Multi-site benchmarking, quartiles and outliers

Benchmarking starts with site-against-site comparison across the whole portfolio, presented as quartile and outlier views so the worst performers surface without anyone building a spreadsheet. Every comparison is drillable, from portfolio to site, building, meter and circuit, down to a single half-hour. Aggregations run hourly, daily, weekly, monthly and annually, so a facilities lead can review last Tuesday night and a finance director can review the full year in the same session. Role-based access decides who sees which sites, with unlimited users per account and permissions set per site, group or function.

A cold January measured against a mild one is not a comparison, it is a coincidence. Insights adjusts consumption using heating degree days and cooling degree days, so year-on-year and month-on-month movement reflects how a building was run rather than the weather it sat in. That matters most after a plant change or a schedule change, when someone has to prove the effect in front of a board. Weather-adjusted figures sit alongside raw consumption rather than replacing it, so finance still sees the number that will land on the bill.

Anomaly detection runs alongside the benchmark rather than separately, and it is usually how a site first appears as an outlier. Weekend usage, overnight spikes, missing reads and holiday baselines are flagged automatically. Thresholds are configurable per meter and per sub-meter in kWh, kW, £ or kgCO2e, and alerts go out by email, SMS or webhook as a daily digest or in real time, with escalation routing.

Internal ranking answers which of your sites is worst. It does not answer whether your best site is any good. For that, weigh your own figures against the industry-typical intensity ranges set out below, which are published sector ranges rather than Purely measurements. As industry-typical figures, offices commonly run 150 to 250 kWh/m2, retail 250 to 450 kWh/m2 and warehouses 90 to 160 kWh/m2. Baseload as a share of total kWh is typically 15 to 25 percent in offices and schools, and above 35 percent it usually points to a BMS schedule or always-on plant fault.

Two published examples show the pattern. A multi-academy trust with 22 schools found 6 sites with weekend baseload above 8 kW, reset the BMS across all 6 and projected a £41k saving for the academic year. A food and drink manufacturer on a single site using 12 GWh put sub-metering on three production lines, isolated a refrigeration setpoint drift, fixed it in one shift and brought baseload down 11 percent year-on-year. In both cases the benchmark identified the site, and circuit-level detail identified the cause.

Every benchmarking view exports. CSV, Excel, PDF and board-pack PDF cover the monthly reporting cycle, and a read-only JSON API is available on request for finance, BI, ESG and BMS integration. The same data set carries Scope 1 gas and Scope 2 electricity emissions on location and market bases, using DEFRA's latest published carbon factors refreshed annually, and feeds SECR, ESOS Phase 4, ISO 50001 evidence packs and CDP. Historical data reaches back up to 6 years where supplier data flows allow, which is what makes long-run comparison possible.