Custom CO2 accounting software development
Carbon accounting is a ledger for emissions. Consumption data from invoices, meters and systems goes through an emission factor and becomes a figure that someone steers by or reports on. That figure is only worth something if you can show which measurement it comes from, which factor was used and which version applied. Appfront builds software that keeps hold of all of that.
Calculation, not reporting
CO2 accounting, or carbon accounting, is the process beneath every sustainability figure. You collect what has been consumed, transported and purchased, convert it to CO2 equivalents using emission factors, and keep every result traceable to its source. The comparison with financial accounting goes beyond the name: there is a ledger, a close for each period, and someone who wants to follow a sample trail afterwards.
Are you in the right place? This page is about that calculation work. If your question is about the broader sustainability reporting and the obligation to do it, about materiality, data points and the sustainability statement, you belong with CSRD and ESG reporting software. They are different systems: one supplies the number, the other the accountability around it. If you start with the report and skip the accounting, you end up, at the first audit question, with a spreadsheet nobody can reproduce.
Consumption is the input, not the emissions
Your sources supply litres, cubic metres, kilowatt-hours and kilometres. The emissions are a derived figure that you calculate.
The factor determines the result
The same litre of diesel gives a different figure with a different factor version. Without a recorded version, a figure cannot be reproduced.
Every figure must trace back to its source
From a line in the report to the invoice or meter reading underneath, without a loose spreadsheet in between.
The plumbing is the largest part of the build
Anyone building a CO2 accounting system for the first time tends to think of the calculation core: consumption times factor. That core is the simplest part of the whole system. The work is in the data feed. Your gas consumption is on invoices from one or more suppliers, your electricity partly on invoices and partly on meters, your fuel on fuel card statements, your business mileage in a trip log or an expenses system, your purchased goods in the ERP or procurement package, and your refrigerants in an installer's maintenance report. Six sources, six formats, six owners within the organisation.
Each of those sources has its own period. An energy invoice runs from meter reading to meter reading and so rarely lines up with your financial year. A fuel card statement runs per calendar month. An annual settlement corrects after the fact what the advance payments got wrong. If you add those series together without thinking, you get an annual total that spans eleven or thirteen months. A usable system therefore records a period with a start and an end for each line, allocates consumption pro rata to the reporting period when an invoice crosses the year boundary, and shows what part of the year is covered per source. If a month is missing, that is a visible gap and not a silent zero.
On top of that, every source has its own key. The energy supplier knows connections, the fuel card provider knows card numbers, the ERP knows cost centres, the property administration knows buildings. To attribute emissions to a site, a department or a project, those keys have to be linked somewhere, and that mapping table has to keep up when a card changes hands or a building is sold. If you don't set that up explicitly, you only notice the problem when someone asks why the South site uses more diesel than there are cars on the road.
Finally, the input must be verifiable. Consumption that suddenly doubles, a meter reading lower than the previous one, an invoice that has been imported twice, a unit that jumps from cubic metres to kilowatt-hours: these are signals the system itself should pick up, with a work list for whoever has to resolve them. Validation at the gate saves a later correction that has to ripple through the whole chain of calculations.
The three scopes, and why scope 3 is the hard part
The division into scopes comes from the Greenhouse Gas Protocol, from the World Resources Institute and the World Business Council for Sustainable Development. Its Corporate Accounting and Reporting Standard is the basis on which almost every other scheme builds. Scope 1 is direct emissions from sources you own or control: your gas boiler, your vehicle fleet, refrigerant leakage. Scope 2 is indirect emissions from purchased energy: electricity, heat, cooling and steam. Scope 3 is all other indirect emissions in the value chain, divided into fifteen categories in the Corporate Value Chain (Scope 3) Standard.
You can, in principle, calculate scope 1 and 2 from data that is already in your own records. Scope 2 has a catch, though: the GHG Protocol requires two calculations side by side, location-based (using the average grid mix) and market-based (using what you actually purchased). Your data model therefore needs to carry two answers per kilowatt-hour, not one.
Scope 3 is a different kind of problem. Here your figure depends on other people's data: suppliers, carriers, processors, sometimes customers. You buy a product, not a dataset. Few suppliers have a figure ready that is tailored to your purchases; what they send is an average per product group. The rest you estimate: on spend per purchasing category, on weight and distance for transport, on an industry average. Those estimates are legitimate, but they must remain recognisable as estimates, even when the figure ends up in a dashboard three systems further down. In transport-intensive chains, the input comes from the trip and shipment records.
Which entities and locations count
Before you can add anything up, you need to establish what you are adding up. This is called the organisational boundary, and the GHG Protocol devotes a separate chapter to it in the Corporate Accounting and Reporting Standard, titled Setting organizational boundaries. There are two routes. With the equity share approach, you count the emissions of an investee in proportion to your stake. With the control approach, you count one hundred per cent of what you control, where control can be financial or operational. Both are permitted, but they give different answers, and the answer changes again if you choose one route one year and the other the next.
In practice, the pain is in the edge cases. A rented building where the landlord buys the energy and passes on the cost. A joint venture in which you hold half but do not make the decisions. Lease cars registered in the name of the leasing company but driven by your people. An operating company acquired halfway through the year. For each case there is a defensible choice, but the choice must be documented and repeatable, because next year someone else may be doing the calculation.
For the software, this means the entity structure is not a dropdown but a model with validity over time. Every entity, site and connection has a period in which it belonged to you, and a basis on which it counts. An acquisition halfway through the year then automatically produces a part-year instead of a discussion. And because certification and reporting schemes may apply their own boundaries, the same base data must be able to serve multiple boundaries without you entering the figures twice. The CO2 Performance Ladder, for example, already counts business travel, a scope 3 category in the GHG Protocol, in the emissions inventory from the lower levels onwards.
Primary and secondary data in the supply chain
For scope 3, there is a clear hierarchy of data, and it is spelled out in the schemes. Handbook 3.1 of the CO2 Performance Ladder asks for emissions data that is as specific as possible and substantiated, so that the assumptions, sources and system boundaries used are clear. Product data preferably comes from studies in accordance with ISO 14067, on the carbon footprint of products, or in accordance with the GHG Protocol Product Life Cycle Accounting and Reporting Standard. If these are not available, you use data from your supplying or purchasing party, provided it is demonstrably representative and backed by underlying studies or calculations. If that is also missing, you fall back on the most specific factors available from the literature. For materials, the same handbook points to the Nationale Milieudatabase (Dutch national environmental database), or to data from an EPD or MRPI certificate.
That hierarchy isn't just theory: it determines what your data model looks like. Every value chain data line carries not only a value but also a level in that hierarchy, a source and a date. That lets you show later on that the share of primary data is growing, which is exactly what a client or certification body wants to see.
That leaves the question of how to get suppliers to provide it. Sending out a questionnaire rarely works: it lands with the wrong person, asks for data that doesn't exist and produces answers you can't verify. What does help is tying in with what is already happening. Ask at an existing point of contact, such as a contract renewal or a supplier review. Ask only for what you actually use in the calculation, and pre-fill the form with what you already know from the purchase order, so the supplier confirms rather than fills in. Reuse an answer for every line it applies to and only ask again once it has gone out of date. Feed back what you do with the figure, because a supplier who can see their own position in your overview will respond faster next time. And bear in mind that there is now a limit to what you can demand from smaller parties, as described below.
What the omnibus changed about who has to report
In just two years, the European reporting obligation has been narrowed substantially. Following the European Commission's omnibus simplification package, the reporting obligation has been narrowed to the largest companies, with an employee threshold combined with a turnover limit. That change affects the Corporate Sustainability Reporting Directive, Directive (EU) 2022/2464. Have your accountant confirm whether the obligation applies to you, because the thresholds and implementation dates have been adjusted repeatedly in recent years.
For the question of whether you need carbon accounting, that tells you less than it seems. The same directive also limits what reporting companies may ask of smaller parties in their value chain. Companies that don't reach an average of one thousand employees are protected undertakings: no more may be asked of them than the voluntary standard allows. So fewer parties publish their own figures, while the demand for emissions data continues to be passed down through contracts and tenders. That is why the focus is shifting from reporting to calculating.
What carbon accounting needs to be able to do
These components are interlinked. Version control without recalculation produces an unusable archive, and an audit trail without a data quality label shows where a figure comes from, but not how robust it is.
Input from invoices, meters and systems
Integrations with your energy supplier, meters, fleet management and procurement, with manual entry where there is no alternative.
Version control on emission factors
Every factor with its source, year and validity period. The calculation stores which version it used, not just the result.
Recalculation as a function
Recalculating a closed year with a different set of factors, alongside the original result, with the reason recorded.
Data quality per line
Measured, derived or estimated, with the method alongside, so a total never looks more certain than the underlying data allows.
Supplier data requests
A portal for value chain data, falling back on a calculated estimate as soon as a supplier provides nothing usable.
Audit trail to the source
From every line in the report back to the measurement underneath it, including who entered and approved what, and when.
What is deliberately missing here is the layer on top: dashboards, targets and a report in the form a reader expects. Those belong too, but they only make sense once the six components above are right. A chart that goes down while half of the underlying lines are estimated and nobody knows which factor version was used is not management information but decoration.
Where emission factors come from and why the version matters
For the Netherlands, the standard factors are published on co2emissiefactoren.nl, an initiative of SKAO, Stimular, Connekt, Milieu Centraal and the Dutch national government, managed by Rijkswaterstaat and Stichting Stimular. The figures draw on, among other sources, the latest IPCC report, research by CE Delft and data from CBS, and are updated annually around the turn of the year. RVO also publishes the Dutch list of energy carriers and standard CO2 emission factors.
That annual update is exactly why an emission factor is not a setting in a configuration screen. The advice for the Dutch list is to use the factor that belongs to the year you are calculating: a factor from January applies to that year's footprint, not to the year before. Only when new scientific insights show that a factor is no longer correct is retroactive application advised. In software, that means a factor table with validity periods, plus a calculation that records which row it used.
Internationally, it gets harder. The Dutch figures apply to the Dutch situation, specifically for electricity, natural gas, biogas and heat here; for sites abroad, your model has to record per measurement point which list the factor comes from. See also energy management software.
- Source, year and version stored with every calculated line
- Validity period per factor, not a single current value
- Location-based and market-based results side by side
- Explicit units, with conversion to CO2 equivalent
- Import a new annual list without overwriting earlier results
- Foreign factors clearly identifiable
Units, system boundaries and the pitfall of double counting
An emission factor belongs to a unit, and that unit is not always what you think. For fuels, factors exist per litre and per kilogram, for gas per cubic metre and per gigajoule, for transport per vehicle kilometre, per passenger kilometre and per tonne-kilometre. Mix one up, and the result is off not by a few percent but by an order of magnitude. On top of that comes the system boundary of the factor itself: does it count only the combustion, or also the extraction, refining and transport of the fuel to get there? That choice must be the same across the whole overview, otherwise you are comparing a lorry on one basis with a van on another.
The danger lies in double counting. The three scopes are mutually exclusive precisely so that the same emissions are not counted twice, but in a system fed from multiple sources it happens anyway. The classic case is the trip that comes in both via your own driver's fuel card and via the shipment line of a carrier accounting for the same load. Or the electricity of a leased business premises that sits both in your own scope 2 and in the recharge to the tenant. Or a lease car that you report in scope 1 while the leasing company reports it as its own fleet, so that two organisations claim the same kilometres.
Within your own books, this is solvable with a rule that is not up for debate: for each activity there is exactly one leading source, and the other sources are there at most for verification. The system should flag when two records touch the same period, the same vehicle or the same connection. Between organisations it is harder, because you cannot dictate what the other party does. What you can do is record on what basis you included an item, so that in a supply chain query you can show that you are not double counting but using a different boundary from your supply chain partner.
There is another form of double counting that is less noticeable: between years. A correction invoice for December that arrives in January belongs to the old year, not the new one. Without period allocation per line, consumption shifts unnoticed, and a reduction appears to have been achieved that is merely a booking difference.
Test your idea first: a working prototype in 1 day
With OneDayBuild, we make your idea tangible in a single day for €1,150, so you know whether further development is worth the investment. Decide to go ahead with the full build? Then we deduct the cost in full.
View OneDayBuild →Recalculation is a feature, not an incident
Sooner or later, an earlier year stops being correct. An emission factor is revised, a calculation method is amended, a site is acquired, or a supplier finally sends the real figure where an estimate used to be. The question is then not whether you have to do something, but whether your system can handle it. If you apply the new factors on top of the old calculation, you lose the result that went out last year.
The schemes are more explicit here than much software. Handbook 3.1 of the CO2 Performance Ladder, managed by SKAO, states that a methodology change in an emission factor is always a reason to recalculate the reference year, whereas technological progress or changed market conditions are not. The recalculation must be clearly documented, and the handbook refers to NEN-EN-ISO 14064-1, the standard for quantifying and reporting greenhouse gases at organisation level. Functionally, that means: being able to rerun a closed year with a different factor set, keeping both results and recording the reason.
Base year and recalculation policy
A reduction target needs a benchmark, and that benchmark is the base year. Everything you say about progress afterwards is a comparison with that single year. That makes the base year the most sensitive number in the entire accounts: if it shifts, your whole story shifts with it. That is exactly why there should be a documented policy that determines when you recalculate retroactively, rather than a decision made case by case by whoever happens to be producing the report.
Such a policy describes at least three things. First, which events trigger a recalculation: a structural change to the organisational boundary through an acquisition or divestment, a change in calculation method, or the discovery of an error in previously reported data. Second, which events do not, such as organic growth or contraction, or switching to a different fuel. Third, the size of deviation at which a correction is worth making, because applying every correction of a tenth of a percent makes the series unreadable without making it any better.
In software, that policy is not a document but a setting. The event is registered, the system proposes the recalculation, someone with the right role approves it, and the new series sits alongside the old one with a reference to the reason. Anyone who later looks up an old annual report must still be able to see which figure was in it at the time, and which steps led from that figure to the current one. Overwriting is not a technical choice here but a way of throwing away evidence.
Estimated, measured and the gap between them
Every CO2 inventory contains three kinds of figures mixed together. Measured values come from a meter or an invoice with a quantity on it. Derived values you calculate from another record, such as kilometres from a trip log. Estimated values are an assumption: spend multiplied by a factor, an industry average, an allocation by floor area. All three are usable, provided the label is on the line and not in a footnote.
That label must count towards the totals. For each scope and category it should be visible what share of the total rests on estimates, because a largely estimated footprint is a different claim from a measured footprint. It also shows where requesting data pays off: in the category that is large and weakly substantiated. If a supplier later provides a real figure, the line changes from estimated to measured, with the date and reason as a visible change.
Treat data quality as a property of every figure, just like the unit and the period. Ideally, a line carries four things with it: where the value comes from, how it was derived, how certain it is, and when it was last confirmed. That last one is often forgotten, yet a supplier figure from an old calendar year is worth something different from a figure from last month. With those four fields, you can produce a report that shows not just the total but also how the total is built up, without anyone having to open the underlying tables.
It also makes the conversation more honest. An organisation that openly shows that a large part of its scope 3 rests on sector averages is more credible than one that presents an exact-looking figure built on the same assumptions. And internally, it gives direction: improvement starts with the item that is both large and poorly substantiated, not with the item that is easiest to measure.
The audit trail, and when an off-the-shelf package is enough
As soon as an auditor or a verification body gets involved, the question changes. They pick a line from the report and want the way back: which consumption lies behind it, which source that measurement came from, which emission factor was used, which version of it, who entered it and who approved it. If a link is missing, the figure cannot be verified. A closed year should be frozen, so that a later correction appears in the trail as a correction.
What a verifier actually asks for
A verification rarely takes the form of a discussion about methodology. It is a sample. The verifier selects a number of items, usually the largest ones and a few random ones, and follows them down until they reach a document that your system did not create: a supplier invoice, a meter reading, a maintenance report, a delivery note. What they want to see is that the path there runs without a manual intermediate step, and that the assumptions, sources and system boundaries are recorded at every step. The standard for organisation-level emissions inventory reporting, NEN-EN-ISO 14064-1, asks the same of the report itself.
In practice, that comes down to a handful of requirements for your system. Source documents must be retained and linked to the line, not kept in a shared folder with its own naming conventions. Changes must leave a trail showing who, when and why, and the system must distinguish between entering, changing and approving, so that one person cannot single-handedly push a figure through. A published version of the report must be kept as a version, including the data it was based on. And the verifier needs read access to everything they have to be able to follow, without being able to change anything. If you only build this in once the first verification has been announced, you end up building it twice.
The footprint of a single product alongside that of the organisation
Sooner or later, a customer will ask not for your organisation's figure but for the figure for what you supply to them. That is a different way of calculating, and it is wise not to let the two run together. An organisational footprint adds up everything emitted in a period within a defined organisational boundary. A product footprint instead follows a single product through its life cycle, from raw material through production and use to disposal, and expresses the result per functional unit: per item, per kilogram, per service delivered. The first is a total for a period, the second a ratio.
To substantiate product figures, schemes refer to their own standards. Handbook 3.1 of the CO2 Performance Ladder cites studies in line with ISO 14067 on the carbon footprint of products, or the GHG Protocol Product Life Cycle Accounting and Reporting Standard, and for materials the Dutch National Environmental Database (Nationale Milieudatabase). The difference from organisational accounting lies mainly in allocation: your gas consumption is a single figure, but how much of it belongs to product A and how much to product B is something you must decide explicitly, for example by machine hours, weight or order lines.
In software, that translates into two models on the same source data. Consumption is collected once and validated once; on top of that sits an organisational model that totals per period and per entity, and a product model that distributes to products or orders via an allocation key. Two separate systems side by side will inevitably produce two figures that cannot be reconciled, and sooner or later you will be asked about it.
The level of that assurance is laid down in European regulation. The recent European simplification kept to limited assurance and dropped the mandate to develop standards for reasonable assurance. Have your auditor confirm which deadline applies to your situation, because those dates have shifted more than once in recent years. For certification schemes, NEN-EN-ISO 14064-3 applies, covering the validation and verification of greenhouse gas statements. The audit trail is therefore not a module you switch on later, but a property of the data model, familiar to anyone who has previously built audit software.
Does that always mean custom software? No. Mature carbon accounting packages exist, with factor libraries, supplier portals and ready-made exports. If your consumption comes from a handful of standard sources and an annual footprint is the goal, you buy it in and your effort shifts from building to configuring. Custom software becomes interesting for allocation: emissions per project, per product, per customer or per trip, in a breakdown that exists only in your organisation. Or when the data has to come from your own operational systems in volumes that nobody uploads manually any more. See also software for the manufacturing industry.
If it's not your own emissions but those of the fuels you supply under EU ETS-2, read more about our software for ETS2 monitoring of fuel supplies.
- Off-the-shelf package if consumption comes from a handful of standard sources
- Off-the-shelf package if an annual footprint is the goal
- Custom software for allocation per project, product or customer
- Custom software if the data has to come from your own systems
- Middle route: a package as the calculation core, custom software for the data feed
Frequently asked questions about carbon accounting
Carbon accounting is the calculation work underneath: collecting consumption data, converting it with emission factors and keeping the result traceable to the source. Sustainability reporting is the accountability layer above it, with materiality, defined data points and publication. You may need the former without falling under the latter, for example because a tender requires figures.
You calculate scopes 1 and 2 from data that is already in your own records: meter readings, gas invoices, fuel cards. Scope 3 covers emissions elsewhere in the value chain, divided into fifteen categories in the Corporate Value Chain (Scope 3) Standard. You buy a product, not a dataset. Whatever a supplier does not provide you have to estimate, and it must remain recognisable as an estimate.
For the Netherlands, they are published on co2emissiefactoren.nl, an initiative of SKAO, Stimular, Connekt, Milieu Centraal and the Dutch government, managed by Rijkswaterstaat and Stichting Stimular. The list is reviewed annually and usually updated around the turn of the year. The advice is to use the factor that belongs to the year you are reporting on.
That depends on the cause. Handbook 3.1 of the CO2 Performance Ladder states that a change in methodology is always a reason to recalculate the base year, whereas technological progress or changed market conditions are not. The recalculation must be clearly documented. So build it in as a feature, showing both outcomes side by side.
By putting the label on the line itself rather than in a footnote to the report. Every value carries whether it is measured, derived or estimated, along with the method. The system adds that up, so for each scope you can see which part of the total rests on estimates and where requesting actual data is worthwhile.
A trail from every line in the report back to the underlying measurement: which consumption, which source, which emission factor and which version of it, who entered it and who approved it. In addition, a closed year must be frozen, so that a later correction shows up as a correction and not as a silent overwrite.
Related services
CSRD and ESG reporting software
If it is about the report itself, with materiality and data points, CSRD and ESG reporting software is the place to start.
Energy management software
If you want to get consumption in order first, with meter integrations and control per site, see energy management software.
Custom ERP system
If the figures need to land within the administration itself, alongside purchasing and production, a custom ERP system is the broader starting point.
Compare your current calculation with our questions
Tell us where your consumption data comes from, which emission factors you use and what happens when someone asks how a figure from last year was arrived at. Those answers usually show whether an off-the-shelf package, custom software or a combination fits.
If you work with the CO2 Performance Ladder, also look at software for the certification file and at the app for CO2 registration on the project.