Loopyah Icon

· 21 min read

Should your conference use facial recognition for check-in?

Featured image for Should your conference use facial recognition for check-in? article

Facial recognition can make conference check-in easier for someone who has opted in, uploaded a usable photo and arrived at a terminal that works. Whether it makes your conference better is a different question. You still have to admit everyone who declines, resolve failed matches, issue badges and protect the information you collected.

For most paid conferences, I'd start with a well-run QR ticket process. Consider facial recognition only when you can name a problem that process doesn't solve, offer an equally usable alternative and demonstrate a worthwhile improvement across the whole arrival experience. A quick match in a supplier demo doesn't establish any of those things.

There is a sensible version of this technology. A returning delegate chooses it because they find it convenient, understands what happens to their information and can change their mind without an argument at the door. Your team gains something measurable, and the extra cost and responsibility make sense. That's the version worth evaluating.

This guide explains how to make that decision, what to ask a supplier and how to test the proposal before you put paying attendees in front of a camera. Legal requirements depend on where and how you operate. The UK, EU and Illinois examples below explain differences you need to review for your own event.

Start with the arrival problem you actually have

Think about your last opening morning. Were attendees struggling to find tickets? Were your printers slow? Did a registration export miss late bookings? Was everyone joining the same queue because the signs weren't visible? Each problem can produce a long line, but each needs a different response.

A face match might reduce the time someone spends finding a booking. It won't make a badge printer run faster or settle a question about which workshop a delegate bought. If those tasks consume most of your processing time, replacing the lookup method may barely change when people enter the conference.

Watch a complete arrival before buying anything. Start where a delegate first needs directions and finish when they have whatever they need to enter. Include the person whose ticket works perfectly and the person whose colleague booked under the wrong email. Your actual problem becomes much easier to describe when you stop timing only the scan.

You might discover that attendees find their tickets quickly, but wait while staff search for lanyards. Or that badge collection flows well until a coach arrives. Address the actual cause. Our guide to event logistics helps connect those arrival decisions to the rest of your operating plan.

Write a sentence your supplier can challenge. "We need to reduce delays caused by delegates finding their booking while carrying bags" gives you something to test. "We want a more innovative arrival experience" gives a sales team room to show you almost anything.

Also ask what your attendees gain. You may like the appearance of a camera terminal. A delegate may prefer opening a familiar ticket and keeping their face out of another database. Neither preference tells you the other person's answer. Research the proposed experience with the people who will actually pay to attend.

Understand what the camera would do

The ICO's explanation of biometric recognition distinguishes verification from identification. Verification checks a claimed identity. Identification searches for a person among stored records. That difference matters when you ask what the supplier means by facial recognition event check-in.

In a verification flow, a delegate might present a booking reference before the system compares their face with the record linked to that booking. In an identification flow, they might approach a camera that searches the enrolled attendee list. Don't accept a demonstration of one as proof that the other will work at your conference.

Ask the supplier to show the enrollment step too. Where does the initial image come from? How does the system connect it to the right ticket? What prevents somebody uploading another person's photograph? A successful comparison tells you that faces appear to match. It doesn't independently prove that the original enrollment was legitimate.

The supplier may convert an image into a biometric template, a mathematical representation used for comparisons. Find out whether it also retains the original photograph. "We store templates" is incomplete if support logs, camera captures or backups contain images as well. Ask for the actual information collected at each stage.

Ticket transfers need particular attention. Imagine a company buying a conference pass for an employee who later leaves. The company changes the named delegate, but the biometric record still points to the original employee. Ask the supplier to demonstrate how the old enrollment loses access and how the replacement delegate receives a fresh choice. Nobody should inherit somebody else's template, and an old enrollment should never quietly remain a second way to use the same ticket.

Keep your proposal limited to the purpose you can explain. Admitting enrolled delegates at a terminal is different from recognizing everyone walking through the lobby, recording session movements or helping exhibitors identify visitors. Don't buy a check-in system and let those other uses appear later as settings someone casually switches on.

A profile photograph supplied for a speaker page or networking directory isn't an invitation to enroll the person in biometric check-in. Establish the appropriate collection and permission process for this separate use. Your marketing team's access to a folder of headshots doesn't settle the question.

Ask how the system deals with a photograph presented to the camera rather than a live attendee. Suppliers may describe checks intended to detect that attempt. Have them explain what those checks cover, what they don't and how they affect people who struggle with the capture instructions. A demonstration using a printed picture can expose a basic weakness, but passing that demonstration doesn't prove protection against every impersonation attempt. The admission decision still needs the ticket rules you agreed.

Keep any other identity requirements separate. Matching an enrolled face doesn't automatically verify age, professional status or a government identity document. If a restricted session requires an additional check, define who performs it and where it happens. Include that step in your arrival timing. Otherwise, your demonstration can look fast because it omits the very check your conference needs, and the team discovers the missing task when the first delegate arrives.

Make declining a normal way to attend

If you're relying on people's agreement, design the alternative before asking for it. A delegate should be able to understand the choice while booking or preparing for the event, with enough time to ask questions. The first explanation shouldn't happen while they're holding up a queue outside the keynote.

The ICO's current guidance on lawful biometric processing says UK biometric recognition requires a lawful basis and a separate special category condition. It identifies explicit consent as likely to be the most appropriate condition in many cases. Where you rely on consent, the choice must be genuine, with a suitable alternative and withdrawal without detriment.

Translate that into the event someone experiences. They can choose ordinary ticket check-in, get the same conference access and receive the same badge. Staff don't ask them to justify their preference. Your directions explain both methods plainly, and enough people work the ordinary route to keep it usable during the busiest arrival period.

An alternative isn't convincing if it exists only in the terms while the registration team steers everyone toward a camera. Nor is it convincing if you remove most ordinary stations because you assumed almost everyone would enroll. Plan that route using realistic demand, including people who enrolled but change their minds on arrival.

Be careful with group bookings. An assistant can buy tickets for a department, but that doesn't automatically establish each delegate's informed agreement to facial recognition. Communicate with the individual whose biometric information you propose to use. Design the booking and invitation process so nobody has to make that decision for their colleagues.

Withdrawal needs an operational answer. Someone opts in, then emails you before the conference to withdraw. Who updates the biometric system? How does the door team know the ticket still admits them? What happens to their template? Test that request while you can still change the setup, rather than asking support to invent the procedure later.

Explain what the ordinary route requires. If attendees can present a QR ticket or get help finding their booking, say so. Don't label the alternative "manual exceptions" in your public directions. People who exercise a valid choice shouldn't arrive feeling as though they've done something wrong.

Check the rules for your event's jurisdictions

Don't buy on the strength of "GDPR compliant" in a proposal. That phrase doesn't tell you whether your particular use is justified, whether the enrollment process works or whether the supplier's contract matches its operation. Ask the person reviewing your event's privacy obligations to assess the actual system and proposed purpose.

For EU events, Article 9 of the GDPR generally prohibits processing biometric data for uniquely identifying a person unless an applicable exception permits it. You also need the relevant Article 6 basis. An ordinary ticket contract or a general legitimate interest doesn't, by itself, answer the special category question.

Assess necessity honestly. If a less intrusive admission method meets the event's needs, describe why you still propose biometric recognition and whether that justification is adequate. Security, convenience and attendance reporting are different purposes. Don't combine them into a broad sentence that makes the system sound indispensable without examining each one.

For the UK, review the ICO guidance against the current law when approving your plan. At the time of writing, the ICO flags its biometric guidance as under review following the Data (Use and Access) Act. That notice is a reason to check the position before launch, especially if your conference is months away.

The Illinois Biometric Information Privacy Act illustrates why a US event needs a separate review. Its requirements include a public retention and destruction policy, specified written information before collection and a written release. Its definition excludes photographs but includes scans of face geometry. Calling something a photo upload doesn't resolve how the resulting biometric processing is treated.

Other states and countries have their own rules, and applicability can depend on more than the venue address. Establish the relevant entities, attendees, processing locations and suppliers with qualified counsel. Don't take an approval for one conference and reuse it unchanged for another market.

In the US, the FTC's biometric policy statement also addresses unfair or deceptive practices. Its concerns include unsubstantiated accuracy claims, unexpected collection, supplier practices, training and monitoring. Your own attendee communications matter alongside the supplier's promises.

Give legal review something concrete. Provide the enrollment screens, camera locations, supplier terms, deletion process and alternative check-in journey. A description saying "optional facial recognition" leaves too much unexplained. The review should be able to change the deployment, including the decision to proceed at all.

Ask where the information goes and when it leaves

You need a clear explanation of who receives the photo, who creates the template and who can retrieve either. Include equipment suppliers and services working behind the main contractor. A supplier's brand on the terminal doesn't necessarily identify every business involved in processing the information.

The ICO's guidance on biometric accountability explains controller and processor responsibilities and the need to consider a data protection impact assessment. This is a review of how the planned processing could affect people and how you would reduce those risks. Complete the relevant assessment before collecting real attendee biometrics.

Ask your supplier to explain processing in language your operations lead can follow. Does matching happen on the terminal or through a remote service? Which countries host the information? Where can support staff access it? What happens when a device is replaced? Get the answers in writing and have the relevant transfer arrangements reviewed.

Then separate the records by purpose. You may need ordinary ticket records after the conference for accounting or disputes. That doesn't establish a reason to keep a facial template for the same period. Decide what the biometric use actually requires and set a justified retention rule for those records.

The ICO's biometric security guidance discusses template protection, original samples and storage limitation. A template shouldn't be treated as harmless simply because it isn't a photograph. Ask the supplier how its design protects against misuse and what evidence supports the answer.

Make deletion a process someone owns. Agree what happens to templates, original images, temporary captures, diagnostic records and backup copies. If backups expire through a separate cycle, understand and describe that accurately. A promise to "delete after the event" is too vague when nobody knows which copies it covers.

Ask explicitly whether the supplier uses enrollment material to train models or improve its own services. Also ask whether information can be reused across your future events or shared with other customers. Those are separate uses to examine, rather than conveniences you should silently accept because they reduce setup next time.

Your attendee explanation must match the answers. Our guide to writing an event privacy notice covers the wider booking journey. Biometric check-in needs its own specific explanation within that work, including the real purpose, recipients, retention decisions and usable route for questions or rights requests.

Before signing, ask what happens if the supplier discovers an exposure during opening morning. Who contacts your event lead, through which channel, and what information will they provide? Agree how they preserve relevant evidence and help your team assess affected records. Your event's incident process should handle any required notification decisions. A generic support response target doesn't explain who answers an urgent biometric information problem while paying attendees are still arriving.

Treat accuracy claims as the start of a test

"Highly accurate" doesn't tell you how the system will behave in your lobby. You need to know what the supplier measured, which system version it tested and whether the test resembles your proposed use. Ask for independent evidence relevant to the matching method, alongside a demonstration with your actual equipment.

NIST's face recognition demographic evaluation explains that false negatives depend strongly on image quality, and poor photography can create demographic differences. Its work also examines false positives. Don't turn laboratory results into a guarantee about a complete check-in installation at your conference.

A false rejection happens when the system doesn't recognize the enrolled delegate standing there. A false acceptance happens when it matches someone incorrectly. The consequences differ. One can delay a valid attendee; the other can bring up the wrong booking or admit the wrong person if the surrounding process permits it.

Ask how the supplier balances those errors. A setting that accepts more uncertain matches may reduce retries while increasing incorrect matches. You don't need to become the software engineer, but you do need an explanation of the tradeoff and authority over changes that affect your admission process.

Include your real arrival conditions in testing. People arrive at different heights, use mobility aids, wear head coverings and glasses, and may find the camera instructions difficult. The test should reveal whether they can use the station comfortably and whether the alternative is obvious when they prefer it.

You also need a way to identify uneven outcomes without collecting more sensitive information than you can justify. Agree the evaluation method with your privacy and accessibility reviewers. A small rehearsal can find obvious failures, but it cannot prove that the system performs equally well for every group attending your conference.

Staff need neutral language for a failed match. "Let's use your ticket instead" keeps the conversation focused on admission. "The system doesn't recognize you" can turn a device problem into an uncomfortable public discussion about someone's identity. Don't keep asking a paying delegate to pose while the queue watches.

Prove the whole process works before opening day

Run a pilot that begins before arrival. Invite a suitable test group through the same explanation, choice and enrollment steps that delegates would use. Track where people need help and whether they understand the alternative. If the photo upload creates support work, include that work in the evaluation.

At the venue, use the actual camera position, lighting, network and badge setup. An empty meeting room demonstration won't expose the same problems as sunlight across a terminal or a crowded registration desk. Ask the venue team to help reproduce the conditions you expect when doors open.

Compare against a properly prepared ordinary ticket process. In Loopyah, door staff can scan QR tickets using their own phones, with check-in access assigned separately. Our ticketing page describes that workflow. Use the exact setup you intend to operate when evaluating whether additional biometric equipment solves a remaining problem.

Time the complete journey through both methods. Include enrollment support, waiting, successful admission, badge collection and help after a failed attempt. Record how many people finish without assistance and how long unresolved cases take. Averages alone can hide the few attendees who spend most of the opening session at the help desk.

Keep the comparison fair. Don't put experienced supplier staff on facial recognition while giving the ordinary lane an unbriefed volunteer and an unfamiliar phone. Use comparable preparation, include busy arrival bursts and record any help the supplier provides behind the scenes. Alternate the order of the tests where practical so people aren't simply faster because they already know where to stand. You want to understand the method's contribution before paying to reproduce the result at your event.

Test difficult bookings deliberately. Use a refunded ticket, a duplicate attempt, a changed attendee and an enrolled person whose access doesn't include the opening workshop. Recognition must connect to the current admission decision. Finding a face in yesterday's attendee list isn't enough to establish a valid entitlement today.

Take the network away during the rehearsal. Find out what keeps working and what stops. If matching continues locally, establish how the station knows whether a ticket was refunded or used elsewhere. If it stops, demonstrate how staff switch to the approved alternative without losing track of admissions.

Don't assume ordinary ticket scanning is automatically a safe fallback under every outage either. Confirm the provider's actual behavior, particularly across disconnected entrances. Our QR ticketing guide explains the questions around ticket validation. Test both systems under the conditions your venue plan relies on.

Agree stopping conditions before the pilot. Repeated wrong matches, an unusable alternative, unexplained data retention or a fallback that staff can't execute should prevent rollout. A lower enrollment rate may instead mean the financial case needs recalculating. Different findings require different decisions, and your supplier should know that a pilot can end with rejection.

Record the approved configuration with the pilot results. Ask the supplier to flag changes to the matching software, capture process or hosting before your event. A different configuration may need more testing or another privacy review. Your agreement should make clear who approves those changes, so the setup you assessed is the setup your door team receives when they arrive to open registration on the conference morning.

Work out what the convenience actually costs

Ask for a complete quote covering the proposed event, rather than a price per recognition. Include equipment, installation, connectivity, integration, support and any minimum purchase. Add your team's time for privacy review, attendee questions, rehearsal and cleanup. Those costs belong to the decision even if they appear in another department's budget.

Be especially careful with projected staffing savings. If you still need greeters, badge staff, ordinary ticket stations and a help desk, a camera may change their tasks without reducing paid hours. Don't call that a saving until your staffing plan identifies which shifts you can actually remove while maintaining the service.

Use the pilot to revise your event staffing plan. Count the people supporting enrollment before the event as well as the people working the lobby. If the supplier sends its own technicians, establish what your team would have to do when those technicians aren't present at a future show.

Lower participation can change the result. You might pay the same equipment and integration costs while fewer delegates choose the service. Keep the ordinary route adequately staffed and test the budget under a less optimistic enrollment assumption. Don't quietly make declining less convenient to rescue a purchase you've already committed to.

There can still be a reason to proceed without direct labor savings. Perhaps a voluntary repeat-attendee group finds the process easier and your pilot shows an improvement they value. Describe that as a service investment, decide what you're prepared to spend and assess it against other ways to improve their arrival.

Avoid assigning speculative revenue to the technology. A faster check-in doesn't prove someone buys another ticket, renews a sponsorship or spends more at the bar. Measure what happened, and keep any broader commercial hypothesis separate until you have evidence. An honest small benefit is more useful than an impressive return calculation built on guesses.

Give your door team an admission plan they can use

Once approved, turn the decision into an operating brief. Your team needs to know who can use each route, where to direct people and who authorizes an exception. Keep the directions short enough to use during a busy morning, with a named person available for questions that need a longer answer.

Keep enrollment away from the main arrival queue unless your tested design specifically supports it. Asking an unenrolled person to read a notice and upload a photo at the terminal can delay everyone behind them. Let them attend through the ordinary route, then consider any later enrollment request through the approved process.

Position cameras so the intended user understands when recognition starts and people using the alternative don't have to enter the capture area. Have the supplier demonstrate the actual field of view. A sign saying participation is optional won't fix a camera that collects people before they've chosen a route.

Preserve an accessible journey through both options. A station might be easy for a standing tester to use but awkward for somebody who can't reach or align with it. Plan assistance without assuming everyone wants to explain their needs in front of other delegates. Keep the alternative nearby and clearly indicated.

Train staff on suspected wrong matches separately from routine failures. A successful-looking screen can still be wrong. They need to stop the affected admission, protect the displayed information and contact the designated lead. They shouldn't improvise by reading out personal details until someone in the queue confirms them.

Give the lead authority to suspend recognition while ordinary admission continues through the tested process. Rehearse the communication with the supplier and venue. If turning off the cameras creates an unmanageable queue, you haven't finished planning the fallback, regardless of how reassuring the software demonstration looked.

Decide what would justify using it again

After the conference, review the original problem alongside the pilot and live results. Did delegates get admitted more easily? Did ordinary ticket holders receive a comparable service? How much support did enrollment require? Did the equipment reduce the delay you identified, or did it move the work into another part of the journey?

Review complaints and awkward moments with the people who handled them. A successful admission count won't tell you that someone felt pushed to participate, or that staff repeatedly asked a delegate to retry before offering help. Those details affect whether the experience was worth repeating.

Ask attendees about the particular service they used. Someone who skipped enrollment because the explanation worried them can tell you something the satisfaction responses from enrolled users miss. Let people answer privately and distinguish objections to biometric collection from problems using the terminal. You may improve a confusing screen, but you shouldn't treat an informed refusal as a problem to overcome. The next decision needs both experiences, including the people who preferred to show their ticket.

Complete the agreed deletion and confirm what remains for separate purposes. Record who checked the supplier's completion evidence and who handles any outstanding requests. Don't leave that job until the next event starts planning, when everyone has forgotten which terminals and temporary files were used.

For your next paid conference, make ordinary ticket admission your starting option. Revisit facial recognition when you have a specific unresolved problem, a willing audience, a workable legal basis and a tested service improvement that justifies the cost. Ask the supplier to demonstrate that complete proposal before you approve a rollout. Your attendees should arrive knowing their choices, and your team should be able to welcome them whichever choice they make.

Loopyah Icon

Author: By the Loopyah Content Team

The Loopyah Content Team shares expert insights, practical guides, and industry updates to help event organizers create unforgettable experiences and stay ahead in the event planning world.