Carbon accounting software still needs clean utility data
Carbon accounting platforms apply emission factors well. They do not fix bad inputs. Here is why activity data quality, not the software, decides whether your emissions number holds up.
Carbon accounting software has become very good at one job: taking a number for how much energy or fuel you used, multiplying it by the right emission factor, and rolling the result up into scopes and categories. That last part, the factors and the math, is largely solved. The EPA's GHG Emission Factors Hub publishes a consolidated, regularly updated set of default factors, and every serious platform builds them in. So if the factors are handled, why do so many emissions numbers still fall apart under scrutiny?
Because the factor is the easy half. The hard half is the activity data: the actual kilowatt-hours, cubic meters and liters that go in front of the factor. A greenhouse gas inventory converts activity data into emissions by applying factors, which means a wrong or missing input produces a wrong output no matter how good the software is.
Garbage in, certified garbage out
Emission factors are a downstream step. The EPA Emission Factors Hub consolidates US sources such as eGRID into one place, and platforms apply them automatically. But the factor cannot tell whether the 12,000 kWh you entered was a real meter read or an estimate, whether it double counts a submeter, or whether it belongs to the site you assigned it to. The software trusts the input. If the input is wrong, you now have a precise, well-formatted, wrong answer, which is harder to catch than an obviously rough one.
Where activity data goes wrong
- Estimated meter reads that were never trued up against an actual read
- Unit and conversion errors (therms vs cubic meters, kWh vs kW)
- Bills mapped to the wrong site, meter or account
- Gaps and overlaps when a billing period does not line up with the reporting period
- Double counting where a submeter is added on top of a main meter
None of these are exotic. They are the everyday texture of utility data, and none of them are visible to an accounting tool that receives a single tidy number per meter per month.
Standards judge your inputs, not just your factors
This is not a stylistic concern. The frameworks companies report against grade the quality of the underlying data. PCAF's financed-emissions method uses a data quality scale from 1 to 5, where using real physical activity data scores better than inferring emissions from spend or sector averages. IFRS S2 requires emissions measured in accordance with the GHG Protocol, and assurance providers checking those figures trace them back to source records. When an auditor asks where a number came from, the answer has to be a defensible activity record, not a spreadsheet cell.
Verified meter data beats an estimated read, an estimated read beats a spend-based estimate, and a spend-based estimate beats a sector average. Better inputs move you up every framework's quality scale.
What carbon accounting software is not built to do
Most carbon platforms assume the activity data arriving at their door is already clean. They are designed to categorize, apply factors and report, not to read a stack of PDF bills, reconcile estimated reads, catch a unit error, or decide which meter a charge belongs to. That data-preparation layer is a separate discipline, and it is where the effort and the errors actually concentrate.
Where MartinAI fits
MartinAI sits upstream of your carbon accounting software. It reads utility bills, meter data and Green Button feeds, structures every field, and validates the result: reconciling usage against reads and billing dates, catching unit and conversion errors, flagging estimates, and tying each record to the correct site and meter. What your accounting platform then receives is clean, validated, analysis-ready activity data. The software does what it is good at, and the number it produces stands up when someone traces it back. For a fuller walkthrough of preparing data for review, see prepare utility data for an ESG audit.
The point is not that carbon accounting software is unnecessary. It is essential. But it is a reporting engine, not a data-cleaning engine, and treating it as both is how organizations end up certifying numbers they cannot defend.
Frequently asked questions
Does carbon accounting software handle data quality?
Carbon accounting software is built to apply emission factors and roll results into scopes and categories, and it does that well. It generally assumes the activity data it receives is already clean. Reading messy bills, reconciling estimated reads and catching unit or mapping errors is a separate data-preparation step it is not designed to perform.
What is activity data in carbon accounting?
Activity data is the physical measure of what you consumed, such as kilowatt-hours of electricity, cubic meters of gas or liters of fuel. Emission factors convert that activity data into emissions, so the accuracy of the final number depends directly on the quality of the activity data.
Why do emission factors not fix bad inputs?
An emission factor multiplies whatever activity number you give it. If that number is an estimate, double counts a submeter, or is assigned to the wrong site, the factor produces a precise but wrong result. Factors are a downstream step and cannot detect an upstream input error.
How do disclosure standards treat data quality?
PCAF uses a 1 to 5 data quality scale that rewards real activity data over spend-based estimates, and IFRS S2 requires emissions measured against the GHG Protocol, with assurance providers tracing figures back to source records. Better inputs improve both your score and your ability to pass assurance.
Financed emissions (Scope 3 Category 15), explained
For banks and investors, Category 15 usually dwarfs every other emission source. Here is how financed emissions work, how PCAF attribution is calculated, and why data quality decides the number.
GHG inventory boundaries: operational vs financial control
Before you count a single tonne, you have to decide what counts as yours. This is how the GHG Protocol organizational boundary works, and how the control approaches change your number.