Service Robot Fleets in ESG Reporting, Scope, Method and the Numbers That Hold
At a glance: A robot fleet entering a facilities ESG disclosure raises a boundary question before it raises a data question, and most vendors answer neither. This guide sets out which scope a fleet's electricity belongs in for owned, leased and RaaS fleets, gives the four quantities you can actually report with confidence, and lists the claims a reviewer will strike.
The request rarely comes from the facilities team that bought the robots. It arrives from finance or from the sustainability function, typically in the run-up to a reporting deadline, as a one-line question: "How do the robots show up in this?" By then the fleet has been running for a year with telemetry nobody exported, and the answer has to be assembled backwards from incomplete records.
This guide is written for the person on the receiving end of that question. It covers the boundary decision that determines whether a fleet number is reportable at all, the four quantities that can be defended with the data a standard fleet already logs, the method that survives review, and the specific claims that should be struck from a draft before a reviewer does it for you.
Why Your Robot Fleet Landed in the ESG Report
A robot fleet is a facilities asset that consumes electricity, consumes water, displaces labor and generates equipment waste at end of life. Under any dual-materiality reporting framework, each of those is potentially in scope, and the facilities function is where the data lives. That is why the question lands with the facilities team and not with the vendor, regardless of what the vendor's sustainability page says.
The framing matters, because it determines what the report needs. A disclosure is not a marketing document. It is a claim set that a reviewer will test against evidence, and the test is not "is the claim attractive" but "can the boundary be stated, the method reproduced, and the number traced to a source." Three of the four quantities below will pass that test for most fleets. The fourth depends on a decision made before deployment, which is the reason this guide starts with boundaries rather than numbers.
Scope 1, 2 or 3? Getting the Boundary Right
Before any number, decide whose boundary the fleet sits inside. The answer differs by commercial structure, and getting it wrong means either double-counting the electricity or omitting it entirely.
| Fleet Structure | Where Fleet Electricity Sits | Who Reports It | The Practical Trap |
|---|---|---|---|
| Purchased (owned asset) | Scope 2, operational control: the reporting entity buys and consumes the electricity | The facilities operator, under its own scope 2 | Sub-metering is often absent, so the figure has to be apportioned from a building total |
| Leased (asset under finance or operating lease) | Follows the operational-control or financial-control test, not the financing label | Whoever has operational control of the site | Two entities each assuming the other reports it, so nobody does |
| RaaS (robots-as-a-service) | Typically scope 2 at the site under operational control, with the service fee itself in scope 3 purchased services | Site operator for electricity; service provider for its own upstream footprint | The service provider's own disclosure and the client's can overlap on the same kilowatt-hour |
The rule that resolves all three cases is operational control. The entity that decides when the machine runs and where the electricity comes from is the entity that reports it. Where a service provider includes its fleet's energy in its own disclosure, that inclusion is about the provider's upstream footprint and does not remove the kilowatt-hour from the client's scope 2 if the client controls the site. State the treatment explicitly in the report, with the reasoning, and the reviewer's question is answered before it is asked.
The Four Quantities You Can Actually Report
Four quantities are defensible from data a standard autonomous fleet already logs. Each has a data source and a stated confidence level, and none of them requires a dedicated sustainability platform.
| Quantity | Unit | Data Source | Confidence Note |
|---|---|---|---|
| Fleet electricity consumption | kWh per robot-year | Charging-point sub-meter, or accumulated charge-cycle data from fleet telemetry | High if sub-metered; medium if apportioned from a building total, and the method must be stated |
| Water consumed | Litres per 1,000 m² cleaned | Tank fills logged against area coverage | High; this is directly observable from the machine's own duty cycle |
| Chemical concentrate used | Litres per year | Issued concentrate volume against metered dosing ratio | High where dosing is metered; low where operators dose by hand |
| Service hours displaced | Hours per year | Task logs against the pre-deployment labor baseline | Medium; displacement is a management decision as much as a measured one, and should be stated as such |
The electricity figure is the one most reports get wrong, and the error is almost always the same: a single robot's rated power multiplied by a nominal 24-hour duty cycle. A service robot does not draw its rated power continuously. It draws a working load during tasks, a lower navigation load in transit, a charging load at the dock and a small standby load when idle, and those four states are not interchangeable. Accumulated charge-cycle data gives the real figure; the rated-power estimate typically overstates it by a wide margin. If you are building a first-year number, use metered charge data, and where none exists, say so rather than substituting an estimate. The detailed breakdown of draw by state is covered in service robot electricity consumption.
Method: Building a Defensible Robot-Fleet Line
The method is four steps, and the ordering is what makes it defensible. Each step maps to something a reviewer can check.
- Baseline before deployment. Record the pre-robot condition on the same boundary you will report afterwards: the area cleaned, the water and chemical consumed by the manual method over a representative window, and the labor hours applied to it. A baseline captured after installation is a reconstruction, and should be labelled as one.
- Meter after deployment, on the same boundary. Same areas, same frequency, same accounting period. If the boundary shifted, the comparison is between two different things and should be presented as two figures rather than one change.
- Keep the units consistent. Per 1,000 m² for water and chemical, per robot-year for electricity, per year for displaced hours. Unit consistency is what makes the line extendable across sites.
- State what you are not counting. Manufacturing footprint, end-of-life equipment, and the embodied carbon of the fleet are all potentially material and all outside a facilities scope 2. Naming them as excluded is stronger than silently omitting them.
The fourth step is the one that distinguishes a disclosure from a claim. A reviewer who finds an unstated exclusion will treat it as concealment; a reviewer who sees it named and reasoned will treat it as a boundary decision, which is what it is.
What a Reviewer Will Ask For
Most ESG questions about robots reduce to a small number of concrete requests. Knowing them in advance is most of the preparation.
The first is a source. For electricity, that means a sub-meter reading or an exportable charge-cycle dataset, not a datasheet figure. For water and chemical, it means consumption against area, not a tank capacity. The second is a boundary statement, in one or two sentences, naming operational control and stating which scopes are affected. The third is a method note, short enough to sit in an appendix, describing the baseline window and the measurement period. The fourth is an exclusion list, naming what was left out and why.
The claims to strike are equally predictable. "Zero emissions" applied to a machine that consumes grid electricity is the most common and the most quickly challenged. "Reduces carbon footprint" presented without a boundary or a number is unverifiable in the direction that matters. "Displaces X percent of labor" stated as a vendor-published average, rather than as the site's own measured change, will not survive contact with the client's own payroll data. And "sustainable materials" applied to a plastic-housed machine without a life-cycle figure is a phrase a reviewer will simply delete. None of those claims is necessary. Each of the four quantities above states more, with evidence attached.
Where Robots Genuinely Help, and Where They Don't
Honesty here is not only ethical, it is strategic, because the areas where a fleet genuinely moves a facilities number are narrower and more specific than a vendor page usually admits.
The largest and most reliable effect is on water and chemical consumption per unit of area. A metered scrubber with recovery applies a controlled volume and lifts most of it back off the floor, where the manual method it replaced applied water freely, recovered none of it, and often needed two or three passes for the same result. That is a real, metres-per-litre reduction and it is defensible from the machine's own duty cycle, as the arithmetic in the water and detergent cost model shows. On large hard-floor sites this is the strongest environmental case a robot fleet has, and it does not require any carbon argument at all.
The smallest effect is on headline carbon, and this is where most vendor claims fail. If the local grid is already low-carbon, the electricity a fleet consumes carries a small emissions factor, so the fleet's scope 2 addition is small in absolute terms. Against that, the displacement of the manual method removes the water heating, the chemical manufacturing and the transport trips that the old method required, which are frequently larger than the fleet's own electricity. The net is often favourable, and it is always specific to the site. It is not favourable because robots are efficient in the abstract. It is favourable because the method they replace was less efficient, and that comparison has to be made in numbers on the actual site.
The dimension that is hardest to report and easiest to overstate is labor. Displaced hours are real, but they are also a management decision: hours removed from a schedule only become a resource reduction if the schedule is genuinely re-planned rather than absorbed. Report the measured change in hours applied to the task, state it as an operational metric, and avoid converting it into a carbon or a cost claim without the intermediate arithmetic. The measurement framework for those hours is set out in fleet KPI benchmarks, and the wider fleet-management picture, including which telemetry is exportable and in what format, is covered in service robot fleet management.
Starting the Robot Line in Your Next Report
For a fleet already in service, the fastest route to a defensible line is to reconstruct the baseline from existing records and begin structured metering now, reporting the first full year as the comparison and labelling the historical baseline as reconstructed. For a fleet being specified, the sequence is different and much easier: decide the boundary, install charging-point sub-metering, capture the pre-deployment baseline on the same boundary, and record the dosing ratio at commissioning.
AOMAN platforms export task logs, coverage records and charge-cycle data in a form that maps onto the four quantities above without a separate reporting layer, and the D1 delivery and C1/C2 Pro cleaning platforms log water consumption against area covered as a matter of routine operation. That data is the difference between a number that survives review and a claim that gets struck. If you are specifying a fleet and want the reporting fields agreed before deployment rather than reconstructed after it, request the telemetry and reporting schema and we will send the export format alongside a sample period. For the budgeting side of the same programme, see service robot budget planning.
