Not all ranking factors apply to each site. The ranking of information sites excludes commercial factors, while online stores are not affected by factors of news sites.
Numerous studies show which signals definitely affect the position of sites. At the MozCon 2019 conference, Rob Osbey presented the results of an interesting SEO A/B test, which tested various factors, and the result of the experiment was very unexpected.
The experiment showed that removing SEO text does not increase traffic on some sites, on others, it shows positive dynamics. But there is a third group, where the removal of SEO texts led to a decrease in traffic.
SearchEngineJournal experts wrote about how to identify the factors that help your site rank.
How to identify factors that affect the position of your site?
Write down everything that you notice in the site: improvements, downgrade in response to your actions, lack of changes. For example, pages with fewer internal links rank worse.
It is not worth checking several factors at the same time since objective data cannot be obtained.
Study the available information on your observations: what does Google say about this factor/phenomenon, independent research, cases.
This will help to concretize the hypothesis and show what exactly is worth checking and what is just a coincidence.
Develop a hypothesis: what, in your opinion, leads to the observed effect. Is there any other evidence for this? How can you explain what is happening?
Do an experiment. Create two groups of pages with the same parameters. In our case, with a minimum number of internal links and low positions. One group remains the control group to exclude the influence of external factors (seasonality, updating the ranking algorithm, industry trend, which is tracked by Google Trends).
5. Data Analysis
Track the dynamics of the tested pages and compare the results with the control group. Following the example from above, determine the position of which pages have improved, downgraded, have not changed. Whether the traffic of the tested pages has changed. How the positions and traffic of the control group changed.
Summarize, collect all evidence or rebuttals. Remember that not always obvious relationships have a causal relationship. For example, you noted that highly ranked pages have lower bounce rates. However, this does not mean that a decrease in the failure rate will lead to an increase in positions.
In the report, be sure to indicate the quantitative data of the experiment in order to understand how objective the picture was.
We recommend reporting for each experiment, even if it’s just internal data. After a while, such a report will save time on the preparation of the next experiment or become excellent evidence for publication, research, and presentation.