds the behavioural diagnosis upfront, then feeds it into the initiatives – the emails, campaigns, and organisation-wide activity designed to shift a specific customer behaviour.
Credentials that don't come from a prompt window
Both founders bring a background that's hard to replicate with a chatbot. Sankey spent close to 15 years running commercial behavioural science for Fortune 500 companies, starting out of a studio apartment in New York – "my desk was about two metres from my bed" – before the practice grew into an office on Madison Avenue. His client list ran from airlines to confectionery to fast food chains, all wrestling with the same question: how to shift human behaviour at scale.
Khorasani's path ran through Formula One. She joined the Williams F1 team in 2014 while studying mechanical engineering, working on simulation technology that, among other things, was later commercialised into transport units for premature babies – the same vibration-dampening engineering built for race cars, repurposed to move the most fragile patients in a hospital system.
The two met during COVID, and Khorasani convinced Sankey that his depth of behavioural understanding was too valuable, and too expensive, to stay locked inside quarter-million-dollar, multi-month engagements only Fortune 500 budgets could reach.
What this means for marketing budgets and timelines
For a marketing leader, that's the number that matters: the kind of behavioural rigour that used to require a quarter-million-dollar, multi-quarter engagement is now something a business can access in weeks, at a fraction of the cost. Sankey was candid on the podcast about starting out sceptical that large language models could touch a process built on 15 years of expert judgement. He changed his mind once he and the team began breaking the workflow into parts and testing what AI could actually systematise.
By his account, engagements that used to take three to six months now run in a couple of weeks, without lowering the quality bar for the enterprise clients Flowing Bee works with in financial services and healthcare.
Getting there wasn't just a case of layering AI over the old process. Khorasani flagged a harder problem underneath the hood: managing bias in how large language models interpret behavioural data, particularly when analysing survey and interview results in sequence, where an early answer can anchor the model's read on everything that follows. It's a version of a problem web analytics teams have lived with for years – Andrew Nash, one of the hosts, pointed to the well-known abandoned-cart myth in ecommerce, where a spike in dropped carts every Friday afternoon turned out to be teenagers browsing after school rather than a checkout flaw. Data can show you the pattern. It takes a behavioural read to know what's actually driving it.
Where Flowing Bee is headed
Flowing Bee is currently Melbourne-based, with more than half the team local and the rest distributed across Australia. North America is next on the list, with the founders saying they're in early discussions about how to break into that market.
For marketing leaders wrestling with data they can see but can't fully explain, the founders' pitch was direct: get in touch. As Sankey said on the podcast, "if there's any marketers out there that are listening or watching, we'd love to have a chat about what your behavioural challenges are."
Listen to the full episode:
