Method
Coverage first, then two parsers on public committee papers. RAG is a delivery rating, not an impact score. Flags mark extracted text and human-checked ratings.
London coverage scan
Not a full audit. Each borough row has a confidence score. ≤ 0.30 means unscanned — treat as unmeasured, not confirmed missing. Scan: 360Giving registry + democracy portal + GrantNav + committee PDFs where accessible.
threesix_status
Whether the council itself publishes award data to 360Giving. Appearing as a recipient does not count.
Published within the last 12 months
Most recent file >12 months old
No publisher profile found
eval_accessibility
Format of publicly available grant performance — what happened after funding, not award decisions.
Per-project rows with extractable fields
Prose or aggregate, not per-project
No public performance trail found
award_eval_join
Whether the award layer and the eval layer could be linked. The core gap indicator.
Both layers exist and identifiers could link
Awards published; no public eval
Eval published; no 360Giving award file
Neither layer found publicly
reporting_cadence
Frequency of public performance reporting to committee or a public register.
Committee item every ~3 months
Annual grant round report
Appears without a regular slot
No public cadence found
join_feasibility
Quality of identifiers inside public eval documents.
Stable IDs + formal org names (e.g. TH 8xxx)
Formal names; some ambiguity
No public eval, or names too informal
Confidence score
0.00–1.00. Rows at ≤ 0.30 are defaults for unscanned boroughs.
Committee PDFs read; data observed
Landing pages; some PDFs checked
Light web search only
Default — not scanned
London Councils
Six Grants Committee packs (Mar 2024–Nov 2025). Method: Green 80–100 · Amber 55–79 · Red 0–54 (performance 70% / quality 10% / compliance 20%). Early packs have Table 2; later packs go narrative, so the RAG matrix is human-checked against the papers. Award IDs from the May 2025 360Giving workbook.
Tower Hamlets MCGP
Appendix 3 project summaries, eight packs (Nov 2023–Dec 2025). RAG and notes parsed from PDF text. Committee project IDs (8xxx/9xxx). No 360Giving award file for the council as publisher — these IDs cannot yet be joined to 360Giving grants. Officer notes are auto-extracted and may be truncated.
Grants Determination Cabinet Sub-Committee
Meeting list · 8 packs
Identifiers we use
We don’t invent a parallel ID space. Align to what already exists.
- Council — 360Giving org ID
GB-LAE-*and ONSE09*on the borough index. - TH project — committee Appendix 3 ID. Not a 360Giving grant ID.
- LC grant — programme ref (
01,03…) plus360G-LondonCouncils-*and charity number where the workbook has them.
Source packs
Official meeting pages plus the text extracts the parsers ran on. Full list on /sources. Amber ⚠ flags on the explorers mark human-checked RAG or auto-extracted narrative.
What this is — and isn't
Working explorers for Tower Hamlets MCGP and London Councils, from public committee papers
A coverage scan of London boroughs — awards vs eval vs join — with 360Giving GB-LAE-* IDs
Source extracts next to official meeting links, with flags where text was extracted or RAG was human-checked
A join between awards and evaluation — Tower Hamlets committee IDs are not 360Giving grant IDs
A live scan of all 33 boroughs — the coverage index is a static light scan; about 15 rows are unscanned defaults
Affiliated with 360Giving, London Councils, or any borough
Caveats
- ~15 boroughs are defaults — treat them as unmeasured, not confirmed MISSING/CLOSED.
- Tower Hamlets join HIGH refers to within-eval corpus matching, not 360Giving join.
- RBKC quarterly cadence is claimed in the VSSF FAQ; per-project appendix not confirmed — confidence 0.55.
- GLA is ACTIVE for its full grant portfolio; VCS community grants modelled here may differ.
- London Councils and GLA are pan-London bodies — they appear as callouts, not map cells.