If your energy data does not link each kWh to a site, meter, bill, and date range, your reporting can go wrong fast. I’d start with one goal: build a source-traced pack that shows consumption, meter IDs, bill status, and contract proof for any renewable electricity claim.
Here’s the short version of what matters:
- Collect the core fields first: site, legal entity, address, MPAN or MPRN, meter serial number, billing dates, kWh, read type, and spend.
- Keep direct supply and landlord supply separate: this stops double counting and boundary errors.
- Split Scope 2 data properly: location-based uses grid averages; market-based needs contract proof like REGOs, PPAs, or supplier statements.
- Check data quality before you model anything: flag estimated bills, partial periods, missing reads, and duplicate invoices.
- Use the right source for each item: bills, half-hourly exports, landlord statements, inverter or SCADA data, EV charging logs, and contract files.
- Store everything in a clear structure: entity > site > meter > contract.
A few checks will catch most problems. For example, half-hourly data gives 48 intervals per day, and if billed use differs from HH or AMR data by more than 2–3%, I’d stop and review it. I’d also check that contract-backed electricity does not exceed total site use, and that CCL kWh matches billed kWh.
Quick comparison
| Area | What I’d collect | Main source | Main check |
|---|---|---|---|
| Consumption | kWh by site and billing period | Supplier bills, HH data | Estimated vs actual reads |
| Metering | MPAN, MPRN, meter serial number | Bills, meter records | Match meter to site |
| Landlord supply | Service charge, sub-meter data, apportionment basis | Landlord statements | Avoid overlap with direct bills |
| Renewable claims | REGOs, PPA terms, supplier attestation | Contract files | Check retirement year |
| On-site generation | Generated, imported, exported kWh | Inverter, SCADA, O&M logs | Match to site and dates |
| EV charging | Charger ID, kWh, vehicle ID | Charging logs, fleet reports | Keep separate from site load where needed |
So if I were doing due diligence, I would not start with emissions factors. I’d start by making sure the raw data is complete, traceable, and tied back to the original records.
Energy Data Collection Process: From Source to Emissions Reporting
What energy data you need
Consumption, meters and organisational boundaries
Start with the basics: site name, legal entity, service address, meter ID, billing period, kWh, read type and spend. Those are the core fields. Each figure should trace back to a bill, a meter and a legal entity before you assign it to the right organisation, reporting period or asset.
You should also record the MPAN (Meter Point Administration Number) or MPRN (Meter Point Reference Number), along with the meter serial number, to help prevent crossed-meter errors [2].
Gas bills usually show kWh, but raw meter reads are often in m³. If raw gas reads are all you have, convert the volume to kWh using the supplier's methodology [2].
For larger half-hourly sites, capture agreed import capacity in kVA as well. And record the VAT rate shown on the bill.
The table below sets out the minimum fields to collect by category:
| Category | Fields to capture | Purpose |
|---|---|---|
| Identity | Site name, legal entity, service address | Assign organisational boundary |
| Metering | MPAN / MPRN, meter serial number | Match asset to billing record |
| Consumption | kWh, billing period (start/end), read type (A/E) | Quantify use and flag data quality issues |
| Financials | Unit rate (p/kWh), standing charge, VAT rate, CCL | Validate costs and check levy treatment |
| Network | Agreed import capacity (kVA), profile class | Assess capacity and load profile |
Utility bills are only part of the picture. You should also collect a few extra fields for cases like these:
- Leased sites: landlord meter reference, service-charge basis, apportioned kWh
- On-site generation: imported kWh, self-generated kWh, export kWh, asset reference
- EV charging: charger ID, charging kWh, site, date range
After you have these fields in place, separate grid consumption from contract-backed electricity for Scope 2 reporting.
Location-based versus market-based electricity data
For Scope 2 reporting, you need to decide which method applies before you do any calculations. The location-based method uses the average emissions intensity of the national or regional grid and applies it to total kWh consumed. The market-based method uses the emissions intensity of the electricity you have bought through contract, which may be lower if you hold REGOs, a PPA or a certified green tariff [3].
The data you need is not the same for both methods:
| Location-based | Market-based | |
|---|---|---|
| Data required | Total grid electricity (kWh) by site and country | Consumption plus contractual instruments (REGOs, PPAs, supplier attestations) |
| Emission factor | National/regional grid average | Supplier-specific or residual mix |
| Primary source | Supplier bills or meter readings | Supplier attestations, contract terms, certificate evidence |
| Typical due diligence use | Default disclosure for SECR and baseline reporting | Demonstrating the impact of renewable procurement |
| Key limitation | Does not reflect specific green energy purchases | Requires rigorous evidence of certificate retirement |
If part of the consumption has no contractual instrument behind it, use residual mix factors for that share. This helps avoid double-counting renewable attributes that another buyer has already claimed [3].
Use the correct method from the start so your evidence trail lines up with the reporting basis.
When annual totals are not enough
Annual totals are fine for high-level reporting, but they won't tell you much about peak demand, billing issues or site-level risk. That's where half-hourly data comes in.
Half-hourly data gives you 48 intervals per day and is required for Profile Classes 05–08 and 00, which cover larger commercial and industrial sites [2]. It is also useful when you need to look at peak demand, tariff exposure or load shape. In practice, this can help you spot data gaps, unusual consumption patterns and periods of heavy use that matter when you're making investment decisions.
sbb-itb-6ca8558
Where to find each data point
Once you know which fields you need, the next step is simple: pull them from the right records.
Supplier bills, portals and half-hourly exports
Start with the supplier invoice. For UK electricity, the bill usually shows the MPAN, billing period start and end dates, total kWh, unit rate in pence per kWh, standing charge, VAT and a Climate Change Levy line item. Gas bills work much the same way, but they use the MPRN and show consumption on a Gross Calorific Value (GCV) basis. Use the matching GCV conversion factor, not the Net CV one [8].
One field often missed during extraction is the read status code. That small letter matters more than it looks:
- A = Actual
- E = Estimated
- C = Customer
- D = Deemed [7]
When you join bills, meters and sites, use the 13-digit MPAN or 6–10 digit MPRN as the key. Account numbers can change when a contract renews. Meter identifiers do not [7].
For half-hourly sites, ask the Meter Operator (MOP) for the raw half-hourly data file as well as the invoices. That gives you a way to spot gaps or scaling errors that a rolled-up bill can hide [7][11].
If there is no direct bill, use landlord records and service-charge documents instead.
Leased sites, landlord energy and service charges
For landlord-served sites, trace the invoice, the sub-meter and the apportionment basis before you start modelling consumption. In practice, that means asking for service charge reconciliations, landlord utility schedules and sub-meter reports [1][5].
For each leased site, map out who gets each invoice and which meter serves which area. Also ask for REGO retirement confirmation for the reporting year [10][9].
If tenant-specific use sits inside a service charge instead of being metered on its own, floor-area apportionment is usually the clearest basis you can defend. Write that basis down plainly. Auditors and assurers tend to pay close attention to boundary decisions in leased buildings [4].
For generation and charging, move beyond the bill and check the asset and contract records.
On-site generation, renewable contracts and EV charging records
Generation and charging data do not usually appear on a standard supplier bill. They come from non-bill sources, but they still need to map back to the site, date range and kWh.
| Energy type | Primary source | Back-up source |
|---|---|---|
| On-site generation | Inverter or SCADA export | O&M logs, PPA contracts |
| Power Purchase Agreement (PPA) | PPA contract + supplier attestation | REGO retirement confirmation |
| EV charging | Workplace charging records | Fleet management reports, fuel-card statements |
For generation assets, keep the generated kWh, exported kWh and any Feed-in Tariff data. For EV charging, keep the kWh and vehicle ID from the workplace charging records [9][7].
For renewable contracts, check that the REGO was retired in the reporting year. If the certificate was retired outside that period, it cannot support a claim for consumption in that same period [10].
How to check and fix imperfect energy data
Once you have the bills and exports, run these checks before any emissions modelling.
Estimated bills, partial periods and missing reads
Flag estimated bills straight away. The read status code tells you what you’re looking at, so check it against the definitions in the supplier-bill section above. An estimated read can throw off both consumption totals and emissions figures, especially when the estimate has drifted away from what was actually used [3][6].
The fix is simple in principle: ask for an actual read and a reissued bill. If that can’t be done in time, compare the billed total with half-hourly (HH) or Automated Meter Reading (AMR) data. If the gap between the invoice and HH data is more than 2–3%, stop and check it before using the number [11]. If there are still missing periods, allocate consumption by days in scope using:
billed kWh × days in scope ÷ total billing days
Record the missing period and note what you did to recover it [4].
Reconciliation checks that catch most errors
A few checks will catch most billing and mapping problems without turning the job into a marathon.
| Issue | How it appears | Likely cause | Immediate fix |
|---|---|---|---|
| Estimated billing | "E" flag next to meter reads | No meter access or failed comms link | Submit actual read to supplier; reconcile with HH data [11] |
| Partial period | Bill dates don't match reporting cycle | Supplier billing cycle misalignment | Pro-rate consumption by days in period [7] |
| Missing reads | Gaps in HH data file | Meter comms failure or data collector error | Request the missing half-hourly or substitute-read data [11] |
| Duplicate billing | kWh totals exceed expected site capacity | Overlapping landlord and direct bills | Reconcile overlapping invoices and remove the duplicate [1] |
| Identifier mismatch | Meter serial number on bill does not match the physical meter | Incorrect site mapping in ECOES or supplier database | Verify via ECOES (electricity) or Xoserve (gas) [11] |
Two checks deserve a place in every dataset, big or small.
- Confirm that contract-backed electricity volumes do not exceed total site consumption. If they do, you may have over-procured Energy Attribute Certificates or included a duplicate bill [9].
- Check that the Climate Change Levy (CCL) kWh on each invoice matches the billed kWh. If it doesn’t, that often points to a mid-period rate change or a charity exemption error [7].
A simple data structure for analysis and reporting
After the data passes QA, store it in a way that lets you trace each figure back to its source.
Use a clear structure: entity > site > meter > contract. At site level, keep a data map that records meter types, MPAN or MPRN numbers, suppliers, billing formats and any known gaps [1]. Add a short method note alongside it so your boundary definitions, conversion factors, and any interpolation or annualisation method are written down the same way each time [5].
Put all consumption into kWh and all emissions into tCO2e. If any figure is estimated, label it clearly and state the method used [1][5]. Also record the DEFRA year for every factor. Use the wrong year and your location-based Scope 2 result can end up off target [9].
Conclusion: build a source-traced energy pack before you model emissions
First, set the reporting boundary. Then build the site and meter register. Every MPAN and MPRN in scope should be on that register before you ask for any bills [1][5].
Once the boundary is fixed, collect the evidence by source. Pull supplier bills and half-hourly data for directly metered sites. Add landlord statements and sub-meter reads for leased space. Bring in on-site generation, EV charging, and contract records for market-based electricity claims [1][12][10]. Keep direct supply and landlord-supplied energy separate from start to finish.
Handle data quality as a live control, not something you leave until the end. As you work through the data, flag estimated reads and partial periods, and log the gap-filling method used [4][3].
The end result is a source-traced pack where every figure is in kWh, the right DESNZ/DEFRA factor year is applied, and every estimate is clearly labelled. That gives you a clean pack ready for emissions modelling.
