loggie
Start free

Google Search Console · Workflow guide

Google Search Console content refresh queue with AI

Have your AI agent compare two Search Console periods through Loggie and build a content refresh queue an editor can act on, with no site changes.

Who this is for: Content leads and SEO operators who want a repeatable way to find useful updates in an existing site.

Before you start

  • A Search Console property verified for the connected Google account and available through Loggie.
  • An agent with read access to search analytics for that property. Publishing changes to your site is a separate permission.
  • Two matched reporting periods, a target country or device scope if relevant, and a minimum impression threshold.

In your Loggie workspace, connect Google Search Console, create an agent identity, and review its access. On the agent’s machine, use Node.js 20 or later and initialize the CLI:

npm install -g @loggie-ai/cli
loggie init
loggie discover

Enter the agent’s Loggie key through the initialization prompt. Keep provider credentials in the connection setup, not in the agent prompt. Use the connection slug returned by discovery in the examples below.

Step-by-step workflow

  1. Start with existing demand

    A content refresh queue works best when it starts from pages already getting relevant impressions. Compare two completed 28-day periods, ending far enough in the past for the data you need to be available. Record whether you requested final or preliminary data.

    Prioritize meaningful lost clicks, high-impression pages with a plausible intent mismatch, and queries close to the topic the page should answer. Keep brand and non-brand queries separate when they would lead to different editorial decisions.

  2. Query the property with consistent dimensions

    Connect Search Console in Loggie and initialize the agent’s CLI. Find the connection slug with loggie discover. Replace the example slug if needed. The siteUrl path parameter must be URL-encoded; sc-domain%3Aexample.com is an illustrative domain property.

    Replace START_DATE and END_DATE with YYYY-MM-DD dates before running. Fetch the same dimensions and search type for both periods. Increase startRow for additional pages, while remembering that Search Console returns top rows and does not guarantee every query-page combination.

    loggie discover google-search-console --method POST --search searchAnalytics
loggie call google-search-console POST /webmasters/v3/sites/sc-domain%3Aexample.com/searchAnalytics/query --body '{"startDate":"START_DATE","endDate":"END_DATE","dimensions":["page","query"],"type":"web","dataState":"final","rowLimit":1000,"startRow":0}' --pretty

  3. Turn movement into an editorial question

    Join results by page and query, then compare clicks, impressions, CTR, and average position. A CTR decline while average position also falls is a different problem from stable position with a weaker search snippet. Neither proves that a title rewrite will recover clicks.

    For the highest-impact candidates, inspect the public page and the current search intent. Create a specific brief: missing explanation, outdated detail, unclear title, or a mismatch between the page and the query. Link the source page and keep observations separate from proposed edits.

  4. Review one refresh before scaling up

    Give an editor a short queue with the target query, period values, proposed change, and a way to evaluate it. Choose one page, record the publication date, and reassess after an appropriate interval rather than expecting an immediate ranking change.

    The agent can prepare a Markdown brief in your existing workflow. Loggie handles the Search Console access; your agent’s scheduler and publishing tools handle recurring runs and any later approved site changes.

Google Search Console agent prompt

Adapt this prompt to your connected account and the scope you have approved.

Compare the two agreed 28-day Search Console periods. Keep property, search type, country, device scope, and dimensions consistent. Rank five existing pages worth reviewing by meaningful opportunity, not percentage change alone. Show both period values, a source page, one editorial question, and a proposed refresh. Do not publish changes or promise ranking gains.

Example Google Search Console report

Example output with invented data.

Refresh queue: last 28 days vs prior 28 (web, US, non-brand)

1. /guides/api-rate-limits
   Clicks 240 → 180 · Impressions 8,000 → 9,000
   CTR 3.0% → 2.0% · Position 7.1 → 7.0
   Position held and CTR fell: the snippet may no longer fit the query.
   Ask: do the title and opening answer "api rate limit best practices"?
2. /blog/webhook-retries
   Position 4.2 → 8.9 for "webhook retry strategy"
   Ask: is the page missing a worked backoff example?

No pages edited. An editor reviews before rewriting.

Success checks

  • Reconcile a sample with Search Console using matching property, dates, dimensions, and filters.
  • Verify that pagination was attempted and describe the top-row coverage limit in the report.
  • Have an editor confirm that each proposed refresh addresses the actual page and query.

Pitfalls

  • Anonymized queries and top-row limits prevent the table from being a complete record of all searches.
  • Average position is an aggregate, not a fixed ranking for every searcher.
  • Seasonality and search-result changes can explain movement without a content defect.

Sources