GridOptic normalises public grid and flexibility data, preserves provenance, and grades the strength of the evidence toward a decision-grade dataset, not a complete-looking one.
GridOptic works from open data published by the six GB DNOs — NPG, UKPN, NGED, SPEN, ENW, SSEN — plus national datasets from NESO, the GB system operator, principally the TEC connection register and the tRESP GSP-level demand pathways. Northern Ireland, a separate electricity market with no GB-style flexibility procurement registers, is out of scope.
Delivery years map to a canonical GB fiscal year (1 April to 31 March). Technology labels map into asset buckets — BESS, generation, demand, mixed, unknown — with combined service-direction strings split into contributing classes. Contract status reduces to a single contracted flag: only “accepted and contracted”, “availability and utilisation” or “utilisation only” are firm deals. Unnormalisable records keep their original labels.
A headline revenue per MW is built from two components, never by multiplying a fee by 8,760 hours. Availability annualises each contracted fee by the PAID service-window hours, with a per-product fallback when window fields are absent. Utilisation divides dispatch revenue by contracted MW and the fiscal years the evidence spans. Where a DNO publishes a procurement-level utilisation price with no dispatch evidence, it is annualised against an empirical activation-hours constant, not a readiness window. Components are reported separately; their sum is published only as an indicative estimate.
Each cell carries a tier — BENCHMARK, INDICATIVE or INSUFFICIENT. BENCHMARK needs at least twenty contracts, one MW, three distinct providers and no quality flag. A cell with fewer than five contracts, under one MW, or fewer than two providers falls to INSUFFICIENT. A SUSPECT quality flag — implausible magnitude, low magnitude, or a coverage mismatch where dispatch could not physically come from the contracted fleet — forces INSUFFICIENT regardless.
Each cell carries a 1-to-5 star price-evidence rating; forward cells also carry a 1-to-5 star requirement-evidence rating. Price stars derive from the tier, the diversity measure and the quality flag. Requirement stars assess whether a DNO has a stated forward need (natively published or held flat) and whether DFES growth, queue activity and upcoming-need signals corroborate it. Historical cells carry no requirement rating.
A cell with no priced availability and no priced dispatch is null across its money fields and renders as a no-data gap, never as zero. A cell whose implied activation hours exceed what its contracted fleet could run is flagged and its utilisation component withheld. GridOptic shows no number rather than a bad one; the gaps mark where published evidence cannot yet support a view.
The six DNOs publish at different cadences and detail: some publish provider identities; others do not. Where identities are absent, diversity is assessed by competition breadth — distinct tenders — with the tier hard-capped at INDICATIVE. The GB rollup is capped at INDICATIVE when fewer than two DNOs clear the floor. Where a forward year has procurement evidence but no stated requirement, it renders as an honest fade, not a guessed saturation.
Each cell exposes its evidence: contributing contracts and MW, distinct GSPs and tender references, and distinct providers where provider identities are published. Source labels resolve through a canonical GSP and substation layer to a single identifier; unmatched rows are kept with an explicit coverage state. The index is rebuilt on a schedule and snapshotted; a forced recompute is gated behind the cron secret. Ingestion is idempotent, deterministic and resumable.
Bring a postcode, coordinates, GSP or portfolio. We will show you what GridOptic can resolve.