Frequently Asked Questions

Pharma FMCG People & Culture Healthcare

Questions about ad testing and EchoDepth

From the science behind FACS and VAD scoring to pharma research applications, culture measurement, GDPR compliance and how to get started. For deeper reading, see our Insights articles and Glossary of terms.

Ad testing

What is ad testing?

Ad testing shows an ad to a sample of people from its target audience before or after launch to find out how it works. Pre-testing happens before the media spend, so you can choose between edits, fix weak moments or drop an ad that doesn't land. EchoDepth pre-tests ads by measuring viewers' facial response second by second, alongside a short survey.

How does facial coding work for ads?

Each viewer watches the ad on their own webcam. Software detects the face on every frame and scores the facial muscle movements defined in the Facial Action Coding System (FACS), such as AU12, the lip-corner pull of a smile. EchoDepth turns those Action Units into valence, arousal and dominance, then averages across viewers to give a response curve for every second of the ad.

Is facial coding accurate?

Automated coding of Action Units is well established, and peer-reviewed work has found that automatic facial coding of people watching commercials relates to their own ratings of the ads and brands. But a facial expression is not a direct readout of a feeling, and people watching alone often show subtle expressions. That is why EchoDepth reports results across many viewers, with the Action Unit detail behind each score, and pairs it with a survey. The UK ICO has warned that emotion AI can be immature, so we publish our method and its limits.

How many viewers do I need to test an ad?

EchoDepth recommends 15 or more viewers per audience segment for each ad or edit. Every group you want to report on separately, such as under-35s and over-35s, is its own segment and needs its own 15. We agree the sample with you based on your audience and the decision you need to make.

What is the Cavefish Emotional Index (CEI)?

The Cavefish Emotional Index (CEI) is a 0 to 100 score of how people feel, read directly from what they write or say. It is the average of five readings, each 0 to 100 where higher is healthier: Advocacy, Goodwill, Engagement, Drive and Assurance. Scores fall into four bands: Strong (80+), Good (60–79), Watch (40–59) and Act (below 40). In an ad test, the CEI is read from what viewers say and write about the ad after watching. It is separate from the Emotional Response Curve, which shows second by second what their faces did.

More detail: how an ad test works and our methodology, the response curve and the CEI.

About EchoDepth Insight

What is EchoDepth Insight?

EchoDepth Insight is a remote research and product testing platform that captures emotional response in real time using facial FACS analysis and VAD scoring. A participant opens a browser link on their own device, consents, and views the stimulus while the platform scores 44 facial Action Units per frame, converts them to valence, arousal and dominance, and plots the Emotional Response Curve against the stimulus timeline. No hardware, facility or specialist operator is required, so a study can run across several countries in a week. It is used for concept and packaging testing, advertising pre-testing and pharma communications research, with consenting research participants. Workforce measurement uses the separate text-based product, EchoDepth Signal, because inferring emotion from biometric data in a workplace is prohibited under EU AI Act Article 5(1)(f), and employee consent does not lift that prohibition.

See how the platform works →

What are the weaknesses of traditional product testing that EchoDepth addresses?

Traditional research methods suffer from social desirability bias (participants say what they think researchers want to hear), the articulation gap (people cannot accurately describe emotional responses), post-rationalisation (conscious reflection distorts emotional memory), peer pressure in group settings, recall bias, geographic limitations, and the absence of moment-to-moment emotional data during stimulus exposure.

What research use cases does EchoDepth Insight support?

EchoDepth Insight supports pharmaceutical patient research, HCP engagement studies, drug messaging testing, FMCG product and packaging testing, concept and idea evaluation, advertising pre-testing with second-by-second emotion curves, healthcare communication research and device usability studies.

Explore all use cases →

Who does EchoDepth Insight work with?

EchoDepth Insight works with pharmaceutical companies, healthcare organisations, FMCG brands, research agencies and innovation teams who need to understand genuine emotional response — not just what research participants say they feel.

The science — FACS and VAD

What is FACS and why does it matter for research?

FACS (Facial Action Coding System) is the scientific gold standard for facial expression analysis, developed by Paul Ekman and Wallace Friesen. It maps every visible face movement to one of 44 numbered Action Units. Because AUs are largely involuntary — many cannot be consciously faked — they provide a reliable measure of genuine emotional response, independent of what the participant says or believes they felt.

Learn more about FACS on EchoDepth →

What is VAD scoring in the context of research?

VAD stands for Valence (positive to negative), Arousal (calm to excited) and Dominance (submissive to in-control). EchoDepth Insight outputs continuous VAD scores throughout a research session, providing a three-dimensional emotional profile at every moment of stimulus exposure — not a single rating at the end.

How does a remote EchoDepth Insight session work?

Participants join via any device with a standard webcam or front-facing camera — no app installation required. Consent is captured digitally. Stimuli are delivered in-session: concepts, packaging, messaging, video or interview questions. EchoDepth captures 44 facial Action Units continuously throughout the session. VAD scores are computed in real time. Key moments are flagged automatically. A structured report is generated with emotion timelines, group aggregates and segment comparisons.

Pharma and healthcare research

How does EchoDepth Insight work for pharma research?

EchoDepth Insight captures moment-to-moment emotional responses in patient interviews, HCP discussions, clinical trial participant sessions and drug marketing material testing. It detects genuine emotional reactions to messaging, packaging, brand and product concepts — without relying on articulation or post-session recall.

Explore pharma use cases in detail →

Can EchoDepth Insight be used for patient journey research?

Yes. EchoDepth Insight is particularly valuable for patient journey research because it captures the emotional response at diagnosis communication, treatment decision points, therapy adherence interviews and care experience evaluation — all moments where patients are unlikely to accurately articulate their true feelings.

Is EchoDepth Insight suitable for HCP (healthcare professional) research?

Yes. HCPs are a notoriously difficult research audience — their professional training makes them highly effective at managing their self-presentation. EchoDepth Insight captures the involuntary physiological signals that reveal genuine response to product messaging, clinical data presentation and brand positioning.

How does EchoDepth Insight improve FMCG product testing?

Traditional FMCG research is distorted by the "interesting but wouldn't buy" gap — participants give safe social responses in group settings. EchoDepth Insight captures involuntary physiological response at the moment of stimulus exposure, revealing which concepts generate genuine desire versus polite curiosity. It supports packaging concept testing, brand evaluation, new product introduction research and advertising pre-testing.

Privacy, Ethics & Data

Is EchoDepth Insight GDPR compliant?

Yes. EchoDepth processes video in memory — no raw footage is stored. Only VAD scores and Action Unit activations are output. All processing is GDPR-compliant by design. Sessions are consented and time-bound with full participant data control. On-premise deployment is available for organisations with strict data residency requirements.

Does EchoDepth claim to detect lies?

No. EchoDepth does not claim to detect deception. It measures involuntary physiological markers — facial Action Units and VAD states — during consented research sessions. All interpretation and recommendation decisions remain with qualified researchers and client teams. This distinction is fundamental and non-negotiable in all deployments.

Read our ethics approach →

How does EchoDepth handle sensitive pharma research data?

No raw video is retained. Only VAD scores and Action Unit activations are output and stored. All sessions require explicit participant consent. On-premise deployment is available for organisations with strict data residency requirements. GDPR-compliant by design.

Company & Team

Who is behind EchoDepth Insight?

EchoDepth Insight is built by Cavefish, a Cardiff, Wales-based emotional AI company founded in September 2023. The platform is built on ISO 9001 and Cyber Essentials certified infrastructure. Cavefish is an FCA Regulatory Sandbox participant and has received 13 industry awards.

Learn more about Cavefish →

Who founded EchoDepth Insight?

EchoDepth Insight was founded by Jonathan Prescott, who previously led digital performance and behavioural analytics teams at the Royal Mint and Assurant. He holds an MBA from Bayes Business School and a B.Eng in Computer Systems Engineering. He is Strategy Director of AI Wales CIC and a Data & AI Council member at Technology Connected Wales.

What is Cavefish's approach to responsible emotional AI?

EchoDepth Insight does not diagnose individuals or claim to read minds. It measures Valence, Arousal and Dominance states during consented, time-bound research sessions. All interpretation and recommendation decisions remain with qualified researchers and client teams. No raw video is retained. Cavefish will not deploy EchoDepth in contexts where demographic bias could produce discriminatory outcomes.

See also: EchoDepth vs iMotions · EchoDepth vs Culture Amp

How EchoDepth compares

When is EchoDepth Insight the right research tool?
EchoDepth adds most value in four specific situations. First, when your existing research (surveys, focus groups, interviews) keeps producing results that don't match what happens in market or in practice — you need the emotional layer beneath the stated response. Second, when you need to validate a concept, product, or campaign quickly before committing to launch — and the cost of being wrong is high. Third, when you have culture survey data that looks stable but you're seeing warning signs that contradict it. Fourth, when you're in pharma or FMCG research and social desirability bias in participant responses is a known, unresolved problem. EchoDepth is not the right tool when you need statistically representative large-scale quantitative data — it is a depth tool that captures what no survey can.
Can I add EchoDepth to my existing focus group or interview methodology?
Yes — and this is one of the most common ways agencies and in-house research teams use it. EchoDepth does not replace your existing qual methodology; it adds an emotional AI layer alongside it. Participants complete their interview or focus group discussion as normal; EchoDepth captures the emotional response simultaneously. The output is a second layer of data — the felt experience — running parallel to the stated experience from the qual. The combination is significantly more powerful than either alone, and the discrepancies between the two layers are often the most commercially valuable finding.
How does EchoDepth differ from tools like iMotions or Noldus FaceReader?
iMotions and Noldus FaceReader require specialist hardware — eye tracking rigs, biometric sensors, laboratory setups, and trained operators. Sessions must be facility-based. EchoDepth runs entirely in-browser on any standard camera-equipped device. No hardware, no facility, no specialist operator required. Participants join remotely from anywhere in the world within minutes. Affectiva (Smart Eye) and Kairos offer facial coding as an enterprise SDK integration — not a turnkey research platform with delivered reports. The practical difference: an iMotions study with 30 participants requires three laboratory visits; an EchoDepth study with the same 30 participants can begin within 48 hours of agreement.
Can EchoDepth replace Qualtrics or Culture Amp for culture measurement?
EchoDepth does not replace Qualtrics, Culture Amp, Glint, or Peakon — it adds an emotional AI layer on top of them. You keep your existing survey platform, your existing questions, and your existing reporting. EchoDepth ingests the open-text responses your survey already collects and scores them for emotional signal — Valence, Arousal, Dominance — separately from their surface sentiment. No cameras are involved. The result is text-emotion divergence analysis: identifying where employees write positively but the language carries negative emotional weight. That signal is invisible to keyword-based sentiment, regardless of how sophisticated its NLP is.
How is EchoDepth different from sentiment analysis tools?
Sentiment analysis tools — whether built into Qualtrics, Brandwatch, or standalone NLP platforms — analyse the words people choose to write. They cannot measure what the person chose not to write, what they felt before they typed, or whether their emotional state matched their written response. For research participants, EchoDepth measures facial response directly. For workforce data, it scores the emotional signature of the written response itself — Valence, Arousal and Dominance — rather than its surface polarity. This is why EchoDepth detects masked dissatisfaction — the gap between expressed and felt emotion — which keyword sentiment tools miss. See: how to measure employee sentiment.
Is EchoDepth like traditional manual FACS coding?
Manual FACS coding — where a trained human coder analyses facial muscle activations frame by frame — typically takes 1–2 hours of analysis per minute of video footage. It requires certified coders and produces data too slowly for most commercial research timelines. EchoDepth automates this process in real time: 44 Action Units are scored per frame at the moment of capture, with no manual coding step. The result is the same scientific rigour as manual FACS — grounded in Ekman & Friesen's taxonomy — delivered at the speed of a live research session.
What makes EchoDepth different from other emotional AI platforms?
Most emotional AI tools stop at the data — they return a score, a chart, or a VAD value and leave interpretation to the researcher. EchoDepth goes further: we translate raw emotional signal into structured insight, and then into specific, actionable knowledge the team can implement. We are data experts who can ingest and analyse whatever data you already hold — not just data generated inside our platform. The engagement also differs: EchoDepth operates as a research partner, not a software subscription. "Using EchoDepth from Cavefish allows us to validate ideas quickly, minimising the risk of launching a product or idea." — Gethin Thomas, CEO, Iterate.

Getting started

How do I get started with EchoDepth Insight?

Contact us to book a 30-minute discovery call. We will explore your research challenge and constraints, then provide a live demo or sandbox API access depending on your needs. Every enquiry is read by a human and we respond within one business day.

Book a discovery call →

Can EchoDepth Insight be used for advertising testing?

Yes. EchoDepth Insight provides second-by-second emotional curves across video advertising — showing exactly which moment lands, which triggers concern, and where attention drops. It supports both pre-testing before campaign launch and post-campaign emotional shift measurement against baseline.

People & Culture

Full culture measurement methodology →
EchoDepth culture radar chart — eight organisational sentiment dimensions plotted as a spider diagram: Leadership, Compensation, Product Quality, Customer Support, Innovation, Work-Life Balance, Pricing, and Reliability. Each axis scored 0 to 10 from emotional AI analysis of survey responses.

EchoDepth culture radar — emotional valence across eight organisational dimensions.

Can EchoDepth be used to measure workplace culture, not just product responses?
Yes — and it is one of the most powerful applications of the platform. EchoDepth applies dual-layer emotional scoring to culture survey open-text responses, combining what employees write with the emotional signals beneath the text. The key output is text-emotion divergence analysis: when diplomatically worded language is paired with high negative emotional markers — Disapproval, Contempt, Distress — EchoDepth flags this as masked dissatisfaction, which is significantly higher-risk than overt complaints. Reports are structured for board presentation.
What does "text-emotion divergence" mean in the context of a culture survey?
Text-emotion divergence occurs when the written content of a survey response and its emotional signature point in different directions. An employee writes "I feel we need to consider colleague wellbeing more" — measured and professional. EchoDepth scores the same response: Contemplation 0.51, Disappointment 0.29, Doubt 0.18. A standard survey platform scores this neutral. EchoDepth scores it as suppressed dissatisfaction and surfaces it as a risk signal. This divergence — between what people say and what they feel — is the most operationally significant data that traditional culture measurement misses.
What is a healthy negative-to-positive valence ratio in a culture survey?
In a well-functioning organisation, we typically see a negative-to-positive ratio below 1.2:1. Ratios between 1.2:1 and 1.5:1 indicate structural friction — usually addressable with targeted intervention. Ratios above 1.5:1 indicate systemic strain with elevated voluntary turnover risk within 12 months. A ratio of 1.8:1 or above represents a workforce that is functionally performing but emotionally depleted — the pre-exit condition most associated with sudden retention crises. See the full culture measurement methodology →

Still have questions?

Book a 30-minute discovery call. We'll walk through your research challenge and show you what emotional capture looks like in practice.

Book Discovery Call Explore the Platform

Key definitions

Employee sentiment analysis
The measurement and interpretation of employee emotions, attitudes, and morale using data-driven methods including AI signal analysis.
VAD model
Valence-Arousal-Dominance — a three-dimensional framework for classifying emotional states developed from Russell's circumplex model of affect.
Sentiment score
A numerical value representing an employee's or group's emotional valence on a scale from negative to positive, derived from multi-modal signal analysis.