F'innews: Back Together Again
November 19th: SF community highlights, next city meetups, synthetic data caution, and a season for compassion (~4 minute read)
SF Happy Hour Recap & What’s Next in Our Community
Two weeks ago we hosted our first San Francisco research happy hour, and about 50 researchers from across the Bay Area joined us.
For many, especially those working remotely, it was a rare chance to be in a room full of people who truly understand the work we do and who are pretty dang fun to be around. Much of the evening was spent simply connecting on a personal level, and we also took time for a thoughtful discussion on ‘what community means in our field right now.’
Some of what we heard:
“We need this; we really need this”
“Love talking shop in an unpolished or fully open way”
“So many of us are working remotely and we miss this”
“You can feel the knowledge in the room right now”
There’s real hunger in our industry for community, connection, and a place to speak freely with trusted peers, so we want to make this a regular thing!
What’s Next
We’re already planning our next meet-ups in:
New York City (early December)
San Francisco (Late January)
Seattle (March)
If you’d like to know when these events are happening and possibly join one in the future, ensure you are subscribed to this newsletter or reach out to us via DM on our LinkedIn.
A Reality Check: Why Synthetic Data Isn’t Ready For Market Research (Yet)
Two recent articles on Greenbook paint an attractive picture of synthetic data as the future of market research by promising faster insights, lower costs, and all at unlimited scale! We at F’inn greatly appreciate innovation and use AI tools ourselves, so we have conducted several synthetic data validation tests, including many of our own synthetic data experiments.
And each time, we found that synthetic data doesn’t capture how real people make decisions.
We had hopes that synthetic data could help us scale marginalized, expensive, and hard to reach populations; but as we dug in, we realized that large language models (LLMs) don’t accurately simulate human decision-making. LLMs are pattern-matching machines trained on text, not sentient beings with lived experiences, complex emotions, and the beautiful irrationality that is our behavior. When an LLM generates a “consumer response,” it predicts what sounds plausible based on trillions of text examples. But it has never experienced the frustration of a broken product, the beauty of the perfect color for a device, the joy of discovering something unique, or the social pressures that can influence real purchasing decisions. It’s never actually doing real human reasoning, so how can we trust it to make business decisions?
Synthetic data doesn’t line up with human data, how useful is that?
The articles say that 52% of cases are using synthetic data as a “full replacement for human input”. That’s scary! What if you launched a product based on synthetic enthusiasm but real consumers end up rejecting, invested in a costly campaign that only resonates with algorithms but not people, or missed safety concerns because synthetic respondents don’t experience real-world product failures? You might even overlook emerging market opportunities because LLMs can’t capture what hasn’t already been widely documented. The Greenbook articles mention various statistical metrics for validating synthetic data, such as distributional similarity, predictive accuracy, and privacy metrics. But there’s a catch.
You can only validate synthetic data against real data you already have. So, for predictive insights about new products, emerging behaviors, or market disruptions there’s no benchmark for validation.
Click here to read Greg’s full piece!
Season of Gratitude & Compassion
As we enter the holiday season, it’s worth remembering that many people around us are feeling the strain. Across the country, rising costs in healthcare, housing, and everyday essentials continue to stretch household budgets. One in eight Americans rely on food assistance, and many families are finding that what used to cover the basics no longer does.
Our research at F’inn has shown that only one-third of Americans feel they’re getting ahead financially, while the rest are struggling to keep up or living paycheck to paycheck. Our mental health research also clearly shows that financial pressure & uncertainty is a leading source of stress.
As we move into this time of year centered on gratitude and connection, let’s also make space for empathy. Whether it’s a client, partner, teammate or friend, remember that what you’re seeing on the surface may not tell the full story. A little kindness and compassion can go a long way.
Places we’re supporting over the holiday season: