GA isn't compatible with a packaged app, so we should be working hard to create a solution based on the feedback of the GA premium support team: 1. Create an iframe which embeds GA 2. Use postMessage to pass page views, events, and variables to the iframe Pretty straightforward, but we need to test this and make sure it doesn't gum anything up, namely getting weird URLs due to passing a full URL (with origin) to the GA page view method.
This will be awesome to do right and we should openly share how we accomplished it. Here is some code examples to help us: https://github.com/GoogleChrome/chrome-app-samples/tree/master/analytics I also read people using this: https://github.com/ggendre/GALocalStorage
https://github.com/mozilla/fireplace/commit/7e706a276c7d04193e9f541fc2d94a76cfb68033 v1 is here, testing to commence shortly.
Testing has started for the second time today. We'll see if there are any more issues that need to be worked out.
This is unexpectedly blocked on bug 897146. Inferno is acting up, which means getting the ball rolling with Universal Analytics might take until next week.
Depends on: 897146
"Hopefully we will be able to deploy potatolytics next week"
Priority: -- → P3
UA landed on -dev today: https://github.com/mozilla/fireplace/commit/a4f72bb608b9a6c4b20ab76e0438bca52f6a97c6 I'll start on the potato side of it tomorrow. It should actually be pretty easy; UA lets you disable storage entirely and instead provide your own client ID. We've already got that part in place, so the only challenge is going to be ripping out the GA-potato code and replacing it with UA-potato code.
This adds potatolytics support to UA: https://github.com/mozilla/fireplace/commit/5b97ae6a6981ea57e4d36d5e66238cb642844640 I'll start testing it locally and maybe flip it on briefly on -dev. Still waiting on gcull to give the thumbs up on the UA data, so I don't want to completely switch it over if there are bugs with the underlying UA abstraction layer.
This is built, assigning to Gareth since we're blocked on his analysis of the current UA implementation. Gareth, any update?
Assignee: mattbasta → garethcull.bugs
Summary: Create Potatolytics (packaged app-patible GA) → Deploy Potatolytics (packaged app-patible GA)
I'll review and will get back to you with an updated analysis later this week. Gareth
(In reply to Gareth Cull [:garethc] from comment #9) > I'll review and will get back to you with an updated analysis later this > week. > > Gareth How'd it go?
Hi Wil, At a high level, the aggregate data looks pretty comparable across a variety of metrics. But, I am seeing some differences when I break it down by some segments. I'm reviewing with Matt tomorrow if you would like to join. Also, I'm not sure if you know this, but Google recently announced that we will be able to migrate our existing Classic Implementation to Universal Analytics with an upgrade tool. This would be great, as we wouldn't lose any historical data when switching over to a UA only account. Something to discuss. More on this here: http://analytics.blogspot.ca/2013/10/an-easy-way-to-upgrade-to-universal_23.html Gareth
Hi Gareth - we're looking into revisiting this shortly. Can you give an update on your last meeting and next steps here?
Hey Wil, Universal Analytics (UA) is out of beta now, and we can migrate our current Classic GA implementation over to UA. Switching over to UA will require us to update any custom variables that we are passing on marketplace in addition to updating the GA snippet. I believe some of your team maybe using some custom variables in your reporting, which would necessitate the need to update. Let me put together an etherpad outlining some of the changes we will need to make. Could you please schedule a quick meeting with me to review sometime next week? Thanks.
functionality can be flipped with potatolytics_enabled flag.
Status: NEW → RESOLVED
Last Resolved: 4 years ago
Resolution: --- → FIXED
Please add STR here or mark it with [qa-] if no QA is needed.
Iirc this was enabled on dev last year. If not, the qa process involves turning it on for dev and comparing incoming data with historical data.
You need to log in before you can comment on or make changes to this bug.