# Monthly AI visibility report

Blank template and completed fictional example

Use the blank template for your own work and the completed fictional example to see how to record decisions, evidence and unresolved fields. Replace the example details before sharing. This manual file does not collect data or approve conclusions.

Pack version 1.0 · 14 September 2026. Template version 1.0.

Guide: https://rankecho.io/resources/prove/agency-monthly-ai-visibility-report

## Part 1 — Blank template

# Monthly AI visibility report

Manual client-report template — version 1.0. Keep every provider's unit separate, retain missing-data states, and distinguish observations after shipped work from outcomes attributable to that work.

## Report control
- Client:
- Reporting period and comparison period:
- Prepared by / reviewed by:
- Report version and UTC issue time:
- Included markets, languages, sites, providers, modes, prompts, pages, analytics properties, and business systems:
- Material scope or method changes since the prior report:

## Executive summary
What changed:
- [Material movement, exact unit, numerator/denominator, comparison, and scope]

What did not change:
- [Important stable or null result; do not omit it]

What we shipped:
- [URL or system, bounded change, owner, ship time, acceptance result]

What it means:
- [Plain-language interpretation with limits; use observed after unless attribution is supported]

What we will do next:
- [Decision, owner, due date, acceptance rule, and next observation window]

## Scope and comparability
| Layer | Population and unit | Current | Prior | Comparable? | Source/evidence pointer | Limitation |
| --- | --- | --- | --- | --- | --- | --- |
| Search-platform visibility | [Provider-defined unit] | | | Yes / No | | |
| Fixed prompt observations | [Observed cells / planned cells] | | | Yes / No | | |
| Identifiable AI referrals | [Sessions and landing pages] | | | Yes / No | | |
| On-site actions | [Declared event and count/rate] | | | Yes / No | | |
| Commercial outcomes | [Attribution rule and eligible population] | | | Yes / No | | |

## Findings
### Changed
For each finding: exact observation; prior/current numerator and denominator; scope; evidence pointer; plausible alternatives; confidence; client implication.

### Unchanged
Record important stable, null, or adverse results using the same fields. An unchanged result is evidence, not empty space.

### Unavailable, failed, excluded, or incomparable
Record each affected provider, prompt, page, metric, and period; the reason; denominator treatment; retry or resolution owner; and whether the client decision changes.

## Work shipped
| Ship time | URL/system | Problem and evidence | Bounded change | Owner/reviewer | Acceptance result | Rollback | Proof window |
| --- | --- | --- | --- | --- | --- | --- | --- |
| | | | | | | | |

Do not list planned work as shipped. Link one-fix before/after evidence to the proof-report record instead of copying its entire ledger into this monthly narrative.

## Decision and action queue
| Priority | Decision or next action | Evidence behind it | Owner | Due date | Acceptance rule | Next observation |
| --- | --- | --- | --- | --- | --- | --- |
| 1 | | | | | | |

## Interpretation and limits
- Provider-native metrics are reported under the provider's current definitions.
- Fixed-panel observations describe the declared prompts, surfaces, repeats, rules, and windows only.
- Referrals omit interactions that do not reach the site or arrive without usable attribution.
- On-site actions and commercial outcomes require their own event, join, and attribution rules.
- Sequence after a release is temporal evidence, not a causal estimate by itself.
- Note provider, model, interface, indexing, site, campaign, analytics, consent, and market changes that could affect comparison.

## Client decision
- Continue / change course / investigate / pause / retire:
- Decision owner:
- Decision date:
- Reason and evidence:
- What result would challenge this decision next month:

## Appendices
- Raw evidence locations:
- Fixed panel and classification rules:
- Search-platform exports:
- Analytics exploration and event definitions:
- One-fix proof reports:
- Version history and corrections:

## Part 2 — Completed fictional example

### What does a completed month look like?

This is synthetic data for Aster & Vale, a fictional agency client. It is not RankEcho customer data, an industry benchmark, or evidence that these movements are typical. The example deliberately includes improvement, no change, a reversal, an unavailable provider view, fewer referral sessions, and one additional key event.

Scope: the fictional report compares 1–31 August 2026 with 1–31 July 2026 for one US-English site, one unchanged 42-cell prompt-and-surface panel, one Search Console property, and one GA4 property. The provider and analytics definitions are held constant; Bing data is unavailable in both periods.

Executive summary — what changed: Google reported 170 more generative Search link impressions, while the fixed panel recorded two more owned-link cells after three gains and one reversal. Identifiable assistant sessions fell by two, and the declared key event increased from three to four. What did not change: the 16-cell comparison subset remained 4/16, and Bing stayed unavailable.

What we shipped: on 18 August, the fictional team corrected one plan-limit definition on /pricing and confirmed visible-page, checkout, and help-center parity; no other panel page changed. What it means: the visibility observations moved in a positive direction, but referral visits did not, the event count is small, and the records do not identify one cause. What happens next: inspect the affected answer sources and pricing-page sessions, preserve the September panel, and do not claim a commercial lift.

| Layer | Prior | Current | Change | Interpretation boundary |
| --- | --- | --- | --- | --- |
| Google generative Search link impressions | 1,240 | 1,410 | +170 (+13.7%) | Provider-defined link impressions; prompts and causes unavailable |
| Bing AI Visibility Insights | Unavailable | Unavailable | No comparison | Property access was not available; excluded, not scored as zero |
| Fixed prompt panel: owned-link observations | 9 / 42 observed cells | 11 / 42 observed cells | +2 cells (+4.8 percentage points) | Three gains and one reversal; same observed-cell panel |
| Identifiable AI-assistant referrals | 31 sessions | 29 sessions | -2 sessions | Attributable visits only; no-click and unattributed activity absent |
| Declared on-site key event | 3 events | 4 events | +1 event | Small descriptive count; no causal or revenue claim |
| Unchanged comparison subset | 4 / 16 owned-link cells | 4 / 16 owned-link cells | No change | Retained as an explicit null result |

### What decision follows from the synthetic month?

The fictional client decision is to retain the corrected pricing fact, investigate why attributable sessions did not move with the visibility observations, and repeat the unchanged panel before proposing another page change. The August pattern is useful enough to guide work, but not enough to attribute an outcome or expand the claim.

| Next action | Owner and due date | Evidence and next observation |
| --- | --- | --- |
| Review the three gained cells, one reversed cell, and the pricing landing-page journey before choosing another fix | Synthetic account strategist · 5 October 2026 | Preserve raw answers and source URLs, Search Console page/device rows, GA4 source/landing-page sessions, and the accepted pricing diff; repeat the fixed panel for 1–30 September 2026 |

## Use this pack

1. Save a copy and complete Part 1 with your own records.
2. Use Part 2 as a teaching example, preserving missing data and unresolved review states.
3. Check sources, comparisons and approvals before sharing your completed version.

Next workflow: [Explore client reporting](https://rankecho.io/product/b2b-command-center).