Independent pilot measurement · Warehouse robotics
The vendor says it hit throughput. The floor manager says it kept stalling near packing. Nobody wrote down what success meant, so the rollout quietly doesn't happen. We measure the pilot against numbers both sides agree on before it starts.
Illustration of report format. Not customer data.
The gap
Every party involved is acting in good faith. The evidence still doesn't settle anything.
Throughput, uptime and intervention counts come out of the vendor's fleet software. The data is usually accurate. It just isn't independent, and the procurement team knows it.
Criteria get written after the pilot has already started, or not at all. Two honest people read the same three months and reach opposite conclusions, with no way to resolve it.
Safety, legal, finance and insurance arrive at the end and ask for evidence nobody was collecting. The rollout slips a quarter while the vendor tries to reconstruct it.
The verified pilot
The sequence matters: the threshold has to be agreed while neither side knows the answer.
Before the robots arrive, the vendor and the operator agree in writing on the metrics, the measurement window, the exclusions, and the number that means "roll out." Both sign it. This is the whole trick — after the pilot begins, every threshold becomes a negotiation.
We connect to the vendor's fleet management system and the site's WMS, and normalize both into one record. Scope stays narrow on purpose: only the fields the agreed metrics require. No blanket telemetry request, no data you can't explain to your own security team.
Continuous measurement across the pilot window, with a reading issued every two weeks to both parties at once. Problems surface while there's still time to fix them, and nobody is ambushed on the final day.
One document: what was claimed, what was measured, the variance, and whether the agreed threshold was met. It goes to the vendor and the operator on the same day, in the same form. Written so a finance team can put it in a model.
What we measure
Each engagement uses a subset. Definitions and exclusions are fixed in the protocol document, not chosen afterwards.
| Metric | How it is read | Source |
|---|---|---|
| Task throughput | Completed tasks per operating hour across the agreed window, with agreed exclusions applied | Fleet software + WMS |
| Uptime | Operating hours against scheduled hours, charging and planned maintenance treated as defined | Fleet software |
| Intervention rate | Remote and on-floor human interventions per 100 operating hours | Fleet logs + site record |
| Task success rate | Tasks completed against tasks attempted, including retries and abandonments | WMS |
| Safety events | Emergency stops, protective-stop activations, near misses and contact events | Fleet logs + site incident record |
| Recovery time | Mean time from fault to resumed operation, split by fault class | Fleet software |
Neutrality
This is the first objection anyone sensible raises, so we designed the answer into the engagement rather than into the marketing.
We don't set the thresholds. The two parties set them together, in advance, before anyone knows which way the result will go. We're the instrument, not the judge.
The document states what was claimed and what was measured. It contains no opinion on whether to buy. The decision stays where it belongs.
Both parties receive an identical report simultaneously. There is no vendor preview, no chance to shape the framing before the operator sees it.
We don't resell hardware, take referral fees, or hold equity in a robotics vendor. Our fee is the same whether the pilot passes or fails.
Who this is for
You already know your system works. The problem is proving it to a buyer who has been pitched the same numbers by four of your competitors, and who has no way to tell any of you apart.
A measured pilot turns your claim into a fact the buyer's finance team can put in a model — and gives you a rollout decision on a date, instead of an evaluation that stays open until the champion changes jobs.
You're being asked to commit floor space, staff time and a multi-year contract on the basis of numbers you have no way to check, produced by the company selling you the robots.
Ask your vendor to run the pilot under independent measurement. It costs them very little. If the system performs the way they say it does, they'll agree — and how they answer tells you something either way.
Where we are
We're early and would rather say so than pretend otherwise. We're looking for three warehouse robotics vendors with a pilot starting in the next quarter. Engagements are paid, deliberately small, and shaped together with the partner. If you've had a pilot stall and you still don't know exactly why, that's the conversation we want to have.
hello@fixrobotic.info