Problem
At an industrial site, diesel leaves the tank through loss, theft and spillage that nobody can see, so nobody can price it. The plant knew its monthly consumption; it did not know how much of it was waste.
Evidence
Site walks at the plant, with the people who dispense fuel as well as the people who buy it. The gaps were at the nozzle and the tank, so the product had to live there: automatic tank gauges, dispensers, flow meters, RFID nozzle tags and tank sensors, 100+ requirements in all.
Decision
A six-month pilot measured against the site's own baseline, with the saving as the metric. The buyer could verify the saving at the meter, which is what makes a hardware-linked product sellable.
Result
Savings did not arrive in month one. As the controls and site behaviour settled, the six-month pilot cut loss, theft and spillage by about 20 paise per litre against the site's own baseline, a run-rate of ~₹4 lakh a month by months 5–6.
What I'd do differently
Set the expectation of a ramp in the contract. A saving that climbs over six months reads as a miss in month two unless the buyer was shown the curve first.
Related work
More case studies
Commercial terms stay private. Figures are rounded.