When you analyze a project with a traffic decrease on specific pages, you might come up with dozens of diverse theories, ranging from changes in ranking factor weights, competitor actions, negative link building, and so on.
One of the theories that must be tested is simply a mistake made by webmasters/content writers/programmers who may have inadvertently broken the page, leading to its decrease in traffic.
To assist in testing this theory, there is a simple online service that relies on data from the Web Archive. You can compare versions of the page when everything was fine with the version when the page started to lose traffic.
To do this, you’ll have to use SEODiff. Using this service is extremely simple, but make sure to read this guide before starting to work with it.
Anyway, in the left part, select the Current/Archived version of the page (twice, for comparison). Choose what to compare — the entire page, the header, the body (with code), or extract the content and compare it. Then check a couple of boxes to indicate whether to check meta tags and show pages fully/changes.
After a while, the service will display all changes to the page between the dates.
An interesting feature is that the result can be processed by ChatGPT for better understanding. There is an approximate prompt there; all you need to do is add the OpenAI API Key, which you can purchase for pennies (see a collection of links to API key sellers) and select the model. According to the author’s assurances, the service consumes a lot of tokens, and he suggests choosing gpt-4-1106-preview (preliminary version of gpt-4-turbo) — this model has the largest context window (128 thousand tokens).
If desired, you can customize the prompt. If it suits you, click the “Analyze Diff” button. This may take some time, so be patient. The end result is a summary with an analysis of the most important changes according to ChatGPT.
