Why XR pilots stall — and how to design them for production

The gap between an XR pilot and an XR program is wider than most enterprises expect. We see the same pattern across Innovation teams at pharma, energy, financial services, and higher education: a celebrated pilot, six months of momentum loss, then a quiet decommissioning. The platform vendor blames the customer. The customer blames the platform. Both are partially right — but the real problem is upstream, in how the pilot was scoped.

Here are the five reasons XR pilots stall before production, and a framework for designing pilots that don't.

1. Headset dependency

Most XR pilots happen on Meta Quest. The team buys ten headsets, runs three sessions, generates a buzz, and writes the report. Then someone in Procurement asks how the pilot scales to 5,000 employees in 23 countries — and the math falls apart.

Headset-only XR forces every production rollout to start with hardware logistics: purchase, distribution, sanitization, IT support, breakage, charging, OS updates. The pilot's success metrics didn't account for any of that.

The fix: design every pilot for browser-first delivery from day one. A successful pilot in the browser proves you can scale to anyone with a laptop. A successful pilot in headsets proves nothing about how production will run.

2. Rebuilding for production

Pilot vendors often deliver custom Unity or Unreal builds. They look great. They demo well. And they require the same vendor — at the same hourly rate — to make any change in production.

Six months into a successful pilot, the customer wants to update the training script. The vendor quotes a six-week change order. The customer's Innovation team realizes the pilot wasn't a platform; it was a one-off.

The fix: pilot on a platform you'd actually run production on. If your team can't change the content without a developer, you're not piloting a platform — you're prototyping a custom build.

3. No analytics story

Most pilots end with a survey. Twenty-three out of thirty respondents agreed it was “engaging.” Procurement is unmoved.

Production XR needs a quantitative case for renewal — completion rates, time-to-competency, cost-per-trained-employee, retention curves at 30/60/90 days. Pilots that don't capture these metrics from day one can't make the production case at quarter-end.

The fix: define the three numbers your CFO needs to see, then make sure your pilot platform captures them automatically. If analytics are a “phase two” item, your pilot is unfundable past the first cycle.

4. No localization path

Enterprise XR is global by default. A pilot run in English with US-based employees doesn't tell you whether the platform will work for your team in São Paulo, your team in Mexico City, or your team in Munich.

Most pilots discover this in production: the platform doesn't support voice translation, the chat is English-only, the captions are auto-generated and unreliable. The customer ends up running parallel pilots in each region, doubling cost and halving cohesion.

The fix: scope the pilot for at least one non-English language from the start. If the platform can't handle Spanish or Portuguese natively, it can't handle global rollout, period.

5. No governance path

The Innovation team runs the pilot. IT, Security, Legal, and Procurement get involved at production. Each finds a reason to pause: data residency questions, SSO integration, accessibility compliance, custom SLA requirements.

Pilots that haven't anticipated these gates lose six to twelve weeks at the production transition — often enough time for the executive sponsor to lose interest.

The fix: invite IT, Security, and Procurement into the pilot phase, even if their formal sign-off isn't required. The questions they'll ask at production are predictable. Address them in the pilot scope.

The framework

Before you scope another XR pilot, answer these three questions:

1. Will the pilot run on the same platform that production will? If no, the pilot is teaching you about a tool you won't keep.

2. Will the pilot capture the metrics that production will be judged on? If no, the pilot can't fund itself past the first cycle.

3. Will the pilot include the constraints that production will face? Languages, devices, security review, accessibility, IT integration. If any of these are punted to production, you've built a pilot that can't graduate.

Most pilots fail these three tests. The ones that pass become programs.

If you're scoping an XR pilot now, we'd love to compare notes. We've shipped enough of these to know which questions matter — and which look important but aren't.