Before any review starts, we ask the agency for its complete audit export, and this page explains in plain language what complete means, why it matters, and what a transparent public record looks like when the standard is met.
An audit reads the record the system kept, so a gap in that record is a place the audit cannot see, and a missing day, a missing log type, or an edited file can hide exactly the activity an audit exists to examine. Nobody has to suspect bad intent for this to matter, because gaps appear for ordinary reasons like export settings, retention windows, and honest mistakes, and yet a report built on a partial file can only make partial statements. Asking for the complete export up front protects the agency as much as it protects the review, since a chief who hands over everything can point to that fact when questions come.
The purpose of this review is oversight that the chief runs over his own unit, paired with reassurance the community can actually check, and both of those depend on the record being whole. Every item our rules raise is printed beside the innocent explanation that could account for it, because a finding is a lead to review and never an accusation, and the review is built to ask questions rather than to make charges.
Here is exactly what we attest to. CivicAttest attests to the completeness of the file we received, and not to the completeness of the underlying system. If the export the agency provides meets the standard below, our report says so, and if activity never reached the system's own logs, no review of any export could surface it, which is a limit we state plainly rather than paper over.
The table below shows the shape of a public transparency record built from a complete export, and it is worth noticing two things as you read it. The weeks run in an unbroken sequence with no missing period, and the deletion events appear right in the record instead of being hidden, because a system that deletes on schedule and says so is demonstrating discipline rather than concealing anything.
Illustrative example with fictional data
| Period covered | Searches by Officer A | Searches by Officer B | Searches with a case number | Deletion events recorded |
|---|---|---|---|---|
| June 1 through June 7 | 41 | 36 | 75 of 77 | 284 records deleted at the 30 day retention mark |
| June 8 through June 14 | 38 | 44 | 80 of 82 | 301 records deleted at the 30 day retention mark |
| June 15 through June 21 | 45 | 29 | 72 of 74 | 276 records deleted at the 30 day retention mark |
| June 22 through June 28 | 33 | 40 | 71 of 73 | 295 records deleted at the 30 day retention mark, plus 2 records deleted by administrator request with reasons logged |
| June 29 through July 5 | 37 | 42 | 77 of 79 | 288 records deleted at the 30 day retention mark |
A reader looking at a real record in this shape can see continuity, can see that nearly every search carried a case number, and can see the deletion schedule doing its job in the open, which is the kind of reassurance no summary paragraph can provide on its own.
You can read about report verification on the verification page, or return to the CivicAttest home page.