Historical data does not seem to work
We're currently working on extending and expanding the data we collect for our website and have recently introduced custom dimensions in our GA implementation.
The dimensions collect various bits of data about the page - article ID, title, author, topic, etc and fire it off to GA for reporting / analytics later down the road.
We have this data readily available for ALL of our historical articles and posts and tried to use the data import tool in GA to pull it into Analytics.
In order to do this, we needed to first match each url to an article ID. Our formats varied over time, but they're easy enough to parse. For example's sake:
We exported a list of all pages tracked by GA since we started collecting and created a CSV containing the URL and Article ID in separate columns to import into GA. Once we have the IDs populated, importing the remaining metadata *ought* to be simple - export from CMS, import to GA.
After importing the historical data, we noticed that it didn't apply it to any of our pages.
On further investigation, it appears that the only urls with the ID field seem to have been visited since the launch of our metadata collection. It hasn't applied any metadata to older articles.
We found this article from 2015 that indicates that historical data is not processed by Google:
Uploaded data only gets added to your data during the processing stage. This means it will not be applied retroactively (Google would have to reprocess all your historic data, and that’s not going to happen). It will only be applied to hits as they’re being received. Also, it may take up to 24 hours before the data shows up in your reports.
However, the support page on Google indicates:
Extended data import adds to (extends) the data already collected and processed, or being processed, for the selected reporting views. Typically, this extended data is stored in a custom dimension or metric, though in some cases you might want to overwrite the default information already gathered (for example, importing a campaign's Source or Medium dimension).
Is the Lunametrics article correct? Does GA only apply the imported data during the processing stage when a page is viewed? Or does google apply it retroactively? Or did we just do something wrong?