Is Compression A Google SEO Myth?

I recently came across an SEO test that attempted to verify whether compression ratio affects rankings. It seems there may be some who believe that higher compression ratios correlate with lower rankings. Understanding compressibility in the context of SEO requires reading both the original source on compression ratios and the research paper itself before drawing conclusions about whether or not it’s an SEO myth.

Search Engines Compress Web Pages

Compressibility, in the context of search engines, refers to how much web pages can be compressed. Shrinking a document into a zip file is an example of compression. Search engines compress indexed web pages because it saves space and results in faster processing. It’s something that all search engines do.

Websites & Host Providers Compress Web Pages

Web page compression is a good thing because it helps search crawlers quickly access web pages which in turn sends the signal to Googlebot that it won’t strain the server and it’s okay to grab even more pages for indexing.

Compression speeds up websites, providing site visitors a high quality user experience. Most web hosts automatically enable compression because it’s good for websites, site visitors and also good for web hosts because it saves on bandwidth loads. Everybody wins with website compression.

High Levels Of Compression Correlate With Spam

Researchers at a search engine discovered that highly compressible web pages correlated with low-quality content. The study called Spam, Damn Spam, and Statistics: Using Statistical Analysis to Locate Spam Web Pages  (PDF) was conducted in 2006 by two of the world’s leading researchers, Marc Najork and Dennis Fetterly.

Najork currently works at DeepMind as Distinguished Research Scientist. Fetterly, a software engineer at Google, is an author of many important research papers related to search, content analysis and other related topics. This research paper isn’t just any research paper, it’s an important one.

What the research paper shows is that 70% of web pages that compress at a level of 4.0 or higher tended to be low quality pages with a high level of redundant word usage. The average compression level of sites was around 2.0.

Here are the averages of normal web pages listed by the research paper:

• Compression ratio of 2.0:
  The most frequently occurring compression ratio in the dataset is 2.0.
• Compression ratio of 2.1:
  Half of the pages have a compression ratio below 2.1, and half have a compression ratio above it.
• Compression ratio of 2.11:
  On average, the compression ratio of the pages analyzed is 2.11.

It would be an easy first-pass way to filter out the obvious content spam so it makes sense that they would do that to weed out heavy-handed content spam. But weeding out spam is more complicated than simple solutions. Search engines use multiple signals because it results in a higher level of accuracy.

The researchers reported that 70% of sites with a compression level of 4.0 or higher were spam. That means that the other 30% were not spam sites. There are always outliers in statistics and that 30% of non-spam sites is why search engines tend to use more than one signal.

Do Search Engines Use Compressibility?

It’s reasonable to assume that search engines use compressibility to identify heavy handed obvious spam. But it’s also reasonable to assume that if search engines employ it they are using it together with other signals in order to increase the accuracy of the metrics. Nobody knows for certain if Google uses compressibility.

Is There Proof That Compression Is An SEO Myth?

Some SEOs have published research analyzing the rankings of thousands of sites for hundreds of keywords. They found that both the top-ranking and bottom-ranked sites had a compression ratio that was essentially the same as the 2.11 compression ratio that the 2006 researchers discovered as being in the range of normal.

The SEOs claimed that the results prove that compression ratio is an SEO myth. Of course, that claim is far from correct and here are two reasons why.

1. The average compression ratio of normal sites in 2006 was 2.11, which means the average they discovered falls well within the range of normal, non-spam websites, which one would expect to see in the search results. Remember, if a site is spammy, it’s supposed to get blocked from indexing.

2. If we assume that Google is using compressibility, a site would have to produce a compression ratio of 4.0, plus send other low quality signals, to trigger an algorithmic action. If that happened those sites wouldn’t be in the search results at all because they wouldn’t be in the index and therefore there is no way to test that with the SERPs, right?

It would be reasonable to assume that the sites with high 4.0 compression ratios were removed. But we don’t know that, it’s not a certainty. So we can’t prove that they were removed.

The only thing we do know is that there is this research paper out there that’s authored by distinguished scientists.

Is Compressibility An SEO Myth?

Compressibility may not be an SEO myth. But it’s probably not anything publishers or SEOs should be worry about as long as they’re avoiding heavy-handed tactics like keyword stuffing or repetitive cookie cutter pages.

Google uses de-duplication which removes duplicate pages from their index and consolidates the PageRank signals to whichever page they choose to be the canonical page (if they choose one). Publishing duplicate pages will likely not trigger any kind of penalty, including anything related to compression ratios, because, as was already mentioned, search engines don’t use signals in isolation.