Change the logic around categorizing non ad clicks to be less restrictive
Categories
(Firefox :: Search, task, P1)
Tracking
()
| Tracking | Status | |
|---|---|---|
| firefox115 | --- | fixed |
People
(Reporter: jteow, Assigned: jteow)
References
Details
Attachments
(1 file)
We're not accurately registering clicks on non-ads on Google.
In SearchSERPTelemetry, we observe network events and check if a channel is a top level load and is a document. This to avoid parsing requests that are unlikely to request a change of the location of a tracked browser.
This seemingly works for ad links and non ad clicks for most SERPs. However, on clicking non-ad links on Google that don't open a new tab, the one time we experience a channel that is is both a top level load and is a document, wrappedChannel.browserElement returns null.
This is annoying because we are using browsers to determine whether the request belongs to a source that we care about.
The webidl comment mentions "for requests that don't originate from a remote browser, this is null."
| Assignee | ||
Comment 1•2 years ago
•
|
||
More descriptive bug title.
| Assignee | ||
Comment 2•2 years ago
|
||
Updated•2 years ago
|
| Assignee | ||
Comment 3•2 years ago
|
||
I've changed the gist of this bug to be about being less restrictive about the categorization of non ad link clicks.
I wrongly made the assumption SERPs are consistent about encoding non ad results with redirects. Bing for example has the ability to enable/disable links to open in new tabs. And Google loads a link in a subframe first before changing the location of the page which throws a wrench in the assumption that only channels that are top level and containing a document should be inspected.
That said, I think we can be safely assume any traffic that a) has an originURL closely matching a SERP, b) is a top level load (except for links that match a nonAdRegexp), c) is a document, should be a click event, and d) originates from a browser that is currently being tracked, is a click event.
The tricky part is to ensure that any redirects are recorded once.
Comment 5•2 years ago
|
||
| bugherder | ||
Description
•