Guías
How to Do SEO Analysis: From Data to a Prioritized Decision
SEO analysis is the process of testing an explanation against several datasets—not exporting every metric into one report.
Ejecuta una auditoría nueva en DomainLens y usa el informe como lista de prioridades.
Begin with a decision, not a dashboard
Write the question precisely: Why did non-brand product clicks fall after May 12? Which unindexed guides should be improved or removed? Which category template change can recover the most revenue? A bounded question determines the dates, segments, URLs, and evidence you need.
Use each dataset for what it knows
| Source | Reliable evidence | Blind spot |
|---|---|---|
| Search Console | Google queries, pages, clicks, impressions, average position, index states | Conversions and all visits |
| Analytics | Sessions, behaviour, conversions, revenue with configured tracking | Google crawl/index decisions |
| Crawler/DomainLens | Current technical response and internal site graph | Historical Google demand |
| Server logs | Actual crawler requests and response behaviour | Search intent and ranking |
| SERP/manual review | Intent, result formats, competing usefulness | Site-wide technical scale |
Normalize the comparison before explaining it
- Compare equivalent date ranges and weekdays; annotate holidays, seasonality, campaigns, and outages.
- Separate brand/non-brand, country, device, search type, directory, and page template.
- Account for tracking changes, consent, URL migrations, and property configuration.
- Use totals and distributions; an average can hide a collapsed template and a growing one.
- Keep correlation language until a release, URL sample, or diagnostic evidence supports causation.
A worked analysis: impressions rise while clicks fall
- 1Segment by query and page. The loss comes from non-brand mobile queries on /guides/, not the whole site.
- 2Check position and CTR. Average position is stable, but CTR fell on queries where a new rich-result format appears.
- 3Review the live SERPs and snippets. Several guide titles are generic and Google rewrites descriptions from repetitive intros.
- 4Audit the pages. They are indexable and canonical, but their openings do not answer the query and structured data does not create the missing result format.
- 5Prioritize pages with high impressions and relevant intent. Rewrite titles/openings for clarity and add only eligible, visible structured data.
- 6Annotate deployment and compare the same query/page cohort after enough data accumulates.
Score actions by expected value and confidence
| Candidate action | Impact | Confidence | Effort | Decision |
|---|---|---|---|---|
| Remove accidental noindex from product template | Very high | Very high | Low | Immediate |
| Rewrite 12 high-impression guides | High | High | Medium | Next content batch |
| Change every URL to include keywords | Uncertain/negative risk | Low | Very high | Reject |
| Optimize one lab score from 97 to 100 | Low | Low | Medium | Defer |
Use DomainLens to test the technical hypothesis
When the data points to a URL cohort, run DomainLens on representative pages to verify current status, indexability, canonical, content signals, schema, links, and performance. If the hypothesis concerns Google’s chosen canonical or indexed rendering, confirm it in Search Console. If it concerns crawl frequency, use logs.
Finish with a short decision record: question, datasets and dates, observed segment, likely cause, alternative explanations, selected action, owner, deployment date, validation check, and review date. That record makes the analysis reusable instead of becoming another slide deck.
When the evidence becomes a site-wide remediation project, organize it with the technical SEO audit checklist .