English Google Webmaster Central Office Hours Hangout: How Can You Improve SEO?

TL;DR

Improve SEO by preserving useful content, strengthening genuinely thin sections, keeping image URLs stable, and validating markup against the requirements of the intended rich result. John Mueller explains that low Google traffic alone does not make a page bad, and even a 10-year-old article may retain unique value. Read on for specific guidance about content cleanup, image indexing, structured data, cloaking, pagination, crawling, ads, and pop-ups.

Transcript

JOHN MUELLER: All right, welcome, everyone, to today's Webmaster Central Office Hours Hangout. My name is John Mueller. I'm a webmaster trends analyst here at Google in Switzerland. And part of what we do are these office hours Hangouts, where webmasters and publishers can jump in and ask their questions around the web search. I don't know, do you ... Read More

Key Insights

• Traffic is not quality: A page receiving little Google traffic is not automatically bad. Its subject may be reasonable and its information useful, while search demand remains low. Using traffic alone as the deletion criterion would therefore confuse limited interest with deficient content. Mueller directs attention toward the actual substance of a page and the broader quality pattern across the site.
• Sitewide proportion changes risk: A small amount of thin material on a gigantic website is usually not presented as critical. Concern increases when a significant portion of the website is genuinely thin and low quality. The relevant assessment is therefore not simply whether any weak pages exist, but whether they form a substantial part of what the site offers.
• Old articles can stay valuable: A news archive naturally contains unusual articles and retained weather reports. Someone researching the past 10 years later may appreciate finding a short article if it contains unique information unavailable elsewhere on the web. Age, limited reach, and thinness do not by themselves eliminate a page’s value to a specific future user.
• Improvement can replace deletion: Once weak areas are identified, publishers can add information to individual pages or combine multiple thin pages into one. Both options aim to create a more substantial result without automatically discarding useful material. This approach follows Mueller’s distinction between content needing improvement and content that merely receives little search traffic.
• Publishing processes matter long term: Even when immediate cleanup is not critical, the website can reconsider how content will be maintained. Mueller suggests implementing processes that improve lower-quality material before it is published. That shifts some effort from repairing a large archive toward controlling the quality of new additions, while preserving older pages that may still have unique value.
• Unused source images offer little: Keeping an original source image at a stable public URL does not solve the indexing issue if that image is not actually used on the page. An image sitemap pointing to it probably will not accomplish much. Google still encounters the changing variant URLs associated with the displayed images that could appear in image search.
• Image changes delay processing: Google is accustomed to images changing less frequently than ordinary page references, so it does not recall them as often. When a new HTML page points to a new image file, processing can take time. Regular URL changes therefore make proper image-search indexing harder, even when the visible image is simply another CMS-generated variant.
• Redirects are the fallback: When image URLs must change, Mueller recommends at least setting up redirects from the old URLs. His preferred solution is to prevent regular URL changes altogether. An occasional change after an editor adjusts display size may not be a major problem, but daily or session-based regeneration would create much greater indexing difficulty.
• Image search value determines priority: Stable image URLs matter most when people might search visually for the site’s content. If the publisher does not care about image search and is confident that users are not looking for the material visually, the problem may not be critical. Mueller leaves that prioritization to the website rather than treating it as universal.
• Validity differs from eligibility: Structured data can be technically valid because it contains the baseline attributes and values, yet still be insufficient for a particular rich result. Different rich-result types can demand more or fewer properties. This explains why validation surfaces may appear contradictory even when none is necessarily evaluating exactly the same eligibility requirements.
• Personalization is not automatically cloaking: The problematic case is showing Googlebot content that users would not see. Personalizing what users receive according to location is acceptable under the existing guidance, provided Googlebot is not given a uniquely privileged version. The distinction depends on whether crawler treatment departs from the content available to corresponding users.
• Crawlability validates pagination: A pagination system should allow crawlers to discover all product pages, not merely the first group displayed. Screaming Frog can simulate crawling to verify that accessibility. Filtering and sorting options should also be controlled because they can generate unnecessary URLs. The practical test is whether the crawl can reach the complete product set.

Questions & Answers

Q: How should a large website handle old thin content?

Do not delete old pages simply because they receive little traffic from Google. First determine whether they are genuinely low quality or merely cover subjects that people rarely search for. If thin sections are identifiable, add useful information or combine related pages into a stronger page. This preserves unique material while addressing the significant sitewide patterns that Mueller says are more concerning.

Q: Should low-traffic articles be deleted for SEO?

Low traffic by itself is not a sufficient reason to delete an article. A page can be reasonable and useful even when few people search for its subject. Old news articles or weather reports may help someone researching the topic 10 years later, especially when the information is unique. Evaluate content quality and usefulness directly because traffic does not establish that a page is bad.

Q: How can publishers prevent future thin content?

Publishers can examine how they intend to maintain content over the long term. Mueller suggests creating processes that improve lower-quality material before it is published. This reduces the need to revisit a large archive after weak pages have accumulated. It also complements selective expansion or consolidation of thin pages that are already on the site.

Q: What should a site do when image URLs change?

Set up redirects from the old image URLs to the new ones. Ideally, change the CMS behavior so displayed image URLs do not change regularly when variants are produced. Google does not recall images very frequently, so discovering and processing a new file can take time. Stable URLs therefore make proper image-search indexing easier than daily or session-based regeneration.

Q: Can an image sitemap identify an unused source image?

An image sitemap probably will not do much for a source image that is not used on the page. Google still has to process the displayed image variants whose URLs are changing. The stable source URL does not remove the problem affecting the images that could be shown in image search. Use redirects for changed URLs and, where possible, keep the displayed image URLs stable.

Q: Why do structured data tools show different errors?

Different results commonly arise from requirements attached to specific rich-result types. Markup may contain valid baseline attributes and values but still omit additional properties needed for the particular result being targeted. Identify the search presentation you want, then work backward from its requirements. Check Google’s developer site to confirm that every required attribute for that result is present.

Q: Is location-based personalization considered cloaking?

Location-based personalization is acceptable under the guidance provided. The problem arises when Googlebot receives content that users would not see. A site should therefore ensure that the crawler is not served a special version unavailable to corresponding users. The distinction rests on unequal crawler treatment, not on personalization itself.

Q: How should pagination be checked for SEO?

Pagination should make all product pages discoverable to crawlers. Use a tool such as Screaming Frog to simulate crawling and verify that it can reach the complete set of pages. Review filtering and sorting options because they can produce unnecessary URLs. This check matters because products hidden from the crawl are not easily discoverable through the pagination structure.

Summary & Key Takeaways

• Introducing Google office hours: John Mueller, a webmaster trends analyst at Google in Switzerland, opens the Webmaster Central Office Hours Hangout as a forum where webmasters and publishers can ask questions about web search. The first submitted case concerns a news website more than 10 years old with many articles. Its owners estimate that roughly 10% could be considered thin and difficult for users to reach, so they want to know whether cleaning up that material deserves their time.

• Evaluating old thin articles: Mueller advises against deleting pages merely because they receive little traffic from Google. A reasonable page may attract few visits simply because people rarely search for its subject. The greater concern is a significant portion of a website consisting of genuinely thin, low-quality content. On a gigantic site, a few weaker areas are usually not a major problem. Identified thin pages can instead be expanded with more information or combined into a stronger page.

• Managing changing image URLs: A CMS creates image variants when an editor changes display size, while retaining the original source image at a stable public URL. Mueller says an image sitemap pointing to an unused source image probably will not resolve the central issue. Google Image Search encounters the displayed image URLs, and changing those URLs complicates processing because images are not recalled frequently. Redirect old URLs, and ideally prevent displayed image URLs from changing regularly.

• Resolving structured data differences: Search Console, a structured markup tool, and an unspecified Chrome tool may appear to provide contradictory results. Mueller identifies requirements for particular rich results as a common source of such irregularities. Structured data may satisfy baseline attributes and values while still lacking extra fields required for the desired search presentation. Start with the result being pursued, then work backward and check Google’s developer site for the attributes that result specifically requires.

• Addressing broader technical SEO: The remaining guidance distinguishes cloaking from acceptable location personalization, confirms that serving a domain from different IP addresses is common with content delivery networks, and explains that a 404 or soft 404 belongs to its URL. Pagination should make every product page discoverable, which can be checked by simulating a crawl with Screaming Frog. Excessive filtering and sorting can create unnecessary URLs. Ads and pop-ups should not obstruct content because poor placement can damage user experience and potentially affect rankings.