Nonverbal Communication Is Everywhere. The Data Isn't.
Nonverbal communication decides sales, service, negotiations and interviews every day — yet almost everything we know about it comes from students judging posed photos in labs. Bodylytics Lab is building the missing field data, and using it to make training measurably better.
The Lab exists to make every claim in a Bodylytics course defensible in front of a sceptical buyer.
A Blind Spot Hiding in Plain Sight
The classic laboratory finding is humbling: people detect deception at barely better than chance, while feeling confident they're good at it. How that plays out in a hotel lobby, on a shop floor, or across a negotiation table — with professionals, in real time, under pressure — has hardly been studied at all. Three gaps keep the field blind:
● Students in labs ● Frontline professionals
Wrong People
Most published studies test university students. The people who read faces for a living — hotel staff, retail teams, salespeople, nurses — are almost absent from the data.
Wrong Conditions
Posed photographs and untimed judgments, when real work means live interactions and split-second reads. Speed, confidence and context barely appear in the literature.
● One-shot studies — Measured over months
No Follow-Through
Almost everything is one-shot. Whether skills actually improve with training — and how fast they fade without practice — is largely unmeasured outside the lab.
Why We're Doing This
Awareness. Workplaces run on folklore — crossed arms, “liars look left”, gut feel. Publishing real numbers on what professionals actually see, miss and believe raises the level of the whole conversation about nonverbal skills at work.
Better training. Every finding flows straight back into our courses, live sessions and materials. If the data shows a myth persisting in hospitality, our hospitality content confronts it head-on. If a skill decays after 90 days, our follow-up practice targets exactly that window.
The research isn't a side project — it's the feedback loop that keeps the training honest and current.
The Claim Register
Every claim taught in a Bodylytics course goes through the register: what the evidence says, the verdict, and what happened to it in the curriculum. Claims that fail are removed — and we say so.
What We Study
Every course, assessment and live training uses established measurement paradigms — emotion recognition, rapid-judgment accuracy, balanced truth/lie base rates — so results are comparable with the published literature. What the lab can't produce is our population: frontline professionals answering in a work context, with response-time precision and follow-up over months.
How the Data Is Collected
Three sources feed the programme — always reported separately — and everything passes through one anonymization barrier before it can be analysed or published.
Free quiz takers
Anonymous answers and response times. No account, name, email or IP — a random identifier ties one attempt together, nothing more. Industry optional, self-declared.
Academy learners
Quiz and assessment responses — correctness, confidence, response time, cues relied on — plus an optional questionnaire recorded in bands, never free text. Opting out takes one click and never affects scores, certificates or pricing.
Team-training delegates
Before-and-after benchmarks. Employers see group aggregates only — never an individual delegate’s score — and every delegate is told exactly that before the first question. People answer honestly when the result can’t follow them into a performance review.
Anonymization barrier
Names, emails and account IDs are replaced by a salted one-way hash that cannot be reversed. The salt exists only inside the research database. Opted-out users are excluded entirely.
Published aggregates
Aggregates only, never any group smaller than 20 participants (k ≥ 20 suppression) — reported openly as industry field data, not population-representative research.
How Our Tests Are Built
A score is only worth what the test behind it can defend. Every BodyLytics assessment is built and maintained under six measurement rules:
One skill per question
Each item tests a single, nameable judgment — no double-barreled questions, no trick wording, and wrong answers that are plausible rather than jokes.
Graded on the server
Answers are scored centrally, so a score can’t be manufactured — which is what makes the dataset, and every certificate built on it, worth anything.
Parallel before/after forms
The post-training test uses different questions of matched difficulty, so improvement measures skill — not memory of the first test. Per-answer feedback is withheld between sittings.
Enough questions to mean something
Every skill dimension is measured by multiple items per sitting — a subscale built on one or two questions is noise wearing a number.
Speed and confidence measured
Response times screen out click-throughs; self-rated confidence is compared against measured accuracy — the gap between the two is one of the most consistent findings in the deception literature.
Versioned, piloted, pruned
Live questions are never silently edited — changes create a new version, so results stay comparable. New banks are piloted first, and item statistics retire questions that fail to discriminate.
The Rules We Publish Under
No Invented Statistics
The famous claim that “93% of communication is nonverbal” is a misreading of two narrow 1967 studies — you will not find it in our materials, marketing or findings. If we cite a number, it comes from our own dataset or a named, checkable source.
No Numbers Before the Sample Earns Them
Percentile comparisons stay hidden until enough real results exist to make them honest. No group smaller than 20 people is ever reported. We’d rather show you nothing than a number we can’t stand behind.
Selection Bias, Disclosed
People who take a body-language test are not a random sample of humanity, and we never pretend otherwise. Every publication states who the sample is, how they got there, and what that means for reading the result.
Corrections, in Public
If an error is found in something we published, we correct the publication itself and say what changed and why — not quietly swap the page. Challenges to our methodology get answered, not deleted.
What We Never Share
Transparency has edges, and it's only fair to say where ours are — and why.
Individual results
Your score belongs to you. Employers, clients and publications only ever see aggregates that clear the k ≥ 20 rule.
Identities
Research analysis runs on one-way pseudonyms. There is no “de-anonymize” button, by design.
Question banks & answer keys
A published test is a dead test — the same reason standardised assessments everywhere keep their items confidential. We publish how tests are built, not the questions.
Client-identifiable results
A client’s benchmark outcomes are theirs; they appear in research only inside industry-level aggregates, never attributably, unless the client asks to be named.
What This Research Is — and Isn't
BodyLytics is a commercial training company doing field research, not a university department. We don't claim institutional ethics-board approval, and we don't dress industry data up as academic peer review. Instead, the entire method is published on this page — collection, anonymization, thresholds, test construction, and the limits of our sample — so anyone, including academics and journalists, can scrutinise the work on its merits. Bona fide researchers who want aggregate data are welcome to ask; requests are considered case by case, under the same anonymization rules.
Security note: rate limits and integrity checks protect the assessments against automated abuse. We don't publish their exact mechanics, for the obvious reason.
Taking Part
Academy members can manage participation — and fill in the optional questionnaire — any time from the Research tab in Account Settings. Questions about the programme, methodology, or press enquiries:
First publications planned once the dataset clears our minimum-group thresholds.
Download the BodyLanguage IQ App
Practice reads, track your accuracy, and keep skills sharp between sessions.