OK… I feel a bit daft asking this question, as I expect the answer to be no, but here goes…
Is there a way to record non traffic related statistics - e.g. time taken to complete a transaction/game, scores on a game, purchases made, etc…
Thanks
Yes you can simply trigger “fake” pageviews with Custom titles… Read Tracking API - Analytics Platform - Matomo
OK, before I go and deep dive…
Example - A known user takes a quiz, takes between 5&10 seconds for each question, completes quiz in 150s, gets a score of 15/20. Can I record all of that via “fake” pageviews?
Thanks Matt!
This is what I’ve been looking for (for a while actually, we were going to write our own system to do this all!!!).
OK, is there anything stopping one presenting the stats to individual users from this?
E.g. a Pie chart showing which type of activities an individual user has completed, a “heatmap” of users locations by no. of activities of a certain sort completed…
Also how goes Scalability now, can Piwik (in this sort of scenario) cope with millions of hits a day?
Yup, I’d seen that.
Was interested in the MongoDB implementation, has this made its way into the trunk yet? Are there any plans for this? (I shall most a similar question there).
Also the release notes indicate improvements with scalability, but I couldn’t see any details for it.
[BTW sorry if I’m asking questions asked/answered elsewhere, I have looked around but not seen what I am looking for, yet]
Thanks
Argon0
Plans for MongoDB are definitely paused currently. I think probably we will work on a new backend engine in early 2012, maybe looking at InfiniDB ?
Isn’t InfiniDB optimised for data reads rather than inserts?
What are you seeing as the biggest resource hog in Piwik - reads or writes?
I would have though that at the millions/day level writes/inserts would have been more important than reads… But I’m as yet unfamiliar with the way it all works, so…
Argon0
both are extremely important of course, “read” is more like “analytics” engine to do segmentation, custom reports, etc.
That is what I thought.
Have you considered something like Redis (VERY fast writes, as it writes to memory then flushes to disk afterwards, asynch I think)?
Or do you want to keep it SQLesque?
(in which case have you thought of using MySQL cluster?, if so what are its limitations?)
[sorry for so much interrogation!!]