Almost every institution wants to try before deciding. Few define what trying means, which is why so many pilots end without a conclusion, the equipment goes back in its box, and nobody can say why.
A pilot is not "having the headsets here for a while". It is a question with a date for its answer.
Before you start: write the question down
A pilot has to be able to fail. If there is no result that would make the institution say "we are not going ahead", it is not a pilot: it is an extended demonstration.
Write one sentence, and write it before the equipment arrives. For example: "After eight weeks, we want to know whether the team can fit sessions into the routine without costing them time they do not have."
Notice what that sentence does not say. It does not say "we want to prove it reduces anxiety". An eight-week pilot in one service has no design, no size and no comparison group to answer that, and promising it does is the fastest route to a report nobody can use.
What a pilot really can answer
These are implementation questions, and a pilot answers them well:
- Does the team use it? The number one question, and the one that fails most pilots.
- How long does it actually take to set up and run a session?
- Where does it jam? Cleaning, charging, Wi-Fi, who fetches the headset, where it sleeps.
- Do patients accept it? How many decline, how many stop midway, and why.
- Who takes ownership? There is almost always one person who picks it up. Or there is not, and that is the answer.
What a pilot cannot answer
Worth saying up front, so there is no disappointment at the end:
- Clinical efficacy. Without a comparison group, without randomisation and with dozens of sessions, there is no efficacy conclusion to draw. That is what trials are, and they are a different thing.
- Real financial return. Eight weeks do not show annual savings.
- Long-term effect. By definition.
Be wary of a supplier who proposes a pilot with clinical outcome objectives. Either they do not understand study design, or they do and are counting on you not to.
A design that works
Duration: six to eight weeks. Less is not enough for the novelty to wear off; more and the pilot dissolves into routine without ever being assessed.
Scope: one service, not three. A pilot spread across three wards produces three half-experiences and no conclusion.
People: names, not "the team". Two or three designated people, trained before it starts. Equipment that is everyone's responsibility ends up in a cupboard: it is the most predictable pattern in this field.
A decision point in the calendar, dated, before the pilot begins. Without that date the pilot does not end: it dissolves.
What to measure, and how
Few things, and simple ones:
- Sessions per week. The curve says almost everything. If it rises and levels off, it entered the routine. If it rises and falls, it did not.
- Who ran them. If it is always the same person, the service did not adopt it: one person did.
- Time per session, including setup and cleaning. That is the number that decides whether it is sustainable.
- Refusals and interruptions, with the reason in two words.
- One sentence from whoever ran it, at the end of each week. Worth more than any dashboard.
RVer records sessions and how each one went automatically, and the Companion app lets one person oversee several headsets, which reduces the effort of collecting points 1 and 3. The rest is conversation with the team.
Before signing: the questions that remain
At the end, three questions settle the decision:
- Did the session curve level off at a rate that makes sense for the service?
- Did more than one person run sessions comfortably?
- Is what jammed solvable, or structural?
If all three answers are good, the business case makes itself. If they are not, the pilot saved you a purchase, which is exactly what it is for.
How we run ours
The device arrives configured and training is included, the system runs without internet during a session, and sessions are recorded for the team to interpret. We do not promise a clinical outcome from a pilot, and if anyone on our side does, tell us.
On who can run the sessions during a pilot, we wrote separately in who can run a session. On what to ask before that, in how to write a tender specification.
Frequently asked questions
How long should a virtual reality pilot last?
Six to eight weeks, in a single service. Less is not enough for the novelty to wear off and to see whether sessions really enter the routine; more and the pilot dissolves into daily work without ever being assessed. The essential part is putting the decision date in the calendar before it starts.
What should you measure in a virtual reality pilot?
Sessions per week (the curve says almost everything), who ran them, time per session including setup and cleaning, refusals and interruptions with the reason, and one sentence from whoever ran it at the end of each week. A few simple things beat a complex dashboard nobody fills in.
Does a pilot prove that virtual reality works?
No, and it is important not to promise that. Without a comparison group, without randomisation and with a few dozen sessions, there is no clinical efficacy conclusion to draw: that is what clinical trials are, and they are a different thing. A pilot answers implementation questions: does the team use it, how long does it take, where does it jam, do patients accept it.
What is the most common mistake in a pilot?
Not designating named people. Equipment that is everyone's responsibility is nobody's and ends up in a cupboard: the most predictable pattern in this field. The second mistake is spreading the pilot across several services, which produces several half-experiences and no conclusion.