INSIGHTS: read what successive campaigns are saying

01In one sentence #sec_13589a0e

One campaign says what is; two campaigns from the same control say what has changed — and INSIGHTS does just that: it produces a reading and a data export. This module does not issue any document and does not replace Reports.

02Before opening #sec_e35b8c26

INSIGHTS reads closed occurrences: executions whose scope has been frozen, whose decisions have been rendered, and whose balance is final. As long as no campaign has closed, the screen says so and sends you back to the campaigns: the first closure will open this screen; the second will make comparisons possible.

The record is daily: Results recorded up to … A closure from today is read tomorrow.

This module is reserved for certain editions. If it is locked, the report card says so.

03Series analysis #sec_90eb508e

Restitution & analysis › INSIGHTS. The page first counts, then lists.

Then Analyzable series, one line per control — first closure, last, number of terms, and the Differential button. Filters by resource type and by policy reduce the list.

A control is what remains identical from one campaign to another: same workspace · same resource type · same policy · same targeting mode · same scope. Two different campaigns that share these five facts form a single series — the origin does not sort anything.

Viewing the excluded lists what does not belong to any series, grouped by who can act, never by severity — an excluded item is not broken.

04The series #sec_7f53f943

Clicking on a control opens its series: the ruler, not equal, arming order — the axis is a rank, not a calendar: the real gap is written in the interval. Each notch is a closed occurrence; the next notch is drawn, empty.

Below, the record of each closed occurrence — immutable, in addition only — no recalculation, no button: arming, opening, closure, expected, requests, concluded, expired, decision distribution. A note is worth reading: concluded ≠ decided — an escalated request passes the hand without concluding.

An arming window restricts the series to a period. Comparing two occurrences proposes the pairs; Series findings opens the cross-reading.

05The difference #sec_212372ba

What has changed between these two executions. The screen starts with the pair — previous term, current term, each with its arming date — and a comparability banner: no change figure is displayed before it. If the two reports were not written under the same rule, it states this: the report is not recalculated under the new version: it is immutable.

Then the three concepts that should never be confused: common object (same identity on both sides), modified common object (content changed between the two freezes), non-common object (present on only one side).

And what the pair learns, in populations whose importance is not equal:

• Returned decision — the heaviest finding: *this is where the review changed*

of opinions — and the directions do not say the same thing: removing a confirmed access is not undoing a revocation.*

• Decided on one side only — what protects the reversal: «no one responded,

then we certified» is not a reversal.

• Identical decision.
• Entered the scope and Exited the scope .

The addition, displayed to be redone: the five lines and their union, so you can recount them. The detail comes next, count first, list next, object by object with its fingerprint — objects, without holder.

06The revocation finding #sec_df939268

From the difference: revoked objects have reappeared in the next execution. This finding does not judge the review. It observes that objects removed from the scope by a revocation decision appeared again in the following frozen scope.

It is only read if the second freeze is after the closure of the first; otherwise, the module refuses rather than returning zero — a zero would be read as underperformance.

And it says what it cannot say: REVIEWIT decides, it does not execute: no remediation engine exists in the product. Three causes produce the same observation — access removed then granted again; access never removed; access in the process of being revoked at the time of the freeze — no hierarchy. The rest is verified on your side: your organization's arrival, mobility, and departure system, and how a revocation decision is taken up and applied there.

07Series findings #sec_9fb44ecc

What the sequence of executions shows, in four findings, each with its written limit:

1. The control that never revokes — from three closures. *A question, never*

a verdict: a healthy fleet exists, and a control that revokes nothing can be a well-tuned control.*

1. The effort and the result — *it took N business requests to obtain one*

revocation.* No credit figure.

1. The quality of the control — half of the decisions are made in less than N hours;*

median and ninth decile, never an average.

1. The concentration of the load — carried by the 3 most loaded, reviewers*

distinct, aggregated, never nominative. No identity, nowhere: neither in the card, nor in the detail, nor in a filter, nor in the export.

08What you will do with it #sec_7d5b2733

The difference feeds the next campaign: a scope that enters and exits a lot is poorly defined; many reversals indicate that the question asked to reviewers is not stable. The revocation finding feeds a conversation with those who execute the removals — outside of REVIEWIT.

None of this is evidence. For evidence, it is the reports page.

09To remember #sec_f3c81656

A series is the same control closed multiple times. The differential counts what has changed and lets you redo the addition; the findings raise questions and write their limits. INSIGHTS does not judge, and does not name anyone.