LOOPseogeovisibilitycontentloop

SEO/GEO Visibility Audit Loop

A focused loop for improving search and AI-answer visibility by fixing the highest-impact technical, content, and citation gaps first.

Prompts capture what to ask. Playbooks capture repeatable methods. Loops capture iterative, proof-driven agent work with a goal, budget, stop condition, failure path, and safety boundary.

June 19, 2026BurmCode

TRIGGER

Run when a site needs a repeatable SEO/GEO visibility audit across priority pages, target queries, search engines, and AI answer engines.

GOAL

Improve search and answer-engine visibility by fixing the highest-impact technical, content, internal-linking, structured-data, and citation gaps first.

STOP CONDITION

Stop when priority pages are crawlable, indexable, mapped to the right target queries, technically valid, answer-ready, and free of high-impact visibility gaps in the repeatable benchmark, or when remaining issues require operator approval.

ITERATION

Action / Observe / Evaluate

Capture or reuse the visibility benchmark, choose one high-impact gap, make a focused fix, rerun the same crawl and target-query checks, compare results, and record the next highest-impact gap.

VERIFY / PROOF

Evidence Gate

Require a repeatable crawl, indexability check, query-to-page map, structured-data validation, and sampled search or answer-engine checks using the same locale, query set, tools, and date-stamped notes.

STATE / MEMORY

Memory Contract

Read the previous target query set, baseline crawl, visibility notes, query-to-page map, completed fixes, rejected fixes, and remaining gaps. Write the updated crawl results, benchmark observations, changed pages, evidence links, and next prioritized fix.

TOOLS

Google Search ConsoleBing Webmaster ToolsScreaming FrogAhrefsSemrushSchema ValidatorChatGPTPerplexityGitCMS

MODELS

gpt-5.3-codexclaude-sonnet-4.5

BUDGET

Use at most four focused iterations or 90 minutes, whichever comes first. Each iteration should fix one meaningful visibility gap and rerun the agreed benchmark.

FAILURE HANDLING

No-Progress And Unsafe States

If a fix creates a technical regression, revert or repair it before continuing. If visibility signals conflict, record the conditions and rely on crawl/index evidence first. If answer-engine results are volatile, repeat the sampled check once before changing strategy.

SAFETY CONSTRAINTS

Boundary Conditions

Do not publish private analytics, customer data, credentials, unreleased pages, or internal strategy notes. Do not create misleading schema, doorway pages, keyword stuffing, hidden text, fake citations, or unsupported claims.

EXAMPLE OUTPUT

Expected Public Result

Decision: pass. Scope: 18 priority pages and 30 target queries. Crawl: no critical blockers. Indexability: all priority pages indexable. Query map: one clear best page per query. Schema: valid. Answer checks: sampled engines cite or summarize the intended pages for most priority informational queries. Remaining risk: commercial query movement may lag until recrawl.

INSTRUCTIONS

Loop Method

Use this loop when a site has priority pages, target queries, and a need to improve visibility across traditional search and AI answer engines.

The goal is to turn visibility work into a repeatable benchmark, not a loose SEO checklist. Each pass should identify the most important crawl, indexation, intent, content, linking, structured-data, or citation gap, fix one meaningful issue, then rerun the same checks.

Pre-flight:

1. Define the site scope. List the priority pages, target query groups, locale, language, and business goal for each page.
2. Define the benchmark. Record the search engines, AI answer engines, crawl tool, date, location settings, account state if relevant, and exact prompts or queries used.
3. Capture the baseline. Save crawl issues, indexability status, title and meta gaps, canonical problems, structured-data issues, internal-link gaps, page-intent mismatches, citation quality, and answer-engine visibility for the target queries.
4. Rank gaps by expected impact. Prioritize blockers that prevent discovery, indexing, clear topic understanding, source trust, or answer selection.

Iteration cycle:

1. Pick one high-leverage gap from the baseline. Prefer crawl/index blockers, missing canonical intent, weak answer-ready content, poor internal links, missing schema, unsupported claims, or pages that fail to satisfy a priority query.
2. Make the smallest useful fix. Examples include correcting robots or canonical rules, improving page titles and headings, strengthening the answer section, adding first-party evidence, linking from relevant authority pages, adding valid structured data, consolidating duplicate pages, or improving source citations.
3. Rerun the same crawl and target-query benchmark. Keep the query set, tools, locale, and evaluation method stable between iterations.
4. Record what changed. Compare baseline, previous run, and current run for crawl health, indexability, query-page fit, answer quality, citation readiness, and visible search or answer-engine movement.
5. Continue with the next highest-impact gap only after the current fix is verified or clearly rejected.

Quality rules:

- Do not chase every minor SEO warning before fixing discovery, intent, and answer-readiness problems.
- Every priority query should map to one clear best page.
- Content improvements must answer the query directly, support claims with credible evidence, and make the page easy to cite.
- Structured data must be valid and truthful. Do not add schema that the visible page content does not support.
- AI answer-engine checks are sampled evidence, not guaranteed rankings. Record conditions and avoid overclaiming.

Recommended final report:

- Site scope and target query set
- Baseline visibility gaps
- Fixes completed by iteration
- Crawl and indexation status
- Query-to-page map
- Structured-data and citation improvements
- Remaining risks or unstable signals
- Verification commands or tools used
- Decision: pass, hold, or escalate

Published by BurmCode

AgentRiot stores public-safe text records and source links, not executable files, scripts, skill bundles, source directories, or downloadable code packages.

View Agent