Centralising emissions data across multiple sites and suppliers takes four steps, in this order: set and document the organisational boundary (operational control, financial control or equity share), define a common data model for every site, automate capture from sources that already exist (invoices, ERP, telemetry, supplier portals), and put governance in place with named owners, a cadence and change control. Skipping the first step is what forces companies to rebuild the whole inventory later.
Why multi site emissions consolidation usually fails
It rarely fails for lack of data. It fails because every site has its own and none of it lines up:
- Each plant uses a different template, with different field names and different units.
- Emission factors from different years and sources get mixed, so comparing sites means nothing.
- There is no shared rule on what is included, so one site reports fleet diesel and another leaves it out.
- Periods do not match: some send calendar years, others fiscal years or loose months.
- Data arrives by email, gets consolidated by hand, and nobody knows which file produced the final number.
- Subsidiaries and suppliers get counted twice, or not at all, because ownership of each figure is unclear.
The result is a number that cannot survive an awkward question and has to be rebuilt from scratch every year. The fix does not start with tooling. It starts with two methodological decisions.
Step 1: set the organisational boundary
Before collecting anything, decide what counts as “the company” for emissions purposes. Chapter 3 of the GHG Protocol Corporate Standard defines two consolidation approaches, with the control approach split into two criteria. ISO 14064-1:2018 allows the same three options. Choosing one and applying it consistently is a requirement of both frameworks.
| Approach | What you consolidate | Good fit when | Friction point |
|---|---|---|---|
| Equity share | The share of emissions matching your economic interest in each operation | You hold many minority stakes and material joint ventures | Requires data from entities you do not control |
| Financial control | 100 per cent of emissions from operations whose financial and operating policies you direct | You want the perimeter aligned with consolidated financial statements | Facilities you operate but do not consolidate fall outside |
| Operational control | 100 per cent of emissions from operations where you have full authority to introduce your policies | You directly run plants, fleets or buildings | Drifts away from the accounting perimeter and complicates financial reporting |
Two practical notes. First, in most cases financial and operational control give the same answer; the differences appear in complex ownership structures. Second, the approach you pick changes how emissions are classified afterwards: a facility outside your organisational boundary can still generate indirect emissions for you through the value chain.
Document the decision with the full list of legal entities, sites and facilities, ownership percentages, who operates each one and whether it is in or out. That document is the first thing a verifier asks for.
Step 2: define operational boundaries and categories
With the perimeter fixed, decide which sources count and how they are classified. Under the GHG Protocol you will use scope 1, scope 2 and the fifteen scope 3 categories. Under ISO 14064-1:2018 you will use categories of direct and indirect emissions. The practical differences are set out in our comparison of the GHG Protocol versus ISO 14064-1, and the category detail in our guide to the 15 scope 3 categories.
In a multi site group, what matters is that the rule is single and written down. One emission source catalogue that applies to every site, plus an instruction for what to do when a site has a source the others do not, removes half the incidents.
How do you avoid double counting between sites and suppliers?
Double counting shows up in four specific places, and all four are solved with written rules:
- Between sites of the same group. If site A invoices services to site B, those emissions already sit in A's scope 1 or 2. They cannot be counted again as B's purchases in the group total.
- Between scope 1 and scope 3. Transport with your own fleet is scope 1. The same trip subcontracted is scope 3. Assigning one route to both is the classic error.
- Between scope 2 and scope 3. Electricity consumption is scope 2, while transmission and distribution losses for that electricity are scope 3, category 3. They are different figures and they do not overlap.
- With suppliers. Your scope 3 is your supplier's scope 1 and 2. That is not double counting inside your inventory, but it becomes double counting if you mix a supplier's primary figure with a spend based estimate of the same purchase.
The operating rule is simple: every euro spent and every physical unit consumed has exactly one data owner and one category. If two people can claim the same consumption, the rule is not written properly.
Step 3: one data model for every site
Centralising is not putting every spreadsheet in the same folder. It is making every activity record share the same structure. These are the minimum fields that make consolidation automatic and auditable:
| Field | Why it matters |
|---|---|
| Legal entity and site | Allows consolidation by company, country and location |
| Emission source and category | Classifies into scope 1, 2 or 3 and prevents duplication |
| Period start and end | Aligns calendars across sites and avoids gaps or overlaps |
| Quantity and normalised unit | Stops litres being mixed with kWh, or short tons with metric tonnes |
| Emission factor, source and year | Makes the calculation reproducible and supports base year restatement |
| Data origin | Metered, invoiced, estimated or supplier provided, so you can report data quality |
| Linked evidence | Pointer to the document behind the figure, for verification sampling |
| Owner and upload date | Closes the accountability chain and lets you chase on time |
With those eight fields, consolidation stops being a project and becomes a query. And when a factor changes, you know exactly which records to restate.
Step 4: automate capture source by source
Not every source automates the same way, and promising daily data for everything is a reliable way to fail. This table sets expectations.
| Data | Where it comes from | How it is automated | Realistic frequency |
|---|---|---|---|
| Electricity and gas | Supplier invoices or customer portal | Automated invoice reading and extraction per supply point | Monthly |
| Metered electricity consumption | Monitoring or smart metering system | API integration with the energy management system | Daily or hourly |
| Fleet fuel | Fuel cards and telematics | Periodic file from the card issuer or telematics API | Monthly, weekly with telematics |
| Purchased goods and services | ERP or procurement system | Periodic purchase ledger extract with category mapping | Monthly or quarterly |
| Subcontracted logistics | Logistics operator or carrier | File exchange or API with tonne kilometre data | Monthly |
| Business travel | Travel agency or expense tool | Automated agency report with routes and class | Monthly |
| Waste | Authorised waste manager | Periodic download of transfer documents and certificates | Monthly or quarterly |
| Supplier primary data | The supplier | Structured campaign with a common form and reminders | Annual |
The golden rule is that the slowest source sets the frequency of the consolidated figure, not the dashboard. A dashboard refreshing every minute on top of monthly invoices is still a monthly dashboard. We develop that idea in our article on real time carbon footprint data.
Where primary data does not reach, the acceptable path is estimating with spend factors or physical proxies, documenting it, and improving coverage year on year. AI helps most with the tedious part: reading heterogeneous invoices and delivery notes, classifying purchase lines and assigning factors. You can see how that works in Manglai's artificial intelligence layer and in our analysis of AI applied to scope 3 calculation with supplier data.
Governance: who does what, and when
A data model without governance degrades within two reporting cycles. These are the minimum roles in a multi site group:
| Role | Responsibility | Cadence |
|---|---|---|
| Site data owner | Upload and validate their site's data and keep the evidence | Monthly |
| Corporate sustainability lead | Maintain the source catalogue, the factors and the consolidation rules | Continuous, reviewed annually |
| Finance or controlling | Align the perimeter with the accounting consolidation and grant ERP access | Quarterly |
| Procurement | Run supplier data requests and add data clauses to contracts | Annual, tracked quarterly |
| Executive team | Approve the inventory, the reduction plan and any exceptions | Annual |
Add three mechanisms that prevent most fires: a monthly close on a fixed date, a data issue log with owner and deadline, and change control that records who edited what after close. Write down the criteria for restating the base year and your history stays comparable when methodology changes or a new site joins.
What changes if you file with MITECO or report under CSRD
- Spain's MITECO carbon footprint registry. Since January 2026 only emissions attributable to Spanish territory can be registered, so a group with plants in several countries needs to isolate the Spanish subset of its global inventory. Registration also requires scopes 1 and 2 as a minimum, and the registry support document treats any source above 5 per cent of total scope 1 plus 2 emissions as significant, so only sources below that threshold can be excluded.
- CSRD sustainability statement. The reporting perimeter aligns with the consolidated financial statements, so financial control usually creates the least friction where the directive applies to your company.
- Supply chain requests. When large customers ask for data, they want the share that relates to them, not your total. Being able to segment by customer, product or service stops being a luxury.
For the first case, the process and requirements are covered in our MITECO registry guide. For the second, start with the CSRD solution.
Centralisation checklist
- Choose and document the consolidation approach, listing entities and sites included and excluded.
- Write one emission source catalogue that applies to every site.
- Define the data model with mandatory fields and ban local templates.
- Centralise the emission factor library, with source, year and unit, and one person who updates it.
- Map every data source to an automatable origin and set a realistic frequency for each.
- Name a data owner per site and confirm it in writing to their line manager.
- Set a close calendar and a deadline for chasing missing data.
- Write the anti double counting rules into the same document as the source catalogue.
- Tag every record with its origin so you can report data quality by category.
- Review coverage annually and move up one step in the data hierarchy for the most material categories.
Frequently asked questions
Which consolidation approach should we choose?
If your priority is aligned financial and sustainability reporting, financial control is usually the easiest because it matches the consolidated perimeter. If you directly run plants and fleets and your goal is operational reduction, operational control better reflects what you can actually change. Equity share makes sense with many material minority holdings.
Can we change consolidation approach later?
You can, but you must restate the base year and the historical series so comparisons still hold, and explain the change in the report. That is why it pays to decide well up front.
How do we bring in supplier emissions without duplicating them?
By deciding, for each purchase, whether you use supplier primary data or a spend based estimate, never both. And by checking that what the supplier reports covers only the part of their activity that serves you, not their whole inventory.
Does every site need the same level of detail?
No. It is reasonable to require primary data at the sites that concentrate most emissions and accept estimates at smaller ones, as long as the criterion is written down, applied consistently and the estimation method is documented.
How often should data be consolidated?
A monthly close is a good balance: frequent enough to catch errors while the evidence is fresh, spaced enough not to overwhelm site owners. The annual inventory is then built on those closes rather than assembled in January.
Can we centralise without changing each site's systems?
Yes, and it is usually the fastest route. What you centralise is the data model and the consolidation layer, not the source systems. Each site keeps its ERP, and integration happens through periodic extracts or APIs.
If you need a first measurement while you build all this, start with the carbon footprint calculator for companies, and if the main challenge sits in the supply chain, the specific approach is in our software for supply chain managers.
