Why the brain trusts some brands and doubts others.

A brain chemical with a job to do

Start with an experiment. Paul Zak, a neuroeconomist who has spent two decades studying the biology of trust, ran a version of what researchers call the trust game. One person decides how much money to send to a stranger, the sum triples in transit, and the stranger decides how much to send back. Zak's lab drew blood before and after the exchange and isolated the trigger with a control condition. When the same sum arrived at random, with nobody choosing to risk it on them, the recipient's oxytocin barely moved. When the same sum arrived because a stranger had deliberately chosen to trust them with it, oxytocin rose. The brain was pricing the gesture the money carried, independent of its size.

A follow-up experiment moved the finding from correlation to cause. Participants received a dose of synthetic oxytocin through a nasal spray, or a placebo, before playing the same game as the party deciding how much to hand over to a stranger. The oxytocin group sent 17 percent more money than the placebo group, and they were twice as likely to hand over the maximum amount available to them. Nobody in the room had been told anything different about the study or the stranger on the other side of the screen. Their chemistry alone had made them more willing to take the risk.

That distinction matters for marketing because it separates two things brand conversations usually merge: how someone reports feeling, and what their brain does next. Oxytocin functions as a safe-to-approach signal, the nervous system's shorthand for "proceed, no threat here." It releases in response to a specific, concrete cue, something a customer can point to as evidence of trustworthiness, and once released it changes the behaviour that follows, not only the sentiment reported afterwards. A customer who has just received that signal reads the next paragraph more generously and forgives a slower reply than they otherwise would. None of that shows up if the only instrument in the room is a five-point scale asking whether someone trusts the brand. Marketing has historically measured the wrong half of that equation, tracking what respondents say about a brand while the body has already decided what to do next.

The brain was pricing the gesture the money carried, independent of its size.

The office evidence, and the bridge to brands

Zak's most cited application of this research concerns managers and employees inside a single organization. In "The Neuroscience of Trust," published in Harvard Business Review in January-February 2017, he reports measuring oxytocin responses across a range of workplace behaviours to find which ones reliably raised it. Among the behaviours he documents: recognizing a person's contribution specifically, rather than with generic praise; giving people real autonomy over how they carried out their work; and sharing information openly instead of restricting it to those presumed to need it. Each gave the recipient's brain a concrete reason to treat the other party as safe. Zak's article does not rank these by the size of the oxytocin response, and this piece will not invent an order his data does not support.

That research measured managers and employees inside one organization, a different relationship from a customer meeting a brand for the first time, so the honest move is to name the gap plainly rather than borrow the finding as if it already applies. Zak's lab did not put a bank logo in front of a subject and measure oxytocin. But the mechanism underneath the finding is not specific to offices. It concerns what a nervous system does when it receives a specific, concrete signal instead of a vague one, and that nervous system does not swap itself out when the source changes from a manager to a bank's mobile app or an insurer's claims letter. A brand asking a customer for trust while offering nothing as concrete as what raised oxytocin among Zak's employees, specific reassurance or real visibility into the process, is asking for a chemical response without supplying what triggers it.

In practice this shows up as a simple test any marketing team can run on its own copy. A claim like "trusted by millions" asks the reader to take a population's word for it, which is closer to the vague reassurance the brain discounts than to the specific cue that raised oxytocin in Zak's workplace data. A claim that names the mechanism, a regulator or a named person who owns the outcome, gives the same brain something concrete to register instead.

When trust is the entire product

Most categories sell on a blend of factors, taste, price, convenience, status, with trust as one ingredient among several rather than the whole formula. A handful of categories work differently. Banking, insurance, healthcare and government services all ask a customer to hand over money, health information or personal data now, in exchange for a promise the institution will perform later, sometimes years later, sometimes only in an emergency the customer hopes never arrives. There is no product to sample in advance. The purchase is a bet on the promise, which makes trust the product rather than a feature of it.

Specificity against vague reassurance in copy is one concrete, testable cue. "We protect what matters most" produces a weaker response than a sentence naming the actual mechanism: the number of days a claim takes to settle, or the specific regulation a bank complies with. The presence of real people is another. A named claims adjuster, with a face and a direct line, reads differently to the nervous system than a stock photograph of an unrelated model wearing a call-centre headset, and that gap is measurable rather than a matter of taste. Friction runs in the opposite direction from both: every extra document, every unexplained delay, every step in an onboarding flow that does not say why it exists, drains the trust balance a fraction at a time. A slow, opaque claims process does more than annoy a customer. It erodes the safe-to-approach signal the brand needs, one form field at a time.

Vague reassurance asks the brain to trust the tone. Specific language gives it something to check.

A bank that publishes its complaints-resolution average in days, rather than describing itself as responsive, is making exactly this trade: a smaller, checkable claim in place of a larger, unverifiable one, and the smaller claim is the one that moves the chemistry. Brand systems built for that kind of scrutiny need consistency across every surface, including the unglamorous ones a hero campaign never touches. The discipline behind that is closer to what we set out in emotion at scale than to a single trust-themed advert.

How to measure a trust reaction

Self-report is a weak instrument here, because the emotion under investigation is exactly the one people are trained to perform politely. Ask someone directly whether they trust a bank, and the answer is shaped by social convention, by what a reasonable customer is supposed to say, by the fact that the interview itself is a mildly formal, mildly watched situation. None of that reveals what the nervous system did while the advert or the web page ran.

Two measures get closer to the truth. Biometric arousal, tracked through heart rate and skin conductance, shows whether something in the stimulus is registering at all, but arousal alone cannot separate calm confidence from a masked, uneasy reaction that produces a nearly identical trace. Facial coding adds a second signal, built to catch the brief, involuntary movement around the eyes and mouth that a managed expression usually cannot fully suppress, and pairing the two improves on either measure alone. Neither reading is a certainty by itself, and a well-practised or naturally low-expressive participant can still produce a trace that looks like ease when it is not, which is a limit our own buyer's guide to these methods is upfront about. The safer use of this pairing checks the result against a behavioural outcome, such as how long a visitor lingers on a page or whether they complete the form, rather than treating the biometric read as a verdict on its own.

Implicit association testing supplies the other half. It measures how quickly a person pairs a brand with a given concept under time pressure, and speed is the tell. Fast pairing with "reliable" indicates a stored association built over real exposure. A brand that only pairs quickly with "big" or "familiar" has bought recognition without buying trust, and those are different assets that do not predict the same behaviour. A brand can be famous and still be doubted at the point of decision.

None of this is exotic instrumentation reserved for pharmaceutical trials. It is the same measurement stack Orbital runs for creative and packaging work, pointed at a narrower question: does this specific claim, in this specific market, produce the chemistry of confidence or the chemistry of doubt. For a bank or an insurer, that question is worth answering before the campaign runs, not after the complaints arrive.

The scale trap in Caribbean markets

Foreign-owned and newly entered brands in Caribbean markets tend to lean on one trust cue harder than any other: scale. A global footprint and a longer history built elsewhere are routinely offered as shorthand for reliability. The instinct is not unreasonable. Size correlates with permanence in a lot of buyers' minds. But scale is a generic signal, closer to "familiar" than to "reliable" in the implicit test described above, and it tends to under-perform against specificity that is locally grounded: a named local underwriter, or language and imagery that read as belonging to the market rather than arriving in it from elsewhere. We have measured this pattern directly in category work across the region. Locally grounded cues often beat imported ones on the trust dimension specifically, even in cases where the imported brand wins comfortably on raw awareness.

The instinct to lean on scale is reasonable as far as it goes. It answers whether a company will still exist next year, a real concern for anyone signing a decades-long policy. The harder question, whether that company will treat this particular customer well, is the one a claims process or a loan application puts to the test, and scale alone rarely answers it. A regional insurer that can name its local claims handler by branch is answering a more specific version of that second question than a multinational's head office usually can, and specificity is what the nervous system responds to.

Where this stops working

Here is the limit, stated plainly. Trust signals are context and category specific, and nothing in this research licenses a single universal trigger that transfers unchanged from one category to another. Formality and credentials raise trust in banking because they answer the specific fear a banking customer carries into the relationship: that the institution might be careless or unaccountable with money that never belonged to it. Import that same formality into a lifestyle or hospitality brand and it can read as cold and distant, suppressing the warmth those categories are selling. A hotel that greets guests in a claims-adjuster tone borrows banking's stiffness and none of its purpose. Government services carry a different version of the same trap. A citizen renewing a licence online is not looking for warmth, and excess friendliness in that interface can read as evasive, as if the process is being smoothed over rather than explained plainly.

The honest position is a method: identify the specific fear or doubt a category's customers carry into the decision, then test which cues address that fear for this brand and this market. The same oxytocin system responds to different triggers depending on what it has learned to watch for.

This is an argument for treating trust as a mechanism with inputs and outputs, testable before a campaign spends a cent, rather than inferred afterwards from a satisfaction score. If a launch depends on being believed as well as noticed, that is worth designing a study around before the media plan is booked. Get in touch and we will scope what a trust study looks like for the specific claim you are asking customers to believe.

Frequently asked questions

Is trust really measurable, or is it just a marketing buzzword?

Yes. Trust changes brain chemistry and subsequent behaviour, not just what a person reports in an interview. Research from Paul Zak's lab found that a signal of trust, more than the transfer of money itself, causes the brain to synthesize oxytocin, and giving people synthetic oxytocin made them trust strangers with 17 percent more money and doubled the odds they sent the maximum amount available. That is a tracked, repeatable behavioural shift.

What is oxytocin's role in trusting a brand or a person?

Oxytocin acts as a safe-to-approach signal. The brain releases it in response to a cue that reads as trustworthy, and the release itself changes what a person does next: how much they risk and how readily they cooperate. Zak's original research measured this in person-to-person exchange, but the same nervous system evaluates a brand's cues using the identical chemistry.

Which industries benefit most from measuring trust directly?

Categories where trust is the entire sale, rather than one factor among several, benefit most: banking, insurance, healthcare and government services. A customer cannot sample a pension product or a claims process in advance, so reliability has to be established through concrete language and through the friction, or the lack of it, in every interaction, which makes those cues worth testing directly.

How do you actually test whether an ad or website builds trust?

Pair biometric arousal data with facial coding to improve on either signal alone, since arousal cannot separate calm confidence from masked unease and a well-managed expression can limit what facial coding catches too. Confirm the read against a behavioural outcome such as time on page or form completion, then run implicit association testing to check whether the brand is unconsciously paired with "reliable" or only with "big" or "familiar", which are different assets that do not predict the same behaviour.

Are trust signals the same across every industry and market?

No. A cue that builds trust in banking, such as formality or credentials, can read as cold or distant in a lifestyle category, and cues imported directly from a foreign market often underperform locally grounded ones in a Caribbean market, even when the imported brand is larger and better known. Trust signals need testing in the specific category and market where they will be used, not assumed from what worked somewhere else.