[] We need to be tracking click through to user, tag and category. If you click through to any of those the search is still a success, we now only track click through to topic AFAIK.
[] We should log fullpage vs mini search in the table (then we can add a “all” / “header” / “fullpage” filter to this page)
This allows you to isolate terms that seem to get click throughs on fullpage and dont on the mini search like sso on meta
[] Left align search terms, left align most viewed topic
It would be nice to be able to click through to the actual search, but its a tricky UI change cause you don’t really want to make all the terms blue and you don’t want to add an icon that repeats 100 time
[] We should sort by unique descending, not by searches
[] Add tooltip on unique: “unique users performing the search”
We should think about ways of allowing people to track “progress” here, for example:
“sso”: Click through 13% (30% last week) - oh no - we changed something that made click through worse
This is probably most relevant in month/week time frames on stuff that is searched more than say 20 unique.
“Click through to a topic” doesn’t look like getting tracked properly.
This is what I did
Searched for a term. Clicked on a topic and scrolled till the bottom of the page.
Accessed /admin/logs/search_logs?period=daily
I can see the term that I searched for. But it shows “0” on “click through count”, “most viewed topic” column is also blank.
I thought something is wrong with “click through tracking to topics” because, when I checked “today’s search logs”, I can see like 100+ searches, but “click through count” and “most viewed topic” column is filled for just 2 search terms. This is when I did a search myself to confirm if something is really wrong. And after my tests “like I mentioned above”, it looks like its not tracking “click through to topics” correctly.
Does unique track IP addresses or user accounts? (since users may use more than one IP)
It sounds like IP addresses here:
And user accounts here:
Our sites are only visible to people with user accounts, and that’s not the case for everyone so probably it’s IP addresses, but maybe there’s some magic going on in the background. Either way, it would be useful for us to know which of the two it is.
@techAPJ can confirm but in your case (“a requires login community”) the unique count is the unique count of users. AFAIK we use user_id and then fallback to ip