The questions we least want to be asked.

“Which of your published results are measured, and which are modelled? Show me the test report for each.”

What a good answer looks like: Most wireless-power claims blend the two. A vendor who can separate them on the spot has an engineering discipline you can audit.

“What hardware revision produced those results, and is it the revision you would ship to us?”

What a good answer looks like: Results from a lab revision that never shipped are marketing, not evidence.

“Under what conditions were peak efficiency figures obtained — distance, alignment, load, temperature?”

What a good answer looks like: Peak numbers without conditions are unusable. Ask for the curve, not the maximum.

“What is the efficiency at the operating point we actually need, not at your best point?”

What a good answer looks like: Your duty cycle is the only one that matters.

“Show me a result where the system underperformed and what you changed.”

What a good answer looks like: A vendor with no failure data has either not tested enough or will not show you.

“When a device requests power, what exactly authenticates it?”

What a good answer looks like: If the answer is “it just receives power”, there is no security model to review.

“What happens if authentication fails mid-session?”

What a good answer looks like: Graceful, logged termination is the only acceptable answer.

“Where do device credentials live, how are they provisioned, and how are they revoked?”

What a good answer looks like: Revocation is the part most vendors have not built yet.

“Can an unauthorised device draw energy within range? Demonstrate the attempt.”

What a good answer looks like: Ask to watch the attempt fail, not to be told it would.

“What is the threat model for the energy link itself — jamming, spoofing, replay, denial of service?”

What a good answer looks like: A written threat model should already exist.

“Has any third party assessed the security of the system? Can we see the report and the remediation status?”

What a good answer looks like: Unremediated findings are more informative than the findings themselves.

“Show me a signed telemetry record for a single energy session, and tell me who can and cannot alter it.”

What a good answer looks like: If energy is metered, billed or audited, the record must be defensible to someone who was not in the room.

“What exactly is recorded per session — identity, energy delivered, duration, conditions, outcome?”

What a good answer looks like: Missing fields become disputes at invoice time.

“What is the measurement accuracy of the metered energy figure, and how is it calibrated?”

What a good answer looks like: Billing on an uncalibrated meter is a commercial risk.

“Where is telemetry stored, for how long, and under whose control?”

What a good answer looks like: Data residency and retention belong in the contract, not the roadmap.

“Can we export the full record set in an open format without your involvement?”

What a good answer looks like: Vendor-locked audit trails fail audits.

“What regulatory approvals does the system hold, in which jurisdictions, and for which power levels?”

What a good answer looks like: Approvals are power-level and geography specific.

“What are the human-exposure limits, and how does the system enforce them?”

What a good answer looks like: Enforcement mechanism, not compliance statement.

“What is the thermal behaviour at sustained maximum duty, and what throttles it?”

What a good answer looks like: Sustained duty is where demos end and infrastructure begins.

“How does the system behave with metal, people, or moving obstacles in the field?”

What a good answer looks like: Real environments are cluttered.

“What interference does the system emit, and what interference degrades it?”

What a good answer looks like: Your RF environment already has occupants.

“What is the failure rate and mean time to repair of the deployed fleet to date?”

What a good answer looks like: If there is no fleet, say so — that is a different risk conversation.

“Who installs, who maintains, and what is the support SLA in our jurisdiction?”

What a good answer looks like: Support gaps kill pilots after they succeed.

“What is the total cost of ownership at our measured efficiency, including energy losses?”

What a good answer looks like: Model this from measured data during the evaluation, not from marketed figures.

“If the pilot succeeds, what are the commercial terms — and can they be agreed now, in week one?”

What a good answer looks like: Terms negotiated after a success take longer than the pilot did.