# Google traffic-drop diagnostic workbook

Prepared and sources reviewed: 8 October 2026.

Use this record to compare a commercial page's measurements, retain evidence and choose the next check. This is an editorial diagnostic procedure, not an official Google recovery framework.

## Download and start

- [Blank CSV](traffic-drop-blank.csv): one instruction row, one empty metric template and one empty decision template.
- [Worked CSV](traffic-drop-worked.csv): one instruction row, nine metric records and three separate decision records.

Save your own copy. CSV files contain static values and no formulas. Editing an input will not update a change value. Use the Markdown template below if a wide spreadsheet is inconvenient. No form, account connection, automated diagnosis or recovery estimate is included.

1. Record the exact property/page, intended buyer job and the first affected date, if known. Preserve original exports before changing filters.
2. Set baseline and observed periods, each source's timezone, comparison timezone, complete-through date and the actual scope. Align filters and date boundaries where possible. Record unresolved differences instead of assuming alignment.
3. Add one metric row per page and measure. Keep Search Console clicks, GA4 sessions, delivered forms and qualified enquiries separate. Use a real zero only when the source establishes zero. Empty numeric/date fields mean unknown until completed.
4. Add a separate decision row for each issue. Link metric IDs and retained evidence, name alternatives, and separate the observation from a proposed action. Record who owns the action and the evidence needed to accept it.
5. Recheck separately after an approved change. A repaired configuration, Google processing a page, later visibility and qualified enquiries are different outcomes.

Rows use record_type values instruction, metric and decision. Filter on that field before analysing. Do not include instruction or decision rows in numeric calculations. On decision rows, numeric fields are not applicable. In an unfinished metric row, empty numeric fields mean unknown. Do not replace either with zero.

For evidence_status, use confirmed, hypothesis or unknown in your own record, with the exact scope of what is confirmed. Worked rows append “in fictional case” because they are invented. For action_status, use proposed, pending, performed or verified, with dated evidence where relevant. A confirmed observation does not make its proposed action complete.

## Blank record to copy

### Comparison context

| Field | Your record |
| --- | --- |
| Business/property and page URL | unknown |
| Intended buyer job and service owner | unknown |
| First affected date and evidence | unknown |
| Baseline dates, inclusive | unknown |
| Observed dates, inclusive | unknown |
| Source timezone and comparison timezone | unknown |
| Reconciliation method or unresolved difference | unknown |
| Country, device, search type and page/query filters | unknown |
| GA4 session source/medium and landing-page filter | unknown |
| Source complete-through date and preliminary markers | unknown |
| Release, consent, analytics, offer or campaign changes | unknown |
| Export/reference and checked-at time with timezone | unknown |
| Enquiry-source attribution and qualification definition | unknown |

### Metric record

| Field | Your record |
| --- | --- |
| Metric ID, page and metric name | unknown |
| Unit and exact source population | unknown |
| Baseline value | unknown |
| Observed value | unknown |
| Absolute change | unavailable until both values exist |
| Percentage change | unavailable until both values exist and baseline is nonzero |
| Source scope, period and completeness limitations | unknown |
| Evidence reference and date checked | unknown |
| What is observed, without choosing a cause | unknown |

### Decision record

| Field | Your record |
| --- | --- |
| Decision ID and linked metric IDs | unknown |
| Page and buyer job affected | unknown |
| Observed issue and evidence reference/time | unknown |
| Evidence status: confirmed, hypothesis or unknown | unknown |
| Alternative explanations and missing checks | unknown |
| Proposed action and responsible owner | unknown |
| Action status | pending |
| Acceptance condition for the technical action | unknown |
| Separate later outcome checks | unknown |
| Next review date or event | unknown |

For an access/indexing issue, preserve the current HTTP/redirect response, intended indexability, robots directives, declared canonical, Google-selected canonical, indexed result, live-test time and last-crawl time. If an exact time was not recorded, say unknown. A live test cannot reconstruct a historical response. Keep Manual Actions and Security Issues as separate checks; an empty current report closes only that report's check.

## Worked example

**Constructed teaching example. All figures, URLs, observations and business details are invented to demonstrate the worksheet. This is not a client case, a test of Google's September update, or evidence of recovery.**

No real account, client, deployment, form or page was inspected. The metric records below and F00 to F06 are constructed teaching inputs. Google documentation supports the method, not these invented observations.

The fictional B2B IT provider uses example.com. Baseline is **10–21 September 2026**, and observed period is **24 September–5 October 2026**, inclusive. Each contains 12 days with identical weekday composition. This is an editorial comparison choice, not a Google-required window. The observed period is inside the September rollout.

F00 assumes data is available through October 5. That cutoff is invented, not a claim about all Search Console properties. For reproducible daily boundaries, F00 assumes **America/Los_Angeles** for all sources and the comparison, with documented reconciliation inside the fictional scenario. No reconciliation was actually performed. Exact extraction times were not supplied and remain unknown.

Search Console is scoped to US, desktop and Web text-based search with the exact page filter. GA4 uses the matching landing page, country/device, session source google and medium organic. Forms and qualified enquiries use all-source receipt-page records, with no supplied country/device restriction. Their acquisition source is unknown. They are not the same population as Google clicks or filtered GA4 sessions.

“Qualified” here means a manually checked business, suitable service requirement and served market. Qualification is not inferred from an analytics event name. No security-audit qualification count was supplied; it remains unknown, not zero.

### Static metrics

Change percentages use the unrounded inputs and are rounded to one decimal here. CSV values retain up to four decimal places.

| ID | Page | Metric | Baseline | Observed | Absolute change | Change |
| --- | --- | --- | ---: | ---: | ---: | ---: |
| M01 | /services/it-support/ | GSC clicks | 200 | 198 | -2 | -1.0% |
| M02 | /services/it-support/ | GSC impressions | 4000 | 4000 | 0 | 0.0% |
| M03 | /services/it-support/ | GA4 Google organic sessions | 170 | 84 | -86 | -50.6% |
| M04 | /services/it-support/ | Delivered forms, all sources | 8 | 7 | -1 | -12.5% |
| M05 | /services/it-support/ | Manually qualified enquiries, all sources | 3 | 3 | 0 | 0.0% |
| M06 | /services/security-audit/ | GSC clicks | 120 | 30 | -90 | -75.0% |
| M07 | /services/security-audit/ | GSC impressions | 2400 | 600 | -1800 | -75.0% |
| M08 | /services/security-audit/ | GA4 Google organic sessions | 102 | 25 | -77 | -75.5% |
| M09 | /services/security-audit/ | Delivered forms, all sources | 4 | 1 | -3 | -75.0% |

M03 represents recorded GA4 sessions under a collection defect in the fictional case. The missing historical sessions are unknown. Its 50.6% decline does not establish the size of the defect's causal effect. Do not sum these page-level impressions into a property-level total.

### Decisions and unfinished actions

#### D01: IT-support measurement defect

- **Evidence:** F05: invented deployment and controlled-check record; M01 and M03. The invented October 8 check finds missing page_view while the form is delivered. A measurement defect is confirmed. GSC clicks fell 1%; recorded GA4 sessions fell 50.6%. The share of missing sessions caused by the defect is unknown.
- **Evidence status:** confirmed_in_fictional_case.
- **Alternatives or missing evidence:** Consent, filters, traffic mix and other changes may also affect the discrepancy. No estimate of lost historical sessions is supported.
- **Proposed action:** Repair analytics configuration, then run a controlled consented journey and verify intended events at their expected trigger points without unintended duplicates. Verify receipt independently.
- **Owner:** Analytics owner (fictional role).
- **Action status:** proposed_not_performed.
- **Acceptance:** Pending: intended events recorded at their expected trigger points without unintended duplicates and form receipt verified. Review future data separately; do not declare historical sessions restored.
- **Next review:** pending: after the proposed repair and controlled check; no calendar date set.

#### D02: Security-audit indexing defect

- **Evidence:** F06: invented noindex deployment and indexed/live inspection record; M06 and M07. The invented deployment added unintended noindex. The indexed report shows exclusion; live inspection confirms the current noindex directive. Clicks and impressions fell 75%. The defect and search decline are confirmed; sole-cause attribution is not.
- **Evidence status:** confirmed_in_fictional_case.
- **Alternatives or missing evidence:** Other ranking, demand, query-mix and deployment changes remain possible. Last-crawl timestamp and exact inspection times were not supplied.
- **Proposed action:** Developer removes the unintended directive, checks intended canonical and access, and retains evidence. Check later indexed state separately.
- **Owner:** Developer (fictional role).
- **Action status:** proposed_not_performed.
- **Acceptance:** Pending technical repair: intended response, indexability and canonical checks pass with dated evidence. Pending separate indexed-state check: Google has processed the intended page. Visibility recovery remains unproven.
- **Next review:** pending: after proposed repair, then separately when indexed information is available.

#### D03: Remaining update hypothesis

- **Evidence:** M01 to M09; F05; F06; S08 supplies real rollout timing only. These records do not establish that the September spam update caused either decline. A complete post-rollout comparison is not represented.
- **Evidence status:** hypothesis_unresolved.
- **Alternatives or missing evidence:** Concurrent measurement and indexability defects are already present. Seasonality, buyer demand, query mix and other changes remain untested.
- **Proposed action:** Retain the update hypothesis as unresolved. Compare subsequent complete periods and review a specific suspected practice against policy only when evidence supports that review.
- **Owner:** SEO owner (fictional role).
- **Action status:** proposed_not_performed.
- **Acceptance:** Pending: comparable complete periods, retained source scope, and explicit evidence for or against each hypothesis. No mass rewrite, disavow or recovery guarantee.
- **Next review:** pending: select later complete comparable periods; no recovery deadline set.

### Constructed evidence references

| ID | What the fictional record represents | Boundary |
| --- | --- | --- |
| F00 | Period/scope/completeness and timezone assumptions above | An invented case setup, not a source export or live availability claim. |
| F01 | Supplied Search Console clicks/impressions for the two pages | Invented totals, not a verified property or query-cohort history. |
| F02 | Supplied GA4 Google organic session totals for the two pages | Invented totals; M03 has missing collection in the scenario. |
| F03 | Form receipt counts by page, from all sources | No traffic-source attribution. Exact filtering/deduplication details were not supplied. |
| F04 | Three manually qualified IT-support enquiries in each period | Fictional qualification as defined above; no sale or revenue result. |
| F05 | September 28 tag deployment and October 8 consented check: missing page_view, form delivered | Invented check. Exact test time is unknown. Proposed repair has not occurred. |
| F06 | September 25 unintended noindex deployment; October 8 indexed exclusion and live noindex observation | Invented inspection. Exact inspection time and last-crawl timestamp are unknown. Proposed repair and later reindexing remain pending. |

## Calculations and boundaries

Absolute change is observed value minus baseline value. Percentage change is (observed minus baseline) divided by baseline, multiplied by 100. If either input is missing or baseline is zero, percentage change is **unavailable**. A measured zero observed value with a nonzero baseline is valid and may represent a 100% decline. Never coerce unknown to zero.

CTR is clicks divided by impressions, multiplied by 100. Use matching page/query population, date range, device, country and search type. If clicks or impressions are missing, or impressions are zero, CTR is **unavailable**. For the fictional IT-support page, CTR is 5.00% then 4.95%, a decrease of 0.05 percentage points. For security audit, CTR is 5.00% in both periods. Unchanged aggregate CTR does not establish unchanged ranking or demand.

The IT-support click change is (198 minus 200) divided by 200 = -1%. Its session change is (84 minus 170) divided by 170 = about -50.6%. Security-audit clicks and impressions each decline 75%; sessions decline about 75.5%. Counts are small for forms and enquiries, so read the absolute figures alongside the percentages.

Do not divide all-source forms or qualified enquiries by Google clicks or filtered GA4 sessions and call it an organic conversion rate. Do not subtract GA4 sessions from GSC clicks to calculate “lost visitors.” They measure different things. No result here establishes an update penalty, a single cause or a recovery outcome.

As of the October 8 source review, Google's incident record states the September spam update began September 24 at 09:15 PDT and completed October 8 at 01:00 PDT. These dates supply context only. A full post-rollout week does not exist in this example. Google's recommendation to wait a full week after a **core update** completes is not a stated spam-update waiting rule, and it does not prevent repair of an independently confirmed defect.

## Method sources

All method sources below were reviewed in the accompanying research on **8 October 2026**. Recheck current documentation before acting. F00 to F06 supply the invented example only; no Google source endorses its invented figures or proves recovery.

- **S01: [Google traffic-drop guidance](https://developers.google.com/search/docs/monitor-debug/debugging-search-traffic-drops).** Investigate several causes. Timing alone does not establish cause.
- **S02: [Search Console Performance report](https://support.google.com/webmasters/answer/7576553).** Keep clicks, impressions, CTR and position distinct. Preserve missing values.
- **S03: [Comparing Analytics and Search Console](https://developers.google.com/search/docs/monitor-debug/google-analytics-search-console).** Align scopes and reporting periods. Clicks and sessions need not be equal.
- **S04: [About Search Console data](https://support.google.com/webmasters/answer/96568).** Standard daily reporting uses Pacific Time. Check availability and privacy limits.
- **S05: [URL Inspection](https://support.google.com/webmasters/answer/9012289).** Indexed and live states are separate. A successful live test does not prove indexing.
- **S06: [GA4 DebugView](https://support.google.com/analytics/answer/7201382).** Inspect a controlled consented event journey and verify form receipt separately.
- **S07: [GA4 recommended events](https://developers.google.com/analytics/devguides/collection/ga4/reference/events).** A lead event, qualification and a closed sale are separate events.
- **S08: [September 2026 spam-update incident](https://status.search.google.com/incidents/XhUDXP7A67iHCD2kmbVu).** Rollout dates provide context, not a diagnosis for this invented case.
- **S09: [Google spam-update guidance](https://developers.google.com/search/docs/appearance/spam-updates).** Compliance learning can take months. No recovery deadline or guarantee.
- **S10: [Google core-update guidance](https://developers.google.com/search/docs/appearance/core-updates).** The full-week waiting recommendation is specific to core updates.

Review separate current reports when relevant: [Manual Actions](https://support.google.com/webmasters/answer/9044175), [Security Issues](https://support.google.com/webmasters/answer/9044101), [page indexing](https://support.google.com/webmasters/answer/7440203), [spam policies](https://developers.google.com/search/docs/essentials/spam-policies), [query filters and dimensions](https://support.google.com/webmasters/answer/17011259), and [Search Console data anomalies](https://support.google.com/webmasters/answer/6211453). Their presence here is not evidence that a live account or every current reporting incident was inspected.