Stop appending " - Web developer guides" to <title> of pages under docs/Web/Guide
Categories
(developer.mozilla.org Graveyard :: Wiki pages, enhancement, P2)
Tracking
(Not tracked)
People
(Reporter: sheppy, Assigned: dflanagan+bugzilla)
Details
(Keywords: in-triage)
Comment 1•8 years ago
|
||
Comment 2•8 years ago
|
||
Comment 3•7 years ago
|
||
| Assignee | ||
Updated•7 years ago
|
| Assignee | ||
Comment 4•7 years ago
|
||
Comment 5•7 years ago
|
||
Comment 6•7 years ago
|
||
Comment 7•7 years ago
|
||
Comment 8•7 years ago
|
||
I looked at our y/y traffic data and it's clear that we dropped 16 percentage points in organic search traffic in the 2 days following this change. Because of the all hands we had not pushed any changes for 10 days before this change and we didn't have a production push for 5 days after that. Here's the full change set: https://github.com/mozilla/kuma/compare/2687c59...4ec6e67
From that list, this change seems to be the most likely to have had SEO impact. Now, I don't know why this change would have that much of an effect, but if the drop is based on our changes, this seems to be the most likely culprit. Can you please revert the change, so we can test this theory?
Kinda hoping it's not this change, but considering the massive drop in traffic, it would be irresponsible to not at least check.
Comment 9•7 years ago
|
||
I think it is just as likely that changing all of our titles again will give another 16% drop in search traffic.
Comment 10•7 years ago
|
||
Interesting. I realized a while ago that there is no "JavaScript" in the titles.
https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/Array is only "Array | MDN". I guess "javascript array", "js array" etc is searched for quite often.
W3schools is first for me now https://www.google.com/search?q=javascript+array
Would a revert make this (again) "Array – JavaScript | MDN"?
Comment 11•7 years ago
|
||
(In reply to John Whitlock [:jwhitlock] from comment #9)
I think it is just as likely that changing all of our titles again will give another 16% drop in search traffic.
In which case we can revert the revert ;)
Comment 12•7 years ago
•
|
||
Sorry if I wasn't clear. I think Google includes the age of the title in the search ranking algorithm, and penalizes us for changing the title.
However, Florian is more convincing, that going from "Array - JavaScript | MDN" to "Array | MDN" is probably a big hit.
djf, do you want to revert?
| Assignee | ||
Comment 14•7 years ago
|
||
Ouch! This was I think my very first patch. Emabarassed to have caused so much damage!
I'll prepare a PR that reverts this.
I could also prepare a PR that then does what the title of the bug requested originally and only remove "Web Developer Guides" from things under "/Web/Guide" but leave the SEO titles for everything else. Or, if I'm special casing the /Web/Guides thing, I could change that to a shorter string like "Guides" or "Tutorials" or "Webdev".
Sorry about this, everyone... SEO is hard!
| Assignee | ||
Comment 15•7 years ago
|
||
There's a PR to revert the whole thing here: https://github.com/mozilla/kuma/pull/5226
| Assignee | ||
Comment 16•7 years ago
|
||
And on second thought, trying to hardcode a different seo title to use instead of "Web developer guides" will get us into localization problems (unless I hardcode it english only.) Looks like the title of /Web/Guide has changed to just be "Developer guides" so that's a bit shorter anyway, even though it probably isn't much good for SEO.
If there was more content under /Web/Guide/, I'd suggest using the third element of the slug as the seo root, and hardcoding that as the special case here. But it doesn't look like it would affect many documents, so I'm guessing it is safer to do nothing at this point.
| Reporter | ||
Comment 17•7 years ago
|
||
Go ahead and revert it. I would actually prefer we come up with a more intelligent solution than this but given time constraints, this will do.
Comment 18•7 years ago
|
||
Comment 19•7 years ago
|
||
Thanks for the quick revert. And no worries David, I shouldn't have made that off hand comment earlier in the bug. I didn't see this in the sprint planning, and unfortunately was on PTO for a few weeks after it was pushed to production. I think we need a light protocol around SEO changes. We need to at least compare the before and after for 7 days before moving on. Ideally we test the change with one area before making a site wide change.
| Reporter | ||
Comment 20•7 years ago
|
||
@atopal - Agreed; we need to have a system in place to compare the before and after. We should have a specific place we record our results and what is being tested so we can keep things straight.
Comment 21•7 years ago
|
||
Sheppy, we can use sanitycheck for the testing. Taking this into the mdnseo slack channel for further discussion.
| Assignee | ||
Comment 22•7 years ago
|
||
I'm going to close this now, since I don't think there is anything else to be done.
Kadir: if reverting this change doesn't fix things, I think we'll need a new bug about that.
Sheppy: if you still want to think about changes to the SEO root titles, I think that probably deserves a fresh bug as well.
Comment 23•7 years ago
|
||
Quick update. Y/Y traffic is positive again for all top 20 countries, except Korea, China and the US, it happened right after this change went into production. In China it's the spring festival/new year, so Y/Y data is not comparable. Same goes for Korean new year. I don't quite understand the data for the US yet, but we can at least see a reduction in the gap. I'll monitor for at least one more week, and give an update then.
Comment 24•7 years ago
|
||
Final verdict: Yeah, title wording and content is a big deal. We are at 12% y/y growth again. The gap for the US has narrowed further, but it's still negative y/y. Discussed this with Sheppy and Raphael, but couldn't think of a particularly good reason. Will keep monitoring to see if that trend continues, but it's not part of this bug anymore. Either way, we ran a massive unintended experiment, would probably be worth writing about it.
Side note: I dug into the data some more and it turns out, the US had stopped showing positive y/y growth right after October 30th 2018. We discussed that too, but couldn't think of a good reason and the data we looked at was not helpful. The change sets before that date don't look like they should impact SEO.
Updated•5 years ago
|
Description
•