Methodology and limits

School Signals joins three kinds of public data so an edtech sales team can see, for each school and academy trust in England, which technology contracts it holds, who holds them, when they end and what has changed in the trust. The data is organisation-level only: no named staff and no personal email addresses.

Sources

Contains public sector information licensed under the Open Government Licence v3.0.

How contracts are selected

1. Buyer filter. A notice is kept when its buyer looks like an English school, academy trust, diocese education body or schools buying organisation, or a council buying for schools. Universities, NHS bodies, FE colleges, and Scottish, Welsh and Northern Irish bodies are excluded. A council notice is kept only when its title is clearly about schools or education, or its description mentions schools more than once. Council-wide social care, health, housing and corporate systems are excluded.
2. Edtech categories. Each notice is put into one of 13 categories by rules on its title, with the main CPV code as a fallback: MIS and school administration software; parent communication, payments and cashless; cyber security, filtering and safeguarding technology; classroom AV and interactive displays; print and MFD; devices and end-user hardware; networks, Wi-Fi and connectivity; telephony; learning platforms, content and assessment; school websites and digital; CCTV, access and visitor management; cloud, licences and productivity; IT managed services and support. Titles about buildings, catering, cleaning, energy, insurance, staffing and similar are excluded.
3. Normalisation. Each notice is parsed by the same open-contracting normaliser the lab uses for its UK tenders data feed. Supplier names that look like individuals or sole traders are withheld. Email addresses and phone numbers are stripped from text.
4. One row per procurement. Opportunity, award and repeat notices for the same buyer and title within about 13 months are merged. The award row is preferred, and missing dates and values are filled from the other notices.

How buyers are tied to schools and trusts

In order of confidence: Companies House number to trust; exact school name plus postcode; trust name, preferring open trusts when names repeat; school name when only one open school has it. Councils are tied to their local authority area. The match column states which rule was used. Unmatched education buyers are dropped rather than guessed.

Renewal timing

• contract_end is the published end date. When a notice gives only a duration, the end is counted from the contract start, award, signature or notice date and flagged in contract_end_derived. Find a Tender notices published before 2025 usually give only a duration.
• Extensions are not added to the end date. extension_possible and max_extension_end show when a notice allows more time.
• est_reprocurement_window_start and est_reprocurement_window_end (end minus 9 months to end minus 3 months) are a planning estimate, not a published date.
• later_notice_same_category points to a later notice from the same buyer in the same category, which may mean the contract was already re-procured.

Trust changes

• Moved to another trust: the school's trust today differs from its trust in the GIAS academy membership file 12 months earlier, matched by URN or through the predecessor link when the school received a new URN.
• Converted to academy: the school joined a trust in the last 12 months and its predecessor was a maintained school.
• New free school opened, Trust closed: from the same two snapshots.
• Pipeline: schools DfE lists as having applied to convert, as approved for conversion, or as matched to a sponsor. DfE does not publish the receiving trust for converter pipeline schools.

Limits (please read)

• Coverage is partial by design. Only notices published on Contracts Finder and Find a Tender are covered. Below the publication thresholds, schools and trusts often buy without a public notice, and framework call-offs are frequently unpublished. A missing contract does not mean the school has none.
• No claim of completeness or accuracy. Categories come from rules and will sometimes be wrong. Values are as the buyer published them: whole term or annual, with or without VAT. Buyers sometimes publish wrong dates.
• Timing changes. Contracts are extended, terminated early or re-awarded without a notice. Check the official notice (notice_url, or record_api_url for the machine-readable record) before acting.
• Snapshot. Each file is a snapshot on the date stated. GIAS changes daily and procurement data monthly.
• Northern Ireland, Scotland and Wales are not covered.

Personal data

None is collected or published. GIAS includes headteacher names and Contracts Finder includes contact names and emails. School Signals drops these fields and strips emails and phone numbers from free text. Rows describe organisations and their published contracts only.

Checks

Every build is checked against independent official sources (pipeline/verify.py). Renewal rows are compared with the Contracts Finder or Find a Tender API record for that procurement (incumbent supplier and end date or duration). Trust moves and conversions are compared with DfE's separate monthly open academies list. Trust sizes are compared with the number of academies DfE lists per trust. The latest result is in qa/verify-report.md.