MeterBrief guides

What Is Half-Hourly Data? Who Holds It After MHHS, and How to Get an Export

2026-09-06

Turn a client's half-hourly export into a branded report in about a minute.

Try it on your data

Half-hourly data is the raw material of almost every serious energy decision an SME makes. Without it, a quote is guesswork, a flexible purchasing conversation is theatre, and a battery feasibility study is worthless. With the Market-wide Half-Hourly Settlement (MHHS) programme well into its migration phase, the parties holding that data have changed, and so has the way you ask for it.

What half-hourly data actually is

For each electricity meter point, half-hourly data is a series of 48 consumption values per day, one for every settlement period, expressed in kWh. A full year for a single MPAN is roughly 17,520 rows. Some files also carry kVArh (reactive power), maximum demand and export values where there is on-site generation.

It normally arrives as a CSV or Excel file, either as a grid of dates by 48 columns, or as a long list of timestamps and values. The timestamp convention matters more than people expect. Settlement periods run on local clock time, so clock-change days have 46 or 50 periods rather than 48. Some exports label a period by its start time, others by its end. Get that wrong and every peak-time analysis you run sits half an hour out.

Who has half-hourly data, and what MHHS changed

Before MHHS, the market split cleanly. Larger sites on profile class 00 with current transformer metering, plus profile classes 05 to 08 moved into half-hourly settlement under BSC Modification P272, sat with a half-hourly data collector and data aggregator appointed by the supplier or by the customer directly. Everything else was settled on estimated profiles from quarterly or monthly reads, even where a smart meter was quietly recording half-hourly consumption behind the scenes.

That split is now going. Elexon confirms suppliers began migrating meters into the new arrangements on 22 October 2025, with around 80 per cent of meters expected to be accepted by October 2026 and implementation completing in May 2027. The shortened settlement timetable follows in 2027. In practice, the great majority of non-domestic meters, including small shops and offices on ordinary smart meters, become half-hourly settled during this window.

Role names have changed with it, and brokers need to recognise them on paperwork:

  • The meter operator role splits into Metering Service Suppliers and Metering Service Agents.
  • The data collector role is replaced by data services: the Smart Data Service for SMETS meters, the Advanced Data Service for advanced and CT meters, and the Unmetered Data Service for unmetered supplies.
  • Data aggregation moves to Elexon central systems, so third-party aggregators no longer sit in the settlement chain.
  • A new Meter Data Retriever role lets an independent party pull smart meter readings via the DCC for settlement, which the Smart DCC has been building into its systems.

The honest answer to "who has my data" during 2026 is that it depends which side of migration your MPAN sits on. Elexon reported in mid-2026 that around a third of meters had migrated and eleven suppliers were actively migrating, with two suppliers roughly three quarters through their programmes. Until your meter moves, the old data collector arrangements still apply.

Where the data physically sits

For a migrated meter, readings flow from the data service into Elexon's Data Integration Platform rather than across the traditional Data Transfer Service flows the industry has used for years. That change worried the I&C market, because a large share of consultants and suppliers relied on the D0036 and D0275 flows through the DTS to obtain consumption data for forecasting and validation.

The gap has been filled commercially rather than by regulation. In March 2026, ElectraLink announced an agreement with Elexon for daily sharing of half-hourly settlement data received through the DIP, so API-based access continues while meters transfer. Elexon has separately consulted on broader access to the consumption data it collects, having flagged in March 2025 that it had built the ingestion process but not yet the sharing process. Access rules for non-domestic data remain contested. In May 2026 the ECA wrote an open letter to Ofgem arguing that authority to act, rather than personal consent, should govern access to organisational MPAN data, since a limited company or charity is not a data subject.

How a business or broker actually gets an export

Four routes are realistic. Pick the one that matches the meter and the deadline.

Route one: ask the supplier. Every supplier has a process for sending historical consumption to the account holder. Email the billing or account management team with the full 21-digit MPAN, the site address, the period required (ask for 24 months, settle for 12) and the format you want. Ask specifically for "half-hourly consumption data in CSV, period start timestamps, local time". Turnaround is typically five to fifteen working days, and it is free of charge for the customer's own data. This is the default route for smaller sites and for anything already migrated.

Route two: ask the data service or data collector directly. Where the site has a CT or advanced meter and the customer holds its own agent contracts, the Advanced Data Service (formerly the HH data collector) can usually issue a full history quickly, often through a customer portal with unlimited self-serve downloads. If the customer pays that agent directly, this is the fastest and cleanest source.

Route three: the meter's own portal. Many advanced meters come with a web portal from the metering provider giving near real-time data. Useful for monitoring, less useful for tender packs, because portal data is not always the settled data a supplier will price from.

Route four: a data aggregation service. Consultants and larger brokers commonly buy API access to half-hourly consumption data through ElectraLink or a similar provider, which avoids chasing suppliers site by site. It carries a cost and requires a valid customer authority.

Whichever route you use, a signed Letter of Authority naming the specific MPANs, the parties and an expiry date is the practical key. Suppliers routinely reject vague or undated LOAs, and post-MHHS scrutiny of data access has made them stricter, not looser. Send the LOA with the first request rather than waiting to be asked.

A checklist you can use this week

  • Confirm the meter type first. Digits one and two of the MPAN top line give the profile class: 00 means it was already half-hourly settled, 05 to 08 means P272 applied, 03 or 04 means it is likely mid-migration under MHHS.
  • Request 24 months of data, not 12. You need a full year plus a comparison year to spot pattern changes and metering faults.
  • Ask in writing for the data "as settled", and ask whether any periods are estimated. Estimated data in a tender pack produces prices that get corrected later, usually upwards.
  • Check the file on arrival: row count, unit (kWh not kW), timestamp convention, and any run of identical or zero values suggesting a communications outage.
  • Log the LOA expiry date. Most suppliers treat twelve months as the maximum, and an expired LOA restarts the whole request.
  • Where a site has export or on-site generation, request import and export series separately, or the net figures will mislead.

Should a supplier stall beyond fifteen working days, escalate to the complaints team in writing and note the date. Micro-business customers who exhaust a supplier's complaints process can take the matter to the Energy Ombudsman, which tends to concentrate minds. For anything more strategic, such as flexible purchasing or a demand response assessment, budget for a paid data feed rather than relying on periodic goodwill exports.

Turn a client's half-hourly export into a branded report

Baseload, capacity headroom, out-of-hours cost and a client-ready narrative, in about a minute. Edit anything, then send it under your name alone. Free preview on your own data.

Try it on your data