A good URL is stable, descriptive enough to understand and consistent with the site’s information architecture. It should serve users first and avoid unnecessary complexity.
What matters now
Google’s current SEO guidance focuses on helping search engines discover and understand useful pages while keeping the experience clear for people; it does not prescribe a magic keyword formula. Google treats canonicalization as a way to select a representative URL among duplicates or near-duplicates; redirects, rel=canonical, sitemaps and internal links should point in the same direction. Google recommends crawlable HTML links with descriptive anchor text because links help both discovery and understanding of related pages. The useful question is therefore not whether one isolated tactic can force a ranking, but whether the page becomes clearer, more useful and easier to discover for the searches it is meant to serve.
A practical workflow
- Use readable words when they help identify the page topic.
- Keep one preferred URL for each canonical piece of content.
- Redirect retired URLs when a clear replacement exists rather than leaving chains of unnecessary redirects.
- Avoid exposing endless parameter combinations as separate indexable pages when they do not add distinct value.
- Link internally to the canonical version consistently.
Example in practice
On a real service or reference page, these decisions should work together: the title sets a clear expectation, the main heading confirms the topic, the visible copy answers the need, internal links provide useful next steps, and metadata accurately summarizes the page. If one element says something different from the others, fix the inconsistency rather than trying to compensate with more keywords.
What to avoid
Avoid changing established URLs merely to insert keywords. URL migrations create risk and should have a clear user or architecture benefit.
How to measure and review
Monitor indexing, canonical selection and traffic after structural URL changes.
Keep the strategy evidence-led
Search behavior changes over time, so treat planning assumptions as hypotheses. Recheck first-party performance data, current search results and the needs of the audience before making large content or taxonomy changes. Where a tool metric and real performance disagree, investigate the difference instead of forcing them into one score.