Title Change To Misleading Title In Serps

[IanTurner]

I know Google has been changing titles in the SERPs for sometime and I've been okay with this when they are shortening titles to make them fit their display and the like. However, the latest change I've seen for one of my pages actually changes the title so that it misrepresents what is on the page in a significant way. So much so that if you read the Google generated title, it would most likely cause you to skip the page as not being what you wanted from the query you made.

Are all these changes auto-generated or are they done by manual reviewers? As in this case it is almost like someone doesn't want the page to do well and has made this change deliberately to lower its click through rate.

[aristotle]

I'm pretty sure that they're auto-generated, but the algorithm that does the generating makes a lot of dumb mistakes. You're not the only victim of misleading titles. This problem has existed for years.

[keyplyr]

Not just Google, Bing has even changed the title of my site to a pseudo-spelling lilely found as text from a back-link.

[engine]

That's more than an annoyance.
Does it show any of the words on the site, in the meta, or odp?

[IanTurner]

Description below title is directly from the meta description on the site - but it has used part of the home page title mixed in to the SERP title for a new page which is similar to the home page in some regards but not in one very important one.

Is there anywhere that this kind of anomaly can be reported?

[johnhh]

I would wait awhile, we had major title changes but it all reverted a few days later.

[Walt Hartwell]

I've actually seen this quite recently where Google appended a "region keyword" to a page title. Makes no sense to me as the region is pretty inconsequential although the keyword is a targeted one. It comes and goes, but seems to be happening more often.

[Robert Charlton]

I've actually seen this quite recently where Google appended a "region keyword" to a page title.

Walt_Hartwell, to me that's fascinating. By "region keyword", do you mean something like, say, "southwest", "mountain states", "greater LA", "northern California", etc? OK to post or exemplify, as long as it's not leading to a specific site.

It strikes me that we've been alerted numerous times by Google that they like to test a lot, and it's very possible that this title rewrite signals yet another attempt to see how results can be sliced and diced to see if it improves relevant clicks.

I find it an intriguing example because I've done a fair amount of geo-related optimization, and in some areas, in combination with some keywords, "regions" of different kinds might be searched more than an actual US Census community name. So, if Google's testing it, if you want to get more specific, I'm curious about it.

[Robert Charlton]

Regarding the Ian's original post, about the "misleading" part... to take it in sections....

Are all these changes auto-generated or are they done by manual reviewers?

The kind of title rewrite you're suggesting is way too granular to scale if done manually, and change like that aren't what manual raters/reviewers do.

They check characteristics of groups of sites, as suggested by search engineers, to help formulate heuristics for developing algorithms and then they evaluate them in blind-tests, comparing sets of results. Rater input is not for specific results or rankings themselves. In many ways, the results are statistically driven, with correlations developed by engineers, based on what the reviewers observe in their "seed sets" of pages and queries. (See video I'll mention below).

...if you read the Google generated title, it would most likely cause you to skip the page as not being what you wanted from the query you made.

What came to my mind when I read the post is pretty much what keyplyr suggested. Many rewrites I've seen are in part inbound-link driven... perhaps in combination with a statistically rare spelling, perhaps also introducing vocabulary or meaning that Google isn't finding on the page itself.

Now, to circle back to Ian's original question about human reviewers... and this is complete conjecture... this rewriting might be incorporated into an algo feature like 'statistically-based spelling changes in page titles', and then the algo, not each individual change, is further evaluated by human quality raters and search quality engineers.

I don't know how they'd rate the rewrites specifically, but I can see that they might rate groups of pages where rewrites had happened, and decide which set was better. I know they couldn't do it one by one. I suspect it's a different review team than the ranking algo review team (again, my conjecture), but the teams and methodology might be the same.

For algo reviews, I recommend that you take a look at a video featuring Paul Haahr, the Google Ranking engineer who has given the best talk about human reviewers I've seen. He discusses their role in the development and the evaluation of the algo. I've just linked to his half-hour video presentation in this thread, on an Oct 18, 2016 post...

Google Quality Rater Guidelines Update March 28roll
April 2016 etseq
https://www.webmasterworld.com/google/4799052.htm [webmasterworld.com]

I highly recommend the video. At about 28:36 into his presentation, Haahr talks about things that can go wrong, and suggests that all ranking algorithms are reviewed. Again, I'm not quite sure about title rewrites... but you might get the idea that it's a thorough process, which isn't to say error free.

I myself would wait a week or two. Without getting into specifics, you can if you like suggest the nature of the rewrite error as you see it and post it here, and maybe we'll have some thoughts. If it is spelling or simply statistically rare vocabulary, I'd make sure that the vocabulary is on your page, and, I'd carefully get some external links, spaced out over time, that include the correct vocabulary, with enough anchor text variation that it doesn't look coordinated.

[IanTurner]

I'll try and give an idea of the rewrite here (this is just an example and not my site)

Say I have a site Black Box Testing and the home page is titled Black Box Testing - All You Need To Know.

Recently I created a new page about White Box Testing with title of the form Other Boxes - White Box Testing.

Google has then rewritten the title of this page to Other Boxes - Black Box Testing - completely changing the impression you get about the page.

So although the page say comes up on page two of the SERPs for the query 'White Box Testing' no one is likely to click it as the title they see has a subject that is not relevant to their query and probably looks like something they are deliberately trying to avoid with their query.

[Robert Charlton]

Ian, your example resonates because it reminds me of something I've noticed in a fair number of recent serps, if they involve what I'm describing as "multiple entities". I posted about it a month ago here...

Massive jumps in GSC legacy crawl errors
https://www.webmasterworld.com/google/4817870-2-30.htm [webmasterworld.com]

Simon_H said that he'd been wondering "if Google has been doing a major application framework or system upgrade over the past few months/year which is now being released pre-Penguin"..., and I mentioned that I've seen results from time to time that were "oddly primitive", as if it were AI still learning....

On Sept 21, 2016, in msg #4819238...

On searches involved multiple entities, Google seems to be struggling to decide which entity is the core query and which are the modifiers. I think I'm seeing the effects of RankBrain, as in many cases the revised sense of the query is way beyond synonyms and really very good, but at other times the results are oddly primitive, probably because there's not yet that much data, and Google needs to grab at whatever obvious clues it has... vocabulary matching and the like.

If what I think I'm seeing is real, it's not just an algo change...Google seems to be building the serps differently. And yes, this would be much more than just a code update.

I don't know whether that would fit what you're describing, but it feels that it might. I'd interpret that what you're describing is a conceptually "opposite" result... the idea of "opposite" picked up from your title term "Other" along with "white"/"black". But it appears that "Other" has been applied by RankBrain to reverse "white" to "black"... not realizing that you'd already reversed that.

Too late at night for me to come up with an example as I might see it in a search, but I hope this makes sense.

[Walt Hartwell]

Robert, my situation was based on the the U.S. state of Indiana, with the two letter state abbreviation being "IN". I have had in the past many pages referencing "IN keyword" and/or "IN keyword variations". Where I believe my issue arose is when I published some fairly heavy content pages utilizing "IN" as a preposition such as "in the house", or "in keyword" which all linked (with different anchors) to the page reflecting the somewhat off-topic title.

It seems to be resolving, but with titles from years ago.

[edited by: Robert_Charlton at 6:39 am (utc) on Oct 26, 2016]
[edit reason] typo changed at poster's request [/edit]

[glakes]

my situation was based on the the U.S. state of Indiana, with the two letter state abbreviation being "IN".

I could only imagine the amount of useless India traffic you received. Hope not, but my guess is you got your share. Blocking India IP ranges is a huge help in reducing the manual scrapers/spammers.

[Walt Hartwell]

No, I did not see any spike of India traffic, it was just Google trying to determine the difference between "IN keyword" and "in keyword" where capitalized letters change the context.

[Robert Charlton]

Walt_Hartwell, the kind of geo-entity confusion I was thinking of is different from your example. I was thinking of a possible rewrite that might add, say, "Twin Cities" to a title when the query was for [minneapolis, st paul] and "Twin Cities" wasn't in either the query or in the existing title.

Your mention of the state abbreviation "IN" did bring to mind a series of discussions we had in this forum a dozen years back, when Google was treating most very short common words as stop words... ie, not really paying any attention to them at all... until some prepositions like "to", "in", "of", etc started becoming distinctive.

I can't find the particular discussion I was looking for, where I started testing the words "to" vs "in" and noticed that "to" and "from" were becoming important, perhaps because of travel, but here's a post of mine from way back about some of the first differences I was seeing regarding the word "in" in a thread about stopwords...

Google Says It Ignores Common Words, But...
... if so, why do the SERPs differ?
June, 2004
https://www.webmasterworld.com/forum3/24580.htm [webmasterworld.com]

If the "someword" theory fully explained it, "of" and "in" would be interchangeable. I've tried a bunch of real world searches, though, with both "of" and "in". Not only do they give different results, but there's another interesting difference as well... "in" triggers Local results at the top of the serps, whereas "of" doesn't.

I assume that Google doesn't index the stopwords, but it may use some of them in queries. I tried using "to" instead of "in", just in case "in" is special. Got different results yet... but, as with "of", no Local results at the top.

It was curious for me that Google now, after a dozen years of progress, seemed for a while to be mixing up uses of "in" again... particularly because Google is so much more context aware now... and your comment is in line with my thought that perhaps it may be Google retraining itself, "as if it were AI still learning"...

Emphasis added...

No, I did not see any spike of India traffic, it was just Google trying to determine the difference between "IN keyword" and "in keyword" where capitalized letters change the context.

As title rewriting comes after the rankings, this no-spike in India traffic would make sense.

Also consistent with what you're seeing, as I was looking at the "IN", "in", and related Indianapolis geo results, I was surprised to see that the use of capital letters signal different context. That too happened way back in a "learning" period the algo had gone through, easily 10-12 years ago.

We observed that there was a change in the serps for a while when caps were used... and then, as I remember, it was discarded. It's possible that this time that change might stick.

You and I tried some of the same tests. Here are several among some queries that I noted produced slightly different results....
- [indianapolis in] slightly different from [indianapolis IN]... surprised me, but made sense when I saw it...
- [indianapolis indiana] and [indianapolis in]... the difference here was expected...
- [indianapolis indiana] and [indianapolis IN] also differ... again not surprising.