After a brief analysis of ranking drops 2 conclusion were drawn:
1. Not all keywords were affected. Differences of positions were noticed for keywords related with subcategories of selected categories. One the grounds of this conclusion it was possible to distinguish parent categories associated with the ranking drops. Consequently, a direction to search the source of the drops was given.
2. The position itself wasn't the only change in Google search engine. Even the subpage indicated by the results page had changed. It was always the page of parent category.
We copied part of a description to Google search engine. The search engine results page displayed lots of its own domain subpages. What made the text content to get crawled on different pages?
After thorough analysis it turned out the lately crawled pages had been created by (seemingly) minor shop engine flaws.
First issue was the "=" or "?" symbol added at the end of category URL address:
What's more important, it was the reason URL addresses did not linked to its categories but to their parent categories: www.exaple.pl/main-category/.
Second issue was the canonical tag used on pages related to product sorting options. For example, the canonical page for www.example.pl/main-category/subcategory/sort=price_a was www.example.pl/main-category/subcategory even though the category link structure had always "/" symbol at the end. The right URL address was www.example.pl/main-category/subcategory/. Incorrect canonical tag had been ignored by Google robots because multitude of URLs containing sorting parameters were indexed.
Third kind of error applies to the improper use of canonical tag implemented on pages including i.a. gclid parameter which refers to an AdWords tag that collects data about website entries. Indicating wrong canonical page triggered off indexing unwanted URL addresses - duplicated categories.
How did the drops affected site visibility in search engine results page? Traffic chart for selected categories:
Website structure of the shop had been fixed from January 20-26th. Some subpages had needed more time to return to their former top10 rank position while other got back in a short time. Not all keywords got back to its original ranking positions. This is done systematically and probably it's caused by the website re-crawling process.
As you can see, internal duplicate content may have negative consequence for keywords ranking in Google search engine. The source of duplicates was the corrupted shop engine but there might be more reasons for that e.g. same text on each page, identical product description, even a printable version or duplicated category description. This is why it is important to take care this aspect of website optimization.