Forum Moderators: Robert Charlton & goodroi
Keyword: Laptop
#1 - Apple
#2 - Dell
Keyword: High Speed Internet
#1 - ATT
#2 - Comcast
Keyword: Quit Smoking
#1 - SmokeFree.gov
#2 - CDC.gov
Keyword: Car Audio
#1 - Crutchfield
#3 - Pioneer
Does anyone else see this in competitive spaces?
One way Google could be discerning the quality of being a brand might be through inbound anchor text. There would be a certain mix of Brand (as opposed to brand.com) and also specific keyword text, over a certain threshoold. But certainly not a hodgepodge of varied anchor text -- there would be a certain staistical balance they would be looking for.
Why Dell and not HP for example?
Deciding the difference between brand searches and non-brand ones is always going to be difficult in some cases.
Roll this out across every site though and it becomes interesting, as historically the external factors that Google has used to rank sites are pretty limited. Now, if all of a sudden they are saying that it's not only about links, but also how much you're being talked about on the web (linked or otherwise) then this is something that can really affect people's link building efforts, and the ways people SEO.
It kind of makes sense, why would Google just give my site credit is someone links to me. If someone decent talks about me, surely that's worthy of something as well. But wait... what if they talk about me in a derogatory way - and says that my site sucks..
Now I'm wondering though.. what is a top brand? Is a top brand something that has been talked about a lot? What if it is talked about in a hugely negative way - does that mean that Google should be giving them more prominence in the rankings? It's pretty subjective all of a sudden..
It's pretty subjective all of a sudden..
If someone decent talks about me, surely that's worthy of something as well. But wait... what if they talk about me in a derogatory way - and says that my site sucks..
Yes, BRANDS can come from any category (ecom/manufacturer/info/comparison/forum/blog/social media), but to become a BRAND, you must transcend that category. Examples include
ecom=amazon
manufacturer=coke
info=wiki
comparison=?
forum=WebmasterWorld(arguably)
blog=?
social media=facebook
I would imagine brands would be names that co-occur near keywords in completely unrelated sites- possibly as shorthand in unrelated niches. So, you might see someone on a social site saying "I bought from amazon"; "I like drinking coke"; "I bought my Dell". Stuff where the brand name is sufficient to give meaning to a sentence, without context.
Now, once a brand transcends it's category, it no longer takes up it's categories' slot on page 1.
I would imagine Social Media and news would be the primary sources for Brand identification.
the idea of giving some trust and authority to unlinked brand citations online seems possible to me
Most people that complain about ranking have trash websites. People think they can sell stuff out of the back of a truck and compete with a big box store.
Well, why would google has to rank for that seen you have found a "search engine" already and are using it (google)..., it's irrelevant when you are already on google. Giving you alternative search engines is correct! That's actually clever and should be the expected result.
If you enter Walmart and ask an assistant: Any shopping malls round here?, the assistant should not say "You are in one" as that will sound dumb, but likely to say, oh there is...and there is...a mile away apart from this one!
The whole ranking thing is to give you the best results so you can click and VISIT them, you are already on the best result page for that term, period!
This reminds me of a Joke about G Bush junior addressing a crowd in the UK and asking "Was any of you guys a student, cos life can be tough for students". Nothing wrong with that you might say, except he was addressing the Oxford Union and no one explained to him it was not a trade union!
I can't see the Google algo differentiating between a "pure play" or online-only brand and a brick and mortar brand's online efforts. If there is a brand component in the algo, I think it would be tone-deaf on that score.
But the very nature of a brick and mortar brand often does seem to work against them. The web gene is just not very strong in their DNA, and the entire corporate structure has usually evolved to address other purposes, sometimes counter-purposes to web success.
And yet, the end user will often expect to see at least some of these offline brands represented in the more generic, non-branded queries. It's still a rare b&m brand that gets the power of that.
I'm often involved with brands that do not even include their drop-dead obvious keyword anywhere on the home page. And no, Google will not just hand them rankings for that keyword.
[edited by: tedster at 5:36 am (utc) on Feb. 25, 2009]
Its backlinks I tell you! ;)
I'm often involved with brands that do not even include their drop-dead obvious keyword anywhere on the home page....
Very true about the corporate DNA. I've had major b&m brands in the course of an optimizing project remove keywords that were added. Often, by the time an optimized site gets through marketing, legal, and various department heads, it's virtually back to what it was before you started.
...And no, Google will not just hand them rankings for that keyword.
This latter point should be noted in relation to the ranking examples cited by the original poster and others within this thread. In the searches mentioned, there were obvious algorithmic reasons why the pages that were ranking were ranking. Nothing mystical or preferential about it.
[edited by: Robert_Charlton at 5:38 am (utc) on Feb. 25, 2009]
If you use a tool like RankPulse (freely available) and look up keywords like
- airline tickets
- auto insurance
- boots
- diets
- health insurance
- online degree
- watches
you will see that all of them had some big brands appear starting January 18th. Also some brands that should rank (but didn't until recently) started ranking...for example...
- Hallmark ranking for gifts
- Radioshack ranking for electronics
In addition to his recenty branding comment, in the most recent conference call Eric Schmidt mentioned something about semantic search and word relationships.
This is something we all know is happening and, more specifically, we know (as a fact unless their guidelines have changed) that certain brands ranking for certain terms is one of the primay things their human testers work to ensure.
>... and have linked to my companies website ...
So I don't. Well, not publicly, at any rate: it's in my affiliations reported to the Open Directory Project community leaders, and the moderators here could probably deduce it, if necessary. For that matter, I don't think it would take much all that much Googling to figure it out.
Not that it's secret. But the identity of my employer is unrelated to my personal identity, and my personal activity online is unrelated to my employer (who isn't unique in having equipment-use policies allowing some kinds of pro-bono work to be done on my time on their equipment.)
But you're likely right in general, that large corporations, even without aggressive artificial website promotion, end up having really really _natural_ inbound links (based on real-life relationships) which is precisely what Google is trying to look for.
Those natural links overwhelm the vast majority of artificial schemes, but "above the vast majority" of 2 million page hits can still fall far below the first page of results.
semantic search and word relationships
There are times when the tilde operator shows that a brand name is identified with a generic word. That is just for the top level of semantic analysis that Google is willing to expose, but Google has MUCH more semantic research that we don't see.
Im jumping in on this one a bit late so apologies if i repeat anything as ive only scanned the earlier posts.
Aaron wall has done an interesting post on the issues youre talking about: [seobook.com...] (maybe he got the idea from this thread?!). Anyway, it was quite insightful and Aaron points to an update that happened in late January that largely went unnoticed.
Personally, I'm not a fan of increased rankings for big brands. One of the examples that Aaron uses is Radioshack, who appear to have shot up to the top page for electronics. This raises a few questions, what have google changed to implement "brand rank" (coined a new phrase there!) and what do seos have to do to leverage that?
Maybe google have started to look at your web presence with regards to how often your brand name appears on other sites (regardless of linking) and in what context? They could use their many spyware tools to assess search count, direct traffic and loyalty?
Isnt "brand rank" gonna make it harder for the smaller business to compete on what has previously been a level playing field? Level playing field my arse! It's getting increasingly uneven if you ask me. (Maybe) soon you'll only be able to achieve internet success if youre a big budget brand, work in a niche area or offer a free service.
The minute you're going to look at mentions of a "brand" and then try to incorporate that into an algorithm, you're getting into an absolute minefield as far I can see. It's taking semantic to a new level, and bringing a huge amount of subjectivity into it. If tilde operators are already associating words such as shoes with certain brands, then all of a sudden you're looking at defining exactly how big a brand that company is, and a whole lot more besides.
I'm not suggesting it's not happening, as there is clearly evidence something is going on - I'm just really wondering how Google is doing it, and how they are "managing" to do it in a way that doesn't go against all their principles.. (if you know what I mean..)
If you make something unique and useful you will win.
If tilde operators are already associating words such as shoes with certain brands...
Actually, I don't see any brands associated with [~shoes] - only "boots", "footwear" and "store". The semantic association with the word "store" (high co-occurence) has the very strange effect of putting Apple on page one for a [~shoes] search!
~apple ~laptop [google.com].
Note that Dell laptops is highlighted as the #3 result.
Another tilde search for me that illustrates a significant degree of co-occurrence correlation is this one...
~ford ~taurus [google.com].
Note the highlighted results showing for Honda Accord. (I know that this particular search result has been around for a while).
Aaron's article is an impressive bit of research, and checking the RankPulse charts makes it clear that something changed in mid-January.
I don't think that co-occurrence between a brandname and keyword, by itself, explains the ranking jump in the searches I've had a chance to examine. There still need to be other factors for a page to rank.
While these two examples, eg, that Aaron cited are not what you'd call conventially well optimized, there are clues provided to Google for the pages that rank...
- Hallmark ranking for gifts
- Radioshack ranking for electronics
In the case of Hallmark, "gifts" does occur in the title and you see it twice in the text cache. In the case of Radio Shack, "electronics" is present in inbound links to the page and is roughly the ninth word in the home page's rather long title.
I haven't had a chance to check enough brands to see whether any rank without apparent onsite or link-driven reasons for ranking. My instinct says that it's unlikely, and that we're seeing the co-occurrence knob being turned up, but operating together with traditional algo factors.
One of the questions I'd ask in relation to this discussion is whether this is a boost applied just to brands, or whether it's an overall co-occurence boost of some kind, perhaps one that only kicks in when there's a strong link profile for the co-occurring term (as there would be on a branded page).
[edited by: Robert_Charlton at 8:27 pm (utc) on Feb. 26, 2009]
You might assume brands have great websites with better content and links than smaller operations. But I have examined several keywords and I can tell you Google has rewarded some brands that do not deserve it. Some of these brands have websites with poor usability, little content and their backlinks (quality and quantity) are not even half as good as the other sites.
It seems there is a new signal at work. I am very surprised at how little attention has been paid to this google change.
Google has rewarded some brands that do not deserve it
I also see this - except I'd express it "Google has rewarded some brands that do not deserve it according to our established ideas of SEO." However, from a total business perspective, these are strongly branded businesses, and Google's users might well expect to see these brands in results for a generic query - so it doesn't surprise me that we are getting a sense of some new 'signal" at work.
Many webmasters may think of Google as running some kind of SEO contest, rather than a service for people who looking to access information.
How can a low-cost-of-entry website imitate a succesfully branded business, whether pure-play or b&m? A business with real, satisfied customers? I don't think that ultimately, it can be imitated. A website can only BE such a business or not. And that seems to me what Google is aiming to judge, by whatever criteria they can summon up... and they've apparently got a new one.
Such an undertaking would be a massive challenge. Some queries are going to have an informational intention, but others may have a buying intention. Google has been sorting that factor, but they need to sort it very well if they're going to integrate some kind of brand power signal and do it effectively.
Not all brands offer e-commerce, but many do. That would be another factor in the mix. The more I consider this, the more challenging it seems.
Matt Cutts:...I talked about this in a video today and we'll try to get that up in a few days. The short answer is that we did change some of our algorithms for some queries, but this was just one of the hundreds of changes that we make each year, so I certainly wouldn't call it an update. The change was discussed on WebmasterWorld for days before Aaron wrote about it, but really not that many people noticed it. That's in line with the fact that scoring changes recently tend to be more subtle and more targeted to improve specific types of queries.
See [sphinn.com...] for much more
However, it's interesting to note that, in the US at least (where I'm guessing most folks are), the search volume for each is roughly the same, with Apple gaining in the past 2 years.
[google.com...]
(I tried running the search initially without a geo filter, but there was to much asian data from Indonesia etc., where I intuitively feel the computer market for those 3 is perhaps not the greatest. Be interesting to find out what the deal is there! HP blows Dell and Apple out the water on Gtrends worldwide.)
Robert Charlton and Signor_John - genius contributions to this thread you two. I love WebmasterWorld for this stuff! :D
p.s. I have a theory on what is driving this brand stuff - type-in traffic and its search equivalent, branded search volume. I won't link to my post on this (though I'd love for any of you to read + comment on it if you can find it), but I've noticed that URLs that get search volume get indexed (Yahoo exhibits this too). However the URLs are prone to bounce in/out of the index on both SEs.
Speaking of Google only now, why stop with indexing off search volume? Why not give a positive boost above a certain threshold of volume? E.g. 500 searches/day?
And this ties in with what you guys were saying about correlations and co-occurrence, but methinks it goes beyond strictly keyword co-occurrence in citations. My guess is that the relationships can be something like:
Viral widget on minisite from company A gets branded search volume for the widget over time (e.g. 1 year +)
CompanyA.com gets a boost.
The reason for that belief is because I've seen a banned domain show up in very-specific branded search situations (e.g. banned.com as a keyword). By contrast, another domain I know of that is banned and was not associated with a popular widget does not get this treatment.
Who the heck really cares if these sites are ranking for those generic keywords? It's a freakin' vanity thing. Very few of those single keyword searches have any real value other than some CEO being able to tell another CEO that we rank #1 for Shoes or whatever. It's an ego thing.
Ya, if a searcher wants to do the drilldown after entering a single word search like that, then they deserve what they have coming to them. And, I applaud Google for making sure that the searcher is presented with an authority for those single word searches and not some crap affiliate site.
[edited by: tedster at 1:43 pm (utc) on Mar. 1, 2009]
Did yall not read what MC said. They manually made some changes on a few search terms that does not mean there is a branding filter. If they did a hand job there is nothing you can do about it to beat the searches they did that on.
