Waste Services — Germany

Why we anonymise

Customer confidentiality is part of our professional standard.

We value the confidentiality of our customers, partners and data owners. Publicly presented test cases are therefore anonymised and do not disclose vehicle identities, vessel names, fleet details or commercially sensitive operating data unless explicit public disclosure has been agreed.

Our principle

Proof should come from validation, not from borrowed credibility.

Our technology should be judged by a transparent test approach and by each customer’s ability to verify NanoEFX under their own operating conditions — not by a wall of past customer logos.

Known limitations

• There is no control group. Untreated sister vehicles running the same routes over the same weeks would separate the treatment effect from the season and from the load shift in one step — this is the single biggest improvement available for this trial, and we are asking Waste Services whether that data can be pulled retrospectively.
• The vehicles ran lighter during the trial. Average capacity utilisation across the fleet fell from 0.87 to 0.78 — down 10.5% — and distance per operating day fell 8.4%. A lighter refuse truck burns less fuel per 100 km regardless of any treatment, so some unknown part of the 18.2% is a load effect rather than a product effect. This is the single most important caveat on this page, and we would not know about it if Waste Services had sent only the fuel figures.
• Against that: the load shift does not track the result. The vehicle whose load fell furthest (Vehicle 02, −19.9%) recorded the smallest saving (−7.9%), the one with the largest saving sits mid-range, and the correlation across the five vehicles is +0.15 — effectively none. If the saving were mostly a load artefact, that correlation should be strongly negative. With five vehicles this is an argument, not a proof.
• The baseline is seasonal. March–May is compared with a 15 June to 24 July window, and summer diesel, fewer cold starts, no cabin heating and drier roads all reduce consumption on their own. Some part of the 18.2% is very likely seasonal rather than product effect.
• The two datasets do not agree week by week. Their totals match to within 0.1%, but individual weeks diverge sharply at three of the five vehicles — Vehicle 04 shows 273 km in the operational records against 823 km in the fuel workbook for week 2, Vehicle 03 shows 1,185 against 878 for week 4. The week definitions in the fuel workbook are also inconsistent (week 1 runs Monday to Friday, week 2 Saturday to Friday, and 27–28 June falls out entirely). We have asked Waste Services to reconcile them.
• The operational records are incomplete in places: the June sheet carries no destination column at all, so route continuity could only be checked for July, and 24 days in July have no distance recorded.
• Weather, traffic and vehicle maintenance status were not systematically controlled during this proof-of-concept phase.
• One vehicle (Vehicle 01) was out of service for repairs during week 6 and is excluded from that week’s calculation.
• The spread between vehicles is wide — from 7.9% to 33.8%. Route changes do not explain it: Vehicle 03, the −33.8% vehicle, has the highest route continuity of all five at 90%. It remains an outlier to be explained rather than a headline.
• This is proof-of-concept data. The trial report itself states that it should be treated as a basis for further controlled validation, not as certified documentation.