Canonical Faux Pas
A small eCommerce website where they have Category (~100) pages and sub product pages and canonicals were set pointing correctly to their actual URL (Canonical was same as URL)
They implemented a new product snippet via JSON LD. During the implementation of the "header tags" section of the code of the page as this would be the location of the Product JSON-LD code too, they also unknowingly introduced a new undocumented feature (also known as error or bug) that went unnoticed for 8-12 months. This new implementation of product snippets was added and tested (albeit only on the product pages) fine. The snippet was correct, Google could read it, and the canonical stayed the same. The new "feature" was that the primary Catagory`s canonical now pointed to the root domain and not to its actual URL (Eg canonical pointed to the home page instead of thedomain.com/catagory.htm ). This IS the faux pas.
During those months of the canonical faux pas, the traffic and SERP rankings increased 20-40%, many new 1st page rankings achieved, etc. Yes, many other things were being done too, such as improved content, link structure, 301 redirects of discontinued products/categories, etc, etc. So to pinpoint anything specific would be difficult, to say the least.
The home page linked to most of the category pages with the bad canonical, so it at least had related text. All the product categories are different but are "Theme" related. This leads to the question, How "Bad" was this? It seems that the canonical tag may have been ignored and treated like a hint/directive rather than an absolute. (Time will tell more as the bug was fixed in last 24hrs) Thoughts?