什么是神经营销？定义与指南

神经营销是研究大脑对营销刺激如何反应的学科。

Technology and Automation

Neuromarketing is the use of neural and physiological measures, such as EEG, fMRI, eye tracking and skin conductance, to study how people respond to marketing stimuli. Some writers reserve the term for brain imaging alone, so check which version a vendor means.

One finding should slow you down. In a recent study of 56 grocery products, adding fMRI data to ordinary market data cut the error in second-year store-level sales forecasts by 18%, while a customer survey cut it by only 7.6%, yet when the question changed to whether store managers would stock a product, the survey won with an 11.3% improvement against roughly 2% for brain data (Bocconi University summary of a Journal of Marketing Research paper, Germany). Same brain data. Same products. Different decision, different winner.

The useful question isn't whether neuromarketing works. It's which decision you're predicting, and whether a brain or body signal carries information about it.

Neuromarketing, Defined

Neuromarketing records responses people don't have to put into words. A viewer watches an ad or scrolls a page while the researcher measures scalp activity, blood flow in the brain, gaze or skin sweat. The marketing question comes first. The sensor comes second.

Neuromarketing vs. Consumer Neuroscience

The two terms overlap, and you'll see them used interchangeably. Still, a widely cited 2013 paper in BMC Neurology argues for a cleaner split: call the commercial application of neuroscientific methods "neuromarketing," and call the scientific study of the same questions "consumer neuroscience." Their reasoning: public discussion of "neuromarketing" has centered on ethical concerns, while the scientific contribution of consumer neuroscience has been largely overlooked.

For a marketer, the distinction matters when you read evidence. A peer-reviewed study reports its sample, method and limits. A vendor case study usually doesn't.

What Counts as a "Neuro" Measure

This entry uses the broad definition, which follows the framework in UNESCO's Recommendation on the Ethics of Neurotechnology, adopted in November 2025. The adopted Recommendation separates two kinds of data:

Neural data: measures of nervous system activity, either direct, such as averaged bioelectric signals in EEG, or indirect, such as blood flow in fMRI.
Non-neural data allowing cognitive state inferences: biometric signals that only indirectly inform about neural activity, including eye tracking, skin conductance, heart rate variability and facial emotion recognition.

UNESCO's point is that the second group isn't strictly neurotechnology, but it raises similar ethical issues once used to infer mental states. That's a good working rule for marketers too: if a tool claims to read attention, emotion or intent from your body, treat it as neuromarketing. Either way, ask any provider which signals they collect.

The Measurement Toolkit

Each signal answers a different question. Mixing them up is where most bad interpretations start.

Neural Signals: EEG and fMRI

EEG records electrical activity at the scalp with fine timing, which suits second-by-second ad testing. fMRI tracks blood flow, so it's slower and costlier, but it reaches deeper regions that EEG can't measure well.

A 2015 Journal of Marketing Research study compared six methods on 30-second TV ads: traditional self-reports, implicit measures, eye tracking, biometrics, EEG and fMRI. According to the abstract, fMRI measures added the most explanatory power for observed advertising elasticities beyond traditional measures, with activity in the ventral striatum as the strongest single predictor of market-level response. Notice the wording. fMRI added information on top of surveys. It didn't replace them.

Non-Neural Signals: Eye Tracking, Skin Conductance, Facial Coding

Eye tracking tells you where people looked and for how long. Skin conductance shows whether arousal rose, not whether the feeling was pleasant. Facial coding infers emotion from expressions, adding another layer of interpretation.

The limits of gaze data are covered in the entry on heat map measurement limits. The short version: looking at something isn't the same as liking or wanting it.

Why Metric Reliability and Sample Size Matter

A signal can't predict anything if it doesn't give the same answer twice. Researchers at Erasmus University tested six EEG metrics commonly used in ad research on 116 participants viewing 13 TV commercials, and the reliability varied sharply, from poor for alpha asymmetry, to good for alpha and beta power, to excellent for intersubject correlation (Journal of Advertising). Most metrics needed 30 to 40 participants to rate an ad reliably, and more to track moment-to-moment peaks.

That matters because alpha asymmetry is often sold as a "liking" or approach score. If its reliability is poor, a dashboard built on it is measuring noise with confidence.

What the Evidence Says About Prediction

Prediction is the commercial promise of neuromarketing. The evidence supports a narrower version of it.

Predicting One Person vs. Forecasting a Market

Predicting what the person in the scanner will choose is one task. Forecasting what thousands of others will do is another. A 2025 study in PNAS Nexus by Genevsky, Tong and Knutson builds on the idea that not every component of individual choice is equally useful for forecasting aggregate choice, and across two experiments it found that neural responses to the affective components of a decision gave more generalizable forecasts of market demand than participants' own choices did.

This finding covers specific components and tasks. It suggests some emotional responses travel well across people, not that any brain metric will forecast your next campaign.

When Brain Data Beats Surveys, and When It Doesn't

The Bocconi summary of the newer Journal of Marketing Research paper gives the clearest marketing example. The team worked with one of Germany's largest grocery retailers, which operated 6,934 stores at the time, and studied 56 food and beverage products using a survey of 1,451 customers plus fMRI data from 44 participants.

Consumer sales per store, second year: fMRI reduced forecast error by 18%; the survey reduced it by 7.6%.
Store manager adoption: the survey improved forecasts by 11.3%; fMRI added about 2%.
Innovative products: for the 35% most innovative products, fMRI-based models predicted sales significantly better than survey-based ones.

The consumer-sales figures are reductions in forecast error and the adoption figures are forecast improvements, all from the same German study, and we relied on the university's summary rather than the paper itself. They aren't contradictory: people in the scanner reacted as shoppers, while store managers weigh shelf space and turnover, which a shopper's brain response is unlikely to capture. The authors' point is that more sophisticated data aren't automatically better; value depends on whether the signal captures the decision you're forecasting.

Where Interpretation Breaks Down

Most neuromarketing mistakes aren't measurement errors. They're interpretation errors.

Reverse Inference

Reverse inference means reasoning backward from "this brain region lit up" to "this person felt X." Russell Poldrack's 2006 analysis in Trends in Cognitive Sciences showed that how useful this move is depends on selectivity, meaning how specifically a region activates for one mental process compared with how often it activates across all kinds of tasks, as described in a published commentary on the paper. A region that responds to many things tells you little when it responds to your ad, so "the reward center lit up" is a weak claim on its own.

Myths Still in Circulation: The 95% Figure and the "Reptile Brain"

You'll often see the line that 95% of purchase decisions happen in the subconscious. Treat it as an unmeasured claim, not a statistic. It is usually traced to a 2003 Harvard Business School interview about Gerald Zaltman's book. The page's summary frames it as 95% of purchase decision making, but the interview question itself refers to 95% of all cognition occurring in the subconscious mind. Those are very different claims, and neither is presented there as a measured market figure.

The "reptile brain" is the other favorite. The idea that newer brain layers were stacked on an older reptilian core, adding more complex functions, has long been discredited among neurobiologists, according to a 2020 paper in Current Directions in Psychological Science.

Vendor Claims vs. Peer-Reviewed Evidence

Some of the sharpest criticism comes from inside the industry. Thomas Zoëga Ramsøy, the founder and CEO of Neurons, a neuromarketing vendor, argued in a 2026 Frontiers in Human Neuroscience paper that low metric reliability, poor consistency across metrics and the wider replication crisis have left the field building predictive systems on measurements whose reliability remains uncertain. Read it as a critique from someone with a commercial stake in the solution, not a neutral verdict, though it lines up with the reliability data above.

Before you buy, find out which metrics are used, how reliable they are, how many participants are tested, and whether predictions were validated against real market outcomes.

Consent, Ethics, and Neural Data Rules

Neuromarketing collects data most people consider deeply personal. Rules are starting to form.

What UNESCO's Recommendation Asks of Neuromarketing

The adopted Recommendation names neuromarketing directly. It asks member states to require disclosure and prior, free and informed consent, with participants actively opting in, and to strictly regulate how the data is used, stored and reused. It also calls for rules that stop marketing techniques from relying on such data from children and adolescents.

Keep the legal status straight. The Recommendation is not legally binding; it asks member states to give it effect through "legislative or other measures" of their own. It's an international standard, not a law that applies to your campaign by itself.

Neural Data in U.S. Policy

In the United States, the MIND Act of 2025 (S. 2925) was introduced in the Senate in late September 2025. Its text directs the Federal Trade Commission to study the governance of neural data and defines neural data as information obtained by measuring the activity of a person's central or peripheral nervous system through neurotechnology. The bill was introduced in the Senate in September 2025; no passage is established by the bill text cited here. It's a study bill. It doesn't itself regulate neuromarketing directly.

When a Survey Is Enough, and When Brain Data Earns Its Cost

Neuromarketing is a research purchase, so judge it like one: does the extra signal pay for itself?

Start with cost. For a project covering around 50 products, external providers quoted roughly €10,000 to €22,000 for a representative survey and €29,000 to €35,000 for fMRI data, according to quotes reported on the Bocconi page (the study ran in Germany, and the page doesn't say where the providers are based). The study's economics suggest fMRI can generate substantial savings when the goal is forecasting sales, but once retailer adoption counts too, surveys can offer greater overall value.

A survey, or the behavioral data you already have, is usually enough when:

The decision is made by a business buyer, distributor or retailer, not by the end shopper.
The product resembles things people already know, so stated preferences rest on experience.
You can run a live A/B testing program and measure actual clicks or purchases within a few weeks.
You already have rich behavioral data on the same audience and offer.
The vendor's metric hasn't been shown to be reliable, or the sample is smaller than the 30 to 40 people most EEG metrics needed in the Erasmus study.

Brain or body data is more likely to earn its place when:

The product is genuinely new, so shoppers can't easily say how they'll react (in one German study, fMRI-based models predicted sales significantly better than survey-based models for the most innovative products).
The outcome you care about is consumer demand, not trade acceptance.
The study tests whether the signal adds information on top of a survey, not whether it replaces one.
You can check the prediction against real outcomes later, at the same journey stage you measured.

For everyday creative work, a lab study can't keep pace: teams shipping new variants every week for their content marketing programs need live testing. Set baselines first, for example against a digital marketing benchmark for Indonesia, then decide whether a neural signal would answer a question those numbers can't.

Indonesian evidence in this research is thin. One 2025 study presented at the IEEE Student Conference on Research and Development, a student conference, surveyed 67 Indonesian participants, whose occupations are not specified in the abstract, using PLS-SEM and found that awareness significantly affected the use of neuromarketing, while innovation and absorptive capacity did not; it's a single small sample, not a picture of the Indonesian market.

Neuromarketing FAQs

Is neuromarketing the same as consumer neuroscience?

Not quite. Some researchers use "neuromarketing" for commercial work and "consumer neuroscience" for academic research. The terms overlap, so check who ran a study and whether it was peer-reviewed.

Does eye tracking count as neuromarketing?

Under the broad definition used here, yes. UNESCO's Recommendation groups eye tracking with other biometric signals used to infer mental states. Some writers limit the term to brain imaging.

Can neuromarketing read a customer's mind?

No. It measures signals such as electrical activity, blood flow, gaze or skin conductance, then infers mental processes from them. Those inferences are only as strong as the link between signal and process.

Is it true that 95% of buying decisions are subconscious?

Treat it as an unmeasured claim. The figure traces to a 2003 Harvard Business School interview whose summary frames it as purchase decisions, while the question refers to all cognition. It isn't a measured market statistic.

Does brain data predict sales better than surveys?

Sometimes. In a German grocery study, fMRI improved store-level sales forecasts more than a survey, but the survey better predicted which products managers would stock. It depends on whose decision you're forecasting.

Is neuromarketing legal?

That depends on where you operate and what data you collect. UNESCO's Recommendation, which is not legally binding, calls for opt-in consent and strict rules on reusing the data. The U.S. MIND Act would only direct an FTC study and the cited bill text does not establish passage.

Related Terms

Heat map and gaze data
A/B testing basics
Customer journey stages
Web analytics measurement
Eye Tracking
Consumer Behavior
Behavioral Economics
Marketing Research