In short: the idea that storytelling is a "neurochemical lever" working through oxytocin (Paul Zak's thesis) doesn't hold up, because studies have failed to replicate it. But that doesn't mean telling stories is useless, far from it. The IPA archive, 996 real cases, shows that emotional campaigns return three to twelve times more than purely rational ones, and the System1 method can even measure the effect before an ad goes live. The difference is timing: stories build the brand over the long run, they don't make people buy on the spot. Knowing where the evidence ends and the myth begins is the first step to not wasting your budget.
The promise: oxytocin, empathy, persuasion
Since around 2010, storytelling has taken over marketing on the back of a seductive promise: stories supposedly trigger oxytocin, the "empathy hormone", and anyone who hears a story becomes more trusting, more cooperative and, above all, more likely to buy. The person who did most to spread this idea is Paul Zak, a neuroeconomist, first through popular articles (Harvard Business Review, 2014) and then in his book Trust Factor (2017), where he points to oxytocin as the chemical explanation for the power of stories.
Meanwhile, the great classics were repackaged for corporate marketing: Annette Simmons's The Story Factor (2001), Joseph Campbell's "hero's journey", Pixar's "22 rules of storytelling". Donald Miller turned all this into StoryBrand, a positioning method that casts the customer as "the hero of their own story". And he built a company on it.
What replication studies say: oxytocin is shakier than it looks
Zak's thesis has drawn heavy criticism in the scientific journals. Lane and colleagues (PNAS, 2016) reanalysed the original data and concluded that oxytocin's effect on trust and cooperation is far smaller, and far more dependent on context, than the popular version suggested. A meta-analysis by Nave, Camerer and McCullough (Perspectives on Psychological Science, 2015) went further: the published results are inflated by the fact that mostly "successful" experiments get published, and the well-designed studies show weak effects.
In other words, the chain "the story releases oxytocin, oxytocin builds trust, trust drives the sale" simply doesn't stand up scientifically. But be careful: this doesn't mean storytelling fails. It means it works for reasons other than the one we were sold.
What the IPA archive says: emotion beats reason
The IPA Effectiveness Databank is an archive of more than 1,500 advertising campaigns analysed with real rigour, controlling for brand size, sector, budget and duration. On this data, Les Binet and Peter Field built a series of studies that have become industry references: The Long and the Short of It (2013), Effectiveness in Context (2018), The Effectiveness Code (2019).
The finding is clear. Over a two- to three-year horizon, campaigns with a strong emotional charge produce business effects two to three times greater than rational ones. On average we're talking about 12% growth versus 4% (Binet and Field, 2013, across 996 cases); and in the more recent 2019 data, the gap is even wider for everyday consumer goods.
One thing needs to be clear, though: the IPA does not confirm the oxytocin theory at all. It proves something else, namely that creative work which genuinely moves people produces measurable sales results. The reason, if anything, is Kahneman's "fast thinking", the distinctive memory traces studied by Byron Sharp, and brand building over time. Not an instant chemical reaction.
System1 Star Rating: putting a number on emotion
System1 (Orlando Wood, Lemon, 2019) developed a test that measures the emotional reaction of a real audience to an ad, before it ever airs. The score runs from 1.0 to 5.9 and predicts how much that ad will grow sales over the long term. The method is simple: 150 people watch the ad and then answer questions about what they felt, how well they recognise the brand, and what comes to mind when they think of it.
Ads scoring 5 stars or more, like Aldi's Kevin the Carrot in the UK, the John Lewis Christmas films, or Apple's old "Mac vs PC", return on average three times the norm. Ads scoring 1 star, rational, with no character, just a list of technical features, produce a tiny lift or even a negative one.
The top-scoring ads keep coming back to the same things: a character, a small story you can follow even in thirty seconds, a positive mood, and a brand that's present but not in your face. It's storytelling, yes, but made concrete and measurable.
When storytelling works
1. To build the brand over time. A consistent campaign running for one, two or three years creates the distinctive cues that make a brand recognisable and keep it front of mind. Think of the long-running brand campaigns from Coca-Cola, Guinness or John Lewis, or in Italy Galbusera, Nutella and Esselunga.
2. For products people buy without much thought. Detergents, snacks, drinks: the decision takes a second, and the story leaves the emotional memory that brings the brand to mind at the moment of choice. This is exactly Sharp's mechanism: it pops into your head because you remember it.
3. When the products all look alike. Compact cars, banks, mobile operators: when the real differences are tiny, the story becomes the way to stand out. Volkswagen's "Lemon" (1959, and still taught today) turns a flaw into a story nobody forgets.
4. In a crisis, and when there's a real value to tell. After a hit to reputation or a major change, an authentic story, not yet another "purpose" stitched on for show, can rebuild trust. The textbook case is Patagonia's "Don't Buy This Jacket".
When storytelling does NOT work
1. Direct sales and response ads. In Google search, in social ads with a call to action, in abandoned-cart retargeting, a story is useless or counterproductive. What converts there is a clear message, a concrete benefit, the lever of scarcity, and credible numbers showing that others have already chosen.
2. Hands-on B2B. Faced with a technical document, a demo or a use case, the person making the decision in a company wants numbers, return calculations and feature comparisons, not a story. In B2B, the story matters earlier, to get known, and later, to get recommended. Not in the middle, when the substance is being judged.
3. Complex products with a lot at stake. Mortgages, pension funds, complicated financial products: here people want transparency, not emotion. A story risks coming across as a distraction, if not outright manipulation.
4. When the brand disappears. A beautiful story with no visible brand just earns views. This is what Romaniuk calls the "vampire effect": the story grabs attention, but nobody connects that emotion to the right brand. You can spot it from one clear signal: fewer than 80% of viewers remember whose ad it was.
The ingredients that switch on "fast thinking"
In the top-scoring ads, System1 keeps finding the same ingredients:
- A character, even a minimal one, even an object given a face. It creates empathy and sticks in the memory.
- A positive mood. Humour, joy, human warmth regularly beat a neutral tone.
- Recognisable music and sound. With its own sonic signature, brand recognition rises by 30-40%.
- A plot you can recognise: the hero's journey in miniature, the classic "set-up plus twist", the "everyday scene the brand resolves".
- The brand inside the story, not slapped on as a logo at the end. When it's woven into the story, as Romaniuk explains, it stays in mind far longer.
How to measure it: four indicators
1. Star Rating (System1): from 1.0 to 5.9, it tells you how much the brand can grow over the long term.
2. Spike Rating: how much the ad can sell right away, in the first seven days after airing.
3. Brand recognition (fluency): how many viewers correctly remember whose ad it is. Below 70-80% is a warning sign.
4. Effectiveness Code (IPA): a grid for judging an investment across six fronts: brand, market, advertising, audience response, financial return and longevity.
If you can't afford a System1 test before airing, there are still shortcuts: a do-it-yourself test on fifty or so people, a comparison against the public data in the Facebook Ad Library (clicks and completion above the sector average), or a paid brand lift study on Google.
Is it worth it for a small or mid-sized business?
A storytelling campaign for a small or mid-sized company carries a real upfront cost: €30,000 to €80,000 for the idea and production, plus €50,000 to €200,000 for media over 8 to 12 weeks. For it to pay off, it has to reach a sizeable share of the target audience, more than 60%, and stay consistent for twelve to eighteen months.
Below €100,000 a year, it almost always makes more sense to spend elsewhere: search visibility, content that genuinely sets you apart, distribution. Above €200,000 a year, on the other hand, storytelling becomes a sound investment for building a brand that lasts. And here, the data backs it up.
FAQ
Is storytelling "scientifically proven"?
That emotional creative drives sales results is well documented (IPA, System1, Cannes Effectiveness Index). The chemical mechanism, the oxytocin Paul Zak proposed, is not: scientifically it's fragile. Keeping the two apart stops you from selling storytelling as the answer to every problem.
Can I do storytelling on a small budget?
Yes, if you have a strong idea. Dollar Shave Club's first video cost $4,000 and got 25 million views; Patagonia's "Don't Buy This Jacket" is, at heart, a newspaper page. What you need is an insight and a recognisable tone, not a big production budget. The one thing a small budget won't buy you is reach: to get past 60% of the audience, you still have to pay for the media.
Does Donald Miller's StoryBrand work?
StoryBrand is a teaching method that simplifies Joseph Campbell. It's useful as a clarity tool for anyone who has never thought in terms of narrative. But it isn't "scientifically validated", whatever the book's marketing implies: it's a handy mental model, not an evidence-based protocol.
Does it work for B2B software (SaaS)?
Early on, yes, to get known: customer cases told as stories, the founder's story, short documentaries on customer results. In the middle, when the tech is being evaluated, no: there you need technical content, return calculations, demos. After the sale, yes again: word of mouth, community, events.
Does generative AI write stories that work?
For working out variations on an idea a person has already shaped, yes: copy, scripts, visuals. For finding the distinctive idea, no: it tends towards the average, towards statistically predictable stories. The best way to use it is this: the person supplies the insight, the System1 method supplies the structure, the AI multiplies the variants.
How do I tell whether my storytelling is working?
Three cheap signals. First: how many people search for your brand on Google over the next six to twelve months; if it grows, you're staying in mind. Second: how often people mention you unprompted online (with tools like Brand24 or Mention). Third: a regular survey of 200-300 people in your audience, to see how many know you with and without a prompt. For a rigorous measure, a paid brand lift study (Meta, Google, YouTube) or a System1 pre-test.
Sources and references
- Binet, L. & Field, P. — "The Long and the Short of It" (IPA, 2013) and "Effectiveness in Context" (2018)
- Wood, O. — "Lemon: How the Advertising Brain Turned Sour" (System1, 2019)
- System1 — "Test Your Ad" methodology and effectiveness benchmarks: system1group.com
- Lane, A. et al. — replication issues on oxytocin, PNAS 2016
- Nave, G., Camerer, C., McCullough, M. — "Does Oxytocin Increase Trust in Humans?" (Perspectives on Psychological Science, 2015)
- Zak, P. — "Trust Factor: The Science of Creating High-Performance Companies" (2017, AMACOM) — primary source, with the due caveats on replication
- IPA Effectiveness Awards Databank — case studies 1980-2024
- Cannes Lions Creative Effectiveness — annual reports
- Romaniuk, J. — "Building Distinctive Brand Assets" (2018), on brand recognition and the vampire effect
- Kahneman, D. — "Thinking, Fast and Slow" (2011), for the fast vs slow thinking distinction
