- The short answer
- The options, by use case
- Comparison at a glance
- What "privacy-first" actually means
- Questions worth asking any vendor
The short answer
For in-person events in 2026 the realistic shortlist is short. Zenus AI is the most established choice in the events industry and the only option here that bundles audience analytics with lead retrieval, access control and a separately licensed facial recognition product. EchoDepth Events is built for measuring reactions on a stand or in a session and acting on them during the show, with an anonymous tier that needs no consent process. iMotions, which absorbed Affectiva in 2025, is the research instrument: multi-sensor, lab-grade, and the wrong shape for three unattended days in an exhibition hall. MorphCast and MoodMe are developer engines — MorphCast runs in a web browser and suits virtual events, while MoodMe runs on a device you build yourself. Realeyes, still named on many older lists, no longer sells emotion analytics at all: it now sells human verification, and its ad-testing business was spun off as Adverteyes.
How this was checked. We make EchoDepth Events, so treat our own entry with appropriate scepticism. Every competitor fact below was taken from that vendor’s own website on 23 September 2026, and quoted in the vendor’s wording where we quote at all. Where a vendor does not publish something — usually price — we say so rather than estimate it.
The options, by use case
Best established events platform: Zenus AI
The name most people in the events industry will recognise, and the vendor with the deepest roster of industry clients: its site lists IAEE, IMEX, PCMA, Freeman, Informa, Dell and ServiceNow. Zenus describes its core product as “ethical facial analysis” on its own smart cameras, measuring impressions, dwell time, positive sentiment and demographics, and states that video is “never stored nor transmitted”. Around that it sells operational tools — automated lead retrieval, attendance tracking and access control, dynamic displays — and, as a separate self-hosted licence, a facial recognition product whose listed uses include event check-in.
Choose Zenus if: you want one supplier for analytics and event operations, you run large shows where its client list carries weight with stakeholders, or you need hardware that keeps working offline. Price: quote only. Read our full Zenus AI review and what drives a Zenus quote.
Best for acting during the show: EchoDepth Events
Our own platform, built for exhibition stands, conference sessions and product launches rather than adapted from a research or retail tool. Video is processed on an edge unit at the venue and each frame is discarded immediately. It runs in two tiers: Standard measures footfall, dwell time, movement and crowd density anonymously with no consent requirement, and Deep adds full FACS Action Unit analysis with delegate opt-in at registration. The distinguishing output is zone-level confusion detection, flagged live, so a misfiring panel or demo can be fixed before the next show day opens.
Choose EchoDepth Events if: you want measurement you can act on during the event, you need an anonymous tier that clears a UK or EU compliance review without a consent process, or confusion — not just positive sentiment — is the signal you care about. We do not sell lead retrieval, check-in or access control; if you need those from the same supplier, that is Zenus. Price: from £1,500 per event (Standard) and £4,995 per event (Deep), ex. travel — full pricing.
Best for research-grade studies: iMotions (now including Affectiva)
iMotions, part of Smart Eye, is a multimodal human-behaviour research platform combining facial expression analysis with eye tracking, EEG, galvanic skin response, ECG and more. Affectiva, formerly a separate emotion-AI company, joined iMotions in 2025, and its SDK and media-analytics products are now sold under the iMotions name. It is aimed at academic labs, UX teams and ad testing with consented participants, and the homepage claims use at nine of the ten top-ranked universities.
Choose iMotions if: you need defensible, publishable, multi-sensor evidence from a controlled study. It is far more capable as an instrument than anything else here and correspondingly more involved to run; it is not built for unattended measurement on a stand. Price: commercial module subscriptions from $3,400 / €3,000 each per year, CORE module required, hardware extra.
Best for virtual events and web content: MorphCast
A JavaScript emotion engine that runs client-side, in the browser or app, and outputs signals “without sending video or picture to any server”. It covers Ekman’s six core emotions plus valence and arousal, and publishes an explicit position on the EU AI Act. Its homepage leads with education and healthcare; it does not describe physical venues.
Choose MorphCast if: your measurement problem is on a screen — a virtual event, a webinar, online content testing. It is also the most transparent vendor here on price. For a physical exhibition hall it is the wrong shape of tool. Price: $199, $499 or $999 per month by stream hours, with a 60-day free trial.
Best for embedding in your own product: MoodMe
An on-device SDK from MoodMe (Belgium and USA). It states its algorithms “run 100% locally on the user’s device” and that faces are never stored or sent to any cloud, classifying seven emotions plus engagement, age and gender, alongside augmented-reality face tracking. Its published customer examples are mostly brand activations and AR campaigns rather than measured trade shows.
Choose MoodMe if: you have engineering resource and want emotion detection as a component inside your own kiosk, app or installation. Price: not published.
No longer an option: Realeyes
Realeyes still appears on many “best emotion AI” lists, but its site now describes a human-verification product, VerifEye, for banks, gaming and marketplaces. Its creative and ad-testing business was spun off in 2025 as Adverteyes, which measures consenting panellists watching ads through their webcams. Neither is an events product.
Comparison at a glance
| Platform | Built for | Processing | Identifies people? | Published price |
|---|---|---|---|---|
| Zenus AI | In-person events, retail | On its own smart cameras; states video is never stored or transmitted | Analytics: no. Sells a separate facial recognition licence for check-in and access control | Quote only |
| EchoDepth Events | Stands, sessions, launches | Edge unit at the venue; frames discarded | No | From £1,500 / event (Standard), £4,995 / event (Deep) |
| iMotions (incl. Affectiva) | Research labs, UX, ad testing | Consented study participants; desktop and cloud | No | Modules from $3,400 / €3,000 per year, ex. hardware |
| MorphCast | Web and app experiences | In the browser; no video sent to servers | No | $199–$999 / month; 60-day trial |
| MoodMe | Embedded apps, AR activations | 100% on-device | No | Not published |
| Realeyes | Human verification (no longer emotion analytics) | — | — | — |
What "privacy-first" actually means here
Privacy-first emotion analytics means measuring how an audience reacts without creating a record that identifies anybody. In practice the distinction that matters is architectural rather than contractual: a system either processes video where it is captured and discards the frame, or it transmits images to a server and relies on policy to govern what happens next. Only the first kind is outside the scope of GDPR's biometric provisions, because there is no biometric data to protect. The test to apply to any vendor on this page is simple and worth asking directly: does a face image ever leave the device, is any identifier retained that could be linked back to a person, and can the system recognise the same individual twice? A platform that answers no, no and no is doing anonymous analytics. A platform that answers yes to any of them is doing something that needs explicit consent.
Questions worth asking any vendor
- Does a face image ever leave the capture device, and what exactly is transmitted?
- Can the system recognise the same person on two different days? If yes, it is identification, not analysis.
- Which product does the quote cover — analytics only, or recognition as well?
- What is an accuracy figure measured against, on what data, and when?
- Under the EU AI Act, does the deployment need transparency notification under Article 50(3)?
- Does the output arrive during the event or after it — and can anything be changed while it still matters?
Vendor details change; verify current capability and compliance posture directly with each company. See how EchoDepth Events works in practice — book a 30-minute demo using real event data, or read our GDPR compliance guide for the architecture behind the anonymous tier.
Last verified: 23 September 2026.