Badge-based check-in, whether via RFID, QR code, or simple visual verification, has always required an attendee to present something — a physical badge, a phone, a confirmation code. In 2024, a small number of large conferences began piloting a genuinely different model: facial-recognition check-in, letting a pre-enrolled attendee walk through an entrance and be identified automatically, without presenting anything at all.

How the Pilots Actually Worked

The trials that ran during this period generally required attendees to opt in explicitly during registration, submitting a photo that was matched against a facial-recognition system at dedicated check-in points, distinct from the standard badge-based check-in lanes that remained available for attendees who didn’t opt in. This opt-in structure, rather than a blanket facial-recognition requirement, reflected both genuine privacy sensitivity around biometric data and, in many jurisdictions, specific legal requirements around biometric data collection that made an opt-out or non-consensual approach a much higher legal risk than a clearly voluntary opt-in trial.

Enrolled attendees using the facial-recognition lane reported meaningfully faster check-in — walking through without needing to present a badge or phone at all — though the overall throughput benefit for an event depended heavily on what share of total attendees actually opted into the trial, since a small opted-in population using a dedicated fast lane doesn’t meaningfully change overall event-wide check-in speed on its own.

Privacy Concerns Limited Enrollment More Than Technical Limitations

The technology itself performed reasonably well in these trials — facial-recognition accuracy for a pre-enrolled, cooperative population under controlled conditions is a comparatively well-solved technical problem. The limiting factor on adoption was overwhelmingly attendee comfort with biometric data collection rather than technical performance, and opt-in rates across the trials that have shared data publicly were meaningfully lower than organizers had hoped, reflecting broader public wariness about facial-recognition technology and its history of more controversial uses in law-enforcement and surveillance contexts.

This wariness existed even though the conference use case was narrower and more clearly bounded than the surveillance applications that had generated the most public controversy around facial recognition generally — attendees evaluating whether to opt in weren’t necessarily distinguishing between different facial-recognition use cases with different privacy implications, and the technology’s broader reputation affected adoption of even a narrowly scoped, opt-in conference application.

It was also worth being precise about where accuracy concerns genuinely applied. Independent benchmarking, most notably the U.S. National Institute of Standards and Technology’s ongoing Face Recognition Technology Evaluation program, has documented that top algorithms perform very well in cooperative, one-to-one verification scenarios like a pre-enrolled attendee walking a check-in lane, while error rates rise substantially in uncontrolled one-to-many identification against large populations — a materially harder problem than conference check-in. The distinction mattered for organizer messaging: the demonstrable accuracy of the specific verification task didn’t, on its own, resolve attendee concerns rooted in the technology’s broader and more controversial identification-at-scale applications, and organizers who conflated the two in their communications tended to lose rather than build trust.

Biometric data collection is subject to specific, and geographically varying, legal requirements beyond general data-privacy law — some jurisdictions have dedicated biometric privacy statutes imposing stricter consent and data-handling requirements than general privacy law requires for non-biometric data. This meant facial-recognition check-in trials were more feasible in some jurisdictions than others, and organizers running multi-city or international conference series sometimes found they could pilot the feature at one event location but not another, purely due to differing local biometric-privacy legal requirements.

Data Retention and Deletion Practices Faced Scrutiny

Given the heightened sensitivity of biometric data specifically, trials that handled this well built explicit, short retention windows for enrollment photos and facial-recognition templates — deleting the data shortly after the event concluded rather than retaining it indefinitely for potential reuse at future events — and communicated this retention policy clearly to attendees considering opting in. Trials that were vaguer about retention and deletion practices generally saw more attendee hesitation and lower opt-in rates, suggesting that clear, favorable data-handling commitments meaningfully affected attendee willingness to participate.

Vendors Positioned This as an Early-Stage Category

Vendors selling facial-recognition check-in systems to conference organizers were generally candid that this remained an early-stage category rather than a mature, broadly proven solution, framing initial deployments explicitly as pilots intended to gather real-world adoption and satisfaction data before recommending wider rollout, a more cautious go-to-market posture than some vendors had taken with earlier, less sensitive event-technology categories.

Frequently Asked Questions

Did facial-recognition check-in require all attendees to participate?

No — the trials that ran in 2024 generally required explicit opt-in during registration, with standard badge-based check-in lanes remaining available for attendees who didn’t opt in, reflecting both privacy sensitivity and legal requirements around biometric data collection.

Was the technology itself the main barrier to wider adoption?

No — facial-recognition accuracy for a pre-enrolled, cooperative population performed reasonably well technically. The main limiting factor was attendee comfort with biometric data collection, with opt-in rates lower than organizers had hoped due to broader public wariness about facial-recognition technology generally.

Why could facial-recognition check-in trials run in some locations but not others?

Biometric data collection is subject to specific legal requirements that vary by jurisdiction, with some regions having dedicated biometric privacy statutes imposing stricter consent and data-handling rules than general privacy law, making the feature more legally feasible in some locations than others.

What data retention practices did more successful trials use?

Trials with clear, short retention windows — deleting enrollment photos and facial-recognition templates shortly after the event rather than retaining them indefinitely — and transparent communication about this policy generally saw better attendee opt-in rates than trials that were vaguer about data handling.

For related coverage, see our piece on RFID vs. QR code lead retrieval.