# GSC / GA4 Monthly SEO Review Template

- Version: 1.1.0
- Released: 2026-08-10
- Source: eigentime's original SEO delivery and review method


> **Evidence level of what is new in this version.** The four blocks 1.1 adds — the three report layers, observation windows, confidence tags with previous-period settlement, and signal-based stage changes — **come from a method that has not been run on a client project**. Source: [SEO monthly reports: what earns a number its place](https://eigentime.org/zh/blog/seo-monthly-report-metric-cuts/) (Chinese). Nothing from 1.0 changed, and its evidence level is unaffected. New material is marked `[UNVERIFIED]` inline. Treat it as a suggestion, not a standard.

## Purpose

This template turns a monthly SEO report into a decision record instead of a slide deck full of exported numbers. Every finding must distinguish observed facts, a plausible mechanism, the people who can act on it, what the evidence cannot establish, and the next verification step.

Use the Markdown document for the narrative and decisions. Use the companion CSV as an action register with one row per finding, owner, due date, evidence link, and review status.

## 0. Report metadata

| Field | Record |
|---|---|
| Reporting window | YYYY-MM-DD to YYYY-MM-DD |
| Comparison window | Month over month, year over year, or a named baseline; explain why |
| Site and market | Domain, country or region, language, and product line |
| Data sources | GSC property, GA4 property, and any verified business source |
| Data cutoff | Export timestamp; exclude incomplete days unless disclosed |
| Definition changes | Filters, events, channel groups, consent, migrations, or site changes |
| Ownership | Report author, data reviewer, and business approver |


## 0.5 A report is three layers, not one [UNVERIFIED]

Separate the layers before writing, or one ruler will damage the wrong thing. Readers and acceptance criteria differ per layer:

| Layer | Answers | Read by | Filter |
|---|---|---|---|
| Delivery | What we did this month | The payer | Not subject to the filter — it is contractual evidence |
| Diagnosis | What happened on the site, and why | Delivery and decision side | "It changed — whose action differs?" |
| Decision | What to do next month, and why this | Decision side | Every line must land on a concrete action |

A delivery list (articles shipped, issues fixed) belongs to the delivery layer. **Do not filter it as if it were a metric.** The filter governs the other two layers only.

The filter itself: a metric earns a place in the body only if it answers "it changed — what does the person reading this report do differently next month?" If it cannot, move it to the appendix — **move, not delete**. When the appendix is incomplete, cutting metrics is not information design, it is reduced disclosure.

## 0.6 How long a window does each metric need [UNVERIFIED]

Single-period movement is mostly noise, especially on small bases (qualified inquiries going 1 → 3 is +200% and statistically nothing). So state an observation window for every metric kept in the body. Window length follows the metric's noise, not the fact that the report is called monthly.

| Metric | Suggested window | Why |
|---|---|---|
| Qualified inquiries | 3-month rolling | Monthly counts are often single digits; a month-over-month delta is unreadable |
| Priority keyword positions | 3 months | Daily volatility is high; a monthly average is already smoothing, so read direction |
| Query-class coverage | 1 month | On a new site direction shifts fast enough to show monthly |
| Indexing status | 6 months | Moves slowly; monthly views hide structural problems |

Every trend chart carries two things: a comparison baseline, and how the start point was chosen. **Pick a trough as the start and any metric becomes a rise.** Both belong in the appendix, with a pointer from the body.

## 1. One-page conclusion

Lead with no more than five decisions:

1. **Most important observed fact**: only what the data directly supports.
2. **Most likely mechanism**: explain the change and label inference separately from proven causation.
3. **Who or what is affected**: markets, query groups, page types, or business stages.
4. **What cannot be concluded**: state attribution limits and missing evidence.
5. **Next action**: include an owner, due date, evidence, and verification method.


## 1.5 Confidence tags, and settling the previous period [UNVERIFIED]

Tag every judgement in the diagnosis and decision layers so the claim and its risk are delivered together:

- **[CONFIRMED]** Directly supported by data; safe to state externally.
- **[INFERRED]** The most likely explanation so far, alternatives not excluded, to be verified next period.
- **[UNKNOWN]** Not known yet; here is what was done this period to make it knowable.

The value is organisational: whoever presents the report can read an [INFERRED] line out as written without personally vouching for it, and if it is falsified a period later, the reconciliation is "this was tagged inferred".

**Tagging without settling is just a disclaimer.** So the first item in each report settles the previous period's [INFERRED] and [UNKNOWN] lines, one result each: confirmed, falsified, or still unknown — and a falsified line states the new explanation.

| Previous judgement | Tag then | Result now | New explanation / next step |
|---|---|---|---|
|  | Confirmed / Inferred / Unknown | Confirmed / Falsified / Still unknown |  |

## 1.6 Change stage on signals, not on month count [UNVERIFIED]

Different stages deserve different metrics. Define the stage by signal, not by months since launch:

- First qualified inquiry attributable to a landing page → early-traction stage;
- Qualified inquiries non-zero for three consecutive months → steady stage;
- **Steady-stage site with three consecutive zero months → back to early traction.** Downgrades matter more than upgrades: when a site is in trouble is exactly when the report should return to structural metrics.

Seasonal industries are the exception: compare against the same month last year, or the site will change stage twice a year for no reason.

| Stage | Keep in the body | Move to the appendix |
|---|---|---|
| New / zero baseline | Query classes covered by impressions (not the absolute number), target pages indexed, first entries into the top 20 | Inquiries, conversion rate, CTR trend |
| Early traction | Inquiries by landing page, top-10 keywords and their landing pages, where clicks concentrate | Site-wide average position, total impressions |
| Steady | Change in inquiry-source pages, priority pages dropping out of the top 10, competitor movement on priority terms | Index count, impressions, average position |

## 2. KPI snapshot

| Source | Metric | Current | Comparison | Change | Interpretation limit |
|---|---:|---:|---:|---:|---|
| GSC | Clicks |  |  |  | Search clicks are not sessions or leads |
| GSC | Impressions |  |  |  | Growth may come from new queries or different ranking ranges |
| GSC | CTR |  |  |  | Interpret with query, position, and device mix |
| GSC | Average position |  |  |  | An aggregate mean is not the position of an individual query |
| GA4 | Organic Search sessions |  |  |  | Depends on channel definitions, consent, and cross-domain setup |
| GA4 | Engaged sessions / rate |  |  |  | Does not directly prove content or commercial quality |
| GA4 | Key events / conversions |  |  |  | Verify that event definitions and triggers stayed stable |

Rules:

- Keep absolute values beside percentage changes; large growth from a tiny base is not a large outcome.
- Use comparable periods and disclose holidays, seasonality, paid campaigns, migrations, and tracking changes.
- GSC and GA4 measure different objects. Their totals are not expected to match.
- Do not translate traffic movement into revenue or lead impact without a reliable business-data link.

## 3. Segment before explaining

Site-wide totals locate a change; they rarely diagnose it. At minimum, review:

### GSC

- Branded and non-branded queries.
- Country or region, device, and search type.
- Query themes and intent groups.
- Home, category, product, article, and other page types.
- New or changed pages against an untouched baseline group.

### GA4

- The exact Organic Search channel definition.
- Landing pages and page groups.
- Country or region and device.
- New and returning users only when definitions remain stable.
- Verified key events, inquiry paths, or other business actions.

## 4. Ask four questions for every finding

### What fact was observed?

Write a reproducible statement such as: “Non-branded query clicks increased 18% month over month, concentrated on three product pages.” Avoid labels such as “SEO improved.”

### What mechanism might it reveal?

State the leading explanation and its evidence level. For example, new specification fields may have expanded long-tail coverage, but seasonality and ordinary ranking movement still need to be ruled out.

### Who can use this mechanism?

Name the decision owner: content may expand comparable pages, product may supply missing specifications, or engineering may repair a template issue.

### What does it not prove?

State the boundary. More clicks do not prove that a redesign caused more inquiries; a lower aggregate position does not prove that every priority query declined.

## 5. Diagnostic order

Investigate in this order so the narrative follows the evidence:

1. **Validate definitions**: properties, filters, time zones, events, and channel groups.
2. **Locate the change**: site, market, device, query group, or a small set of pages.
3. **Check external factors**: seasonality, news, algorithm changes, campaigns, and competitors.
4. **Map project actions**: content launches, template changes, technical fixes, and internal links.
5. **Look for counterexamples**: did unchanged pages move too, and what evidence would falsify the explanation?
6. **Define verification**: obtain missing data, inspect samples, recrawl, or wait for the next complete period.

## 6. Action register

| Finding | Evidence | Mechanism hypothesis | Cannot conclude | Next action | Owner | Due | Review status |
|---|---|---|---|---|---|---|---|
|  |  |  |  |  |  |  | Planned / Review due / Verified / Closed |

Every action must be closable. Replace “keep monitoring” with the metric, end date, and result that will trigger a decision.

## 7. Pre-publish QA

- [ ] Reporting and comparison windows are complete and comparable.
- [ ] GSC and GA4 metrics, properties, and filters are explicit.
- [ ] Every percentage can be traced to absolute values.
- [ ] Facts, inferences, and unverified hypotheses are labeled separately.
- [ ] Correlation is not presented as causation.
- [ ] Brand, non-brand, and priority markets are not hidden by site-wide totals.
- [ ] Client, team, and individual ownership are represented accurately.
- [ ] Every recommendation has an owner, due date, and verification method.
- [ ] Screenshots and exports contain no customer, user, or inquiry data that should remain private.

## License and limits

This template is licensed under [CC BY 4.0](https://creativecommons.org/licenses/by/4.0/). You may copy, adapt, and use it commercially with appropriate attribution and an indication of changes.

Suggested attribution: `CG-X / eigentime.org — GSC / GA4 Monthly SEO Review Template v1.1.0`.

This template is not a ranking, traffic, revenue, legal, or compliance guarantee. Validate GSC, GA4, and business-system definitions, sampling, consent, time zones, and data delays for each project.