The share of the documents that produce a body’s cases which carry text a machine can extract — measured by fetching every catalogued document, not by sampling one per year. It is a statement about legibility, not possession: a scan we hold and cannot read counts against this number, and a document held somewhere other than our own store can still count for it. Reference layers have no document corpus at all, so the column says so rather than showing a zero that would read as a failure.
Coverage
Our coverage is a table, not a claim.
Every source we hold, named by jurisdiction and by body, with how deep we are in each and whether an approval rate can honestly be computed from it. Including — especially — the rows where the answer is no.
15
Sources held and classified
2
Can carry a published approval rate
13
Cannot — each row says why
Most of this table says no, and that is the honest shape of it rather than a gap we are apologising for. A source earns a published rate only when the record it comes from could have shown us a refusal, and when we hold enough of that record to divide by. Those are two separate tests and a source has to pass both.
So a body that only recommends is not rate-bearing no matter how complete our holdings are — it advises, and another body decides. A ledger of what was adopted cannot show a refusal, so a rate from it would be 100% by construction. And a source whose documents are only 88.7% readable is not one we will divide by, even though it records every outcome: the unreadable tenth is sitting right there as a scan, and nothing says a meeting we cannot read decided the way a readable one did.
| Jurisdiction | Body | Record kind | Readable text | Rate? | Why |
|---|---|---|---|---|---|
| Berkeley County | GIS zoning ledger | approved_only | not a document corpus | No | The source lists what was adopted. A refusal cannot appear in it, so a rate computed from it would be 100% by construction. |
| Charleston County | GIS parcel and sales layer | not_a_decision_source | not a document corpus | No | A reference layer, not a record of decisions. |
| City of Charleston | Board of Zoning Appeals | full_disposition | 99.2% | Yes | The source records every outcome, including refusals. |
| City of Charleston | City Council | full_disposition | 88.7% | No | Has readable text for 88.7% of its documents, below the 90% needed to divide by. |
| City of Charleston | Planning Commission | recommendation | 70.1% | No | The body recommends; another body decides. A recommendation is advice, not a result. |
| Dorchester County | GIS parcel and sales layer | not_a_decision_source | not a document corpus | No | A reference layer, not a record of decisions. |
| Dorchester County | GIS zoning ledger | approved_only | not a document corpus | No | The source lists what was adopted. A refusal cannot appear in it, so a rate computed from it would be 100% by construction. |
| Mount Pleasant | GIS zoning layer | not_a_decision_source | not a document corpus | No | A reference layer, not a record of decisions. |
| North Charleston | City Council | approved_only | not a document corpus | No | The source lists what was adopted. A refusal cannot appear in it, so a rate computed from it would be 100% by construction. |
| North Charleston | Planning Commission | recommendation | 96.5% | No | The body recommends; another body decides. A recommendation is advice, not a result. |
| North Charleston | Board of Zoning Appeals | full_disposition | 98.7% | Yes | The source records every outcome, including refusals. |
| Summerville | GIS rezoning and annexation ledger | approved_only | not a document corpus | No | The source lists what was adopted. A refusal cannot appear in it, so a rate computed from it would be 100% by construction. |
| Summerville | Town Council minutes | full_disposition | 10.9% | No | Has readable text for 10.9% of its documents, below the 90% needed to divide by. |
| Summerville | Planning Commission minutes | recommendation | 9.1% | No | The body recommends; another body decides. A recommendation is advice, not a result. |
| Summerville | Standing Committee minutes | recommendation | 13.2% | No | The body recommends; another body decides. A recommendation is advice, not a result. |
Reading this table
What “readable text” counts, and what it doesn’t.
Document coverage
The 90% gate
Below 90% readable we do not publish a rate from a body. What we cannot read is not unreadable at random, and nothing says a meeting recorded as a scan decided the way a readable one did — so a rate computed over the readable part quietly describes the part that happened to be legible.
The n ≥ 20 floor
Even where a rate is publishable, we render a percentage only at twenty or more final outcomes. Below that we show the count and list the cases: “seven of nine approved” is honest at any size, and “78%” at nine cases is noise wearing precision.
Final outcomes only
A rate divides ratified by denied. Deferred and withdrawn cases are counted and listed separately, never folded into the denominator — a pause is not a refusal, and counting it as one would understate every approval picture on this site.
This table is generated from the same registry the reports read from, so it cannot drift from what the reports actually do. When a source moves, this page moves with it.
What isn’t here
A jurisdiction we haven’t named is one we can’t answer for.
There is no row on this page for a body we have not classified, and there is no regional claim anywhere on this site. A commercial parcel can border two counties, and only the named ones can be answered for.
A source that has not been classified yet cannot carry a rate at all — that case fails closed on purpose, because adding a jurisdiction must never silently authorise a rate over it. How we grade what we hold →