Google's Rank Modifying Patent for Spam Detection

[tedster]

Patent hawk Bill Slawski has published an article [seobythesea.com] about a newly granted Google patent - one that seems to explain a lot of the odd ranking behavior that has been reported in recent times. The patent is named, quite simply, Ranking documents [patft.uspto.gov]

As I understand it, the idea here is to identify, by algorithm, what looks like an attempt to manipulate rankings in a spammy way. The patent details things like keyword stuffing, invisible text, link-based manipulations and so on.

Rather than allow the rankings to respond immediately and directly to those changes, the patent explains a system that would change rankings in unexpected, counter-intuitive ways - while the rankings change from a first position through transition positions and to the final "target rank" position. In other words, significant changes in position continue to happen even though there is no change in page's ranking factors!

Bill explains:

During the transition from the old rank to the target rank, the transition rank might cause:
a time-based delay response,
a negative response,
a random response, and/or
an unexpected response

If these transitional ranking shifts are followed by what looks like a responsive action from the website, the URL or website which was previously only SUSPECTED as spam, might now be positively tagged as spam.

Read Bill's full article [seobythesea.com] (or the patent itself [patft.uspto.gov]) and see if it doesn't explain some of the odd behavior people have been reporting in recent months:

[aristotle]

The assumption I see is not that "all SEO is spam"

Tedster, maybe that's true within Google, but Matt Cutts seems to be equating SEO with spam in his public statements. For example, he originally referred to Penguin as an over-optimization penalty, then later started calling it an anti-spam action. Also, at the start of this thread, you wrote "If these transitional ranking shifts are followed by what looks like a responsive action from the website, the URL or website which was previously only SUSPECTED as spam, might now be positively tagged as spam.", a statement which again associates SEO with spam.

Of course, I don't agree that "all SEO is spam" myself, but (to me at least) Google seems to be promoting this idea in its public statements.

[Leosghost]

What is of issue here and what I am addressing is how the Industry does things.

Internet marketing community. It's about the Industry, capital I

The industry has been lead by the so called SEO and marketing experts, many of whom are far too cosy with Google, and or who don't actually know half of what they claim too, when telling others , clients or novice webmasters what to do..

They are the ones who've been in denial this past twelvemonth or more..or who have tried to shut down those of us who said it was happening , that G were randomising for multiple concurrent reasons..

So yes, it is about me, and some others like me, who don't post in order to get clients, so have no interest in posting, other than to aid others to see what we see..

Changing the occasional things on their site or pages harms no webmaster "directly", but SEO gurus can't get paid so easily if they can't claim to know secrets that the client doesn't or to be on first name terms with someone at the plex, or going to all the right meetings and events..

Some of us just look at the serps, especially outside our own "niches" and see what is happening rather than speculate and dissect the musings of Matt..

The SEO community, particularly in the USA is particularly incestuous, every one praising everyone else, everyone knows everyone else, everyone links to each others blogs and sites, and some are waaay too close to G to be objective, even if they wanted to be..

IMO, that is not good for their clients, nor the "industry", nor for objective debate, here or elsewhere.."conflicts of interest" do not aid one in being objective when scrutinising G or other search engine's serps in order to advise clients or novices..

[TypicalSurfer]

Leos ghost hit it pretty good, I don't care if people want to schmooze and BS their way into SEO fees when they truly have no clue about what is going on but when they adopt the social agenda that google is promoting (webmasters are the enemy, SEO bad, googles too smart to do anything "bad" etc.) then they have crossed over from serving an industry to being partners with the entity that seeks to destroy it.

Jumping on board with petty attacks on other webmasters is unfortunately the new norm for some who hold themselves out to be "expert", google doesn't even have to do it, plenty of proxies waiting with baited breathe to do the dirty work.

[1script]

Personally, I try not to make any changes but it looks like whether YOU make any change or not does not matter: Google will still push you down because there are OTHER people making changes after which they have to enter the transitional period and the random luck of the draw gave them a positive push, so they jump ahead of you regardless of whether they deserve it or not. Given that there are hundreds of websites vying for a place on SERP1, someone is bound to be making some changes EVERY DAY, so, presumably, if this patent-pending technique is in use, the SERPs are NEVER RIGHT. Especially given the ostensibly long (20+ days) transitional periods for each of the changing sites. (Thank you Bill Slawski, again, for unearthing the additional info)

So, there you go: Google is applying for a patent that says: Google's SERPs are never the true representation of what Google thinks should be the best Web page matches for any query. Taken another way, it says that there is never even a good will in trying to show best results - the entire "production process", so to speak, guarantees that you get a product that is less than perfect.

It's as if Coca-Cola had started to add salt to their drinks. They know it's better without the salt, but that's what the production process demands. Instead of changing the production process, they just add the salt. This would've been crazy to contemplate anywhere in the physical world, is Internet really so much different? Google can, with impunity, screw with their end product and be so proud of it that they want to take out a patent on how to screw the end product properly?

[seoskunk]

The SEO community, particularly in the USA is particularly incestuous, every one praising everyone else, everyone knows everyone else, everyone links to each others blogs and sites, and some are waaay too close to G to be objective, even if they wanted to be..

Yep its getting that way all over. Not what you know but who you know. But we are mere minions that follow the word of G.

I have to be honest, I am just here to learn and try and manipulate the search engines off the patents, to second guess G intentions would be impossible !

[gadget26]

I think we have just heard the explanation for both zombie traffic and domain crowding.

If google implemented the algorithm outlined in the patent, some pages would get randomly (and undeservedly) pushed up in the serps for certain search terms. This would generate bouncy and poorer converting traffic. (They may be good pages, just bad for the terms they are being artificially elevated for.)

Couple this with traffic shaping and you suddenly have, on average, very poor quality traffic visiting your site.

This might also explain the domain crowding (or whatever we're calling it) with the same site appearing over and over again. With all the pages being pushed down by this test, something has to take their place or there wouldn't BE a down. The duplicate pages could be one of the "lucky" ones that is being randomly moved up. (Which would explain the seeming lack of relevance to the search term.)

Note that this also means that, since domain crowding seems to be more prevalent in some niches than others, those niches must have more of this kind of testing being performed.

They can't 'fix' domain crowding because it's a necessary side-effect of their own testing.

So what I'm wondering now is: How do I make ANY change to a website without Google thinking it's in response to a ranking change? How many changes and of what scope am I allowed per year? Are news and topical sites exempt from this kind of testing?

The way ahead is very murky. But at least the past is making a bit more sense.

Cheers,

Jeff

[tedster]

Again, I don't read this patent as labelling ANY change as a sign of spam - it's a series of changes, each one apparently in response to a ranking change.

Yes, there's every chance that any first change may be followed by a set of randomized ranking shifts by Google. That's been happening for a while. So don't be casual about post-publishing changes. And if they need to be made (and sometimes they certainly do) now we know what might happen. Better to know than not to know, right?

Do I like this? Not one bit - not since it first apepared. But I certainly do understand Google's need to protect against ranking manipulation and reverse engineering.

[bluntforce]

"the rank of a document slowly responds to a positive change in its link-based information"

Seems to me the "link based" part might be a little more significant than has been discussed.

[tedster]

Not to mention "slowly" - eh? This patent definitely mentions link manipulation as one of it's targets. I do think a natural spike in backlinks will see a quick "freshness" response in the SERPs, but that kind of catching fire is not common in most areas.

[bluntforce]

I really do believe Google is trying to provide the best results for users. That may conflict with my efforts from time to time, but I don't think they're going to change their program to suit me.

I launched a site on Thursday that doesn't have a whole lot of text, it's mostly images that convey a lot of words. The SEO was done pre-launch so I don't envision any changes to the 70+/- pages, link growth will be measured and on topic unless there's some social boost.

Clean site with no baggage, it will be interesting to watch how it grows in the current environment.

[slawski]

@ Bluntforce

The description of the patent uses link-based ranking examples, but the patent makes it clear that it isn't limited to link-based signals:

While the description to follow may describe the conventional ranking algorithm as being based on just link-based information, the conventional ranking algorithm can be based on factors other than or in addition to link-based information. The phrase "ranking factor" or "ranking factors" might be used to refer to any type, or all types, of factors that might be used in determining the rank of a document, such as link-based information, an information retrieval score based on a match of a search query term to the content of a document, an indicator of document freshness, information regarding the manner in which a document's content changes over time, information relating to user behavior associated with the document, etc.

[bluntforce]

I certainly have pages that appear to have been labeled as possible spam due to on-page changes. I've been dwelling on reorganization of the page content, renaming the internal anchors and the page name, make another page to receive redirect juice. Currently, I'm just waiting as pages will rank well for one term but other terms just won't move.

There are pages I won't touch.
There are pages I'll change on-page factors.
There are pages I'll change off-page factors.
I don't consider it prudent to mix and match.

Thanks for your clarification and your original article, I hadn't seen much mention here of the link aspect, which I consider significant.

[diberry]

But this patent does make me wonder whether a manual inspection flag can also get thrown just by regular tweaking. If I were in a betting mood, I'd bet yes. In fact, this patented action may be an attempt to automate some of that manual inspection.

I would bet yes too. Because often, the same tweak that's good for visitors is also good for SEO. Even if Google's shockingly good at making that distinction, there's just no way they'll be 100% accurate. It's too subtle. (And from their perspective, there's no incentive to even try to be 100% accurate - false positives don't noticeably affect the user's experience.)

One thing many people noticed after Penguin was that sites which had obviously been untended for years rising to the top. I'm thinking the intent of this patent might explain that. I'm not sure if those sites have dropped now - people haven't been mentioning them so much, but then all the Penguin talk has slacked off over time.

[Maurice]

"Are you now or have you ever been a spammer" :-)

But it did not turn out well in the end for Joe Mc Carthy even though in hindsight there where areas - where the US had been pentrated (read the official history of MI5)

It was his hysterical acusing of all and sundry that did for him in the end.

[smithaa02]

Interesting patent...so if google scrambles/pushes down positive search results temporarily to confuse SEO 'spammers'...do they do the opposite for negative results?

Say the SEO does something really stupid...instead of google telling them they something really stupid by tanking the rankings...they increase the rankings...then tank them later?

[bordering]

I'm even more confused now...

G allows us to access data via WMT, and makes 'HTML suggestions'. Does acting on this, and perhaps fixing dup <titles> and <descriptions> mark us out as manipulative SEO-ers?

Also, my advertisers can log in and amend content. Some do it frequently. Almost always this is to ensure that the latest prices and deals are there for site visitors to find - abuse of this to keyword-stuff or do anything else dodgy is really unusual. So should I put a stop to the updating of content?

[Sgt_Kickaxe]

This is nearly a moot subject until the top 80 positions aren't entirely held by just three(3) sites. :-/

[backdraft7]

I just cleaned up my site (again) yesterday, finally removed ALL keywords, cleaned up descriptions and any superfluous text. Focused again on useability. Guess what? today the site is dead. In an unrelated event, I had filed a complaint to Adwords complaining of the penguin update and poor adwords performance post update...they sent me this:

I understand you are disappointed with the changes the we've made via the Penguin update. The idea was so that high quality websites, which are valuable to users, offering good quality content may be rewarded. This is so that user experience is enhanced.

Really? So what are there so many useless, one page, keyword stuffed trashed pages on page one now?

[richelectron]

I improved rankings for a certain page through A/B testing different variations of just the title tag with great results to a point - Beyond that it felt as if changes had no further effect and Google started displaying it's own variation of my page title in the SERPS. I know there are several factors involved, particularly changes to the algorithm over time, but it did feel like I had crossed a line.

[aakk9999]

@gadget26

If google implemented the algorithm outlined in the patent, some pages would get randomly (and undeservedly) pushed up in the serps for certain search terms. This would generate bouncy and poorer converting traffic. (They may be good pages, just bad for the terms they are being artificially elevated for.)

This is not the way I understood the patent. The way I read it was:
- a change is made by a webmaster in order to push the page UP
- algo calculates a new ranking which is UP
- the patent *may* be now applied (and more likely to be applied perhaps if the jump up is bigger or if perhaps the UP results from certain suspected actions such as new links, extra keywords etc as noted in posts before) so that:
--> either the ranking UP is delayed
--> or the ranking UP goes "smaller way up" first
--> or the ranking goes DOWN despite the algo calculated new ranking was UP

Each of these may have time delay.

Eventually, the page *may* move UP to its newly calculated deserved place - unless there is a further manipulation by a webmaster during this "patent applying" process, which made Google suspect that the original change that resulted in algo calculated UP was purposely obtained by spamming. In this case the new "deserved" place is DOWN (kind of penalised).

Reading the patent, I do not think that if your change results in ranking going DOWN that the patent would be aplied, but then perhaps I might have missed something.

[Dymero]

For a website I work on, I make changes to the page every so often because the content needs to reflect seasonal changes. Yes, for more popular items, I do check the rankings, and naturally I wonder if I messed something up when it suddenly drops five pages.

So I may go back and make some changes, but I'm not spamming. I might be editing content for relevance or making sure my keywords actually make sense to use for the context of the content.

Maybe from now on I just won't change the content on the page, or titles, or descriptions, unless there's something so starkly obviously wrong that not making a change could mean something worse is coming for my site.

[tedster]

From the patent:

17. The method of claim 9, where the rank transition function causes the transition rank to initially decrease and to increase, after the decrease, during the transition period.

That sounds like a potential ranking increase could trigger an initial decrease. And most of all, it sounds like the webmaster cannot predict.

[Nostalgic Dave]

This explains a lot of what I've been seeing on the pages I've been editing lately. I just overhauled one that was ranking on the first page for my target phrase, not just because I wanted to move it up higher, but because it badly needed an update (for better UE)... now it's nowhere to be found in the first 10 result pages. I haven't touched it since, and now I think I will just leave it alone and see what happens in the coming weeks.

[PR_Brian]

I may have actually seen this in action recently.

I have a commerce site built with products from an api. It has been live for over a year, has around 9 million pages indexed and brings in daily revenue. It's built under a 14 year old message forum.

On July 2nd I redesigned the api site completely. Same basic url structure, but the navigation elements, on page content, design, etc were updated.

About a week after the change, earnings stopped completely. Rankings tanked, and traffic was virtually gone. I "almost" put the site back to the old design, but decided to wait it out. I didn't make any further changes. Within a couple weeks, the number of pages indexed fell drastically (even though 100 percent of the old url's were still valid and working).

Fast forward to today. The site now has 12.2 million pages indexed, and sales are higher than ever.

Looking back, i'm wondering if I saw this rank modifying algorithm in action on this domain? I also wonder if by not making further changes I was able to somehow skip further penalization? If so, it appears to take about 45 days from start to finish.

Again, take this all with a grain of salt. it does fit the profile of what is discussed in the patent, but this is a single domain. hardly scientific.

[JohnRoy]

is it 12.2 millions words, or 12.2 characters indexed?

[CainIV]

If these transitional ranking shifts are followed by what looks like a responsive action from the website, the URL or website which was previously only SUSPECTED as spam, might now be positively tagged as spam.

Sounds like another way to throw random movements into the mix so that webmasters cannot anticipate. Another way to eventually eliminate reverse engineering.

Let's face it, slight chaos in Google rankings - more often - probably doesn't distract the user as much as it distracts businesses.

Reduced real estate, 7 packs and more chaos lead to record breaking quarterly results.

[Martin Ice Web]

PR_Brain,

didnīt you update your stock info in this time?


I wonder if there is some threshold that this "algo" would trigger.
Every business site has to bee updated in order of stock, laws, changes in item descriptions.
What about price compare sites, are they left outsite this "algo" ?

[simonlondon]

I am probably wrong, so please correct me.

In the list of spamming techniques in Bill's article, keyword stuffing and meta tag stuffing got my attention. To me it didn't suggest that title tweaks, or tweaks of anything else is a problem unless your tweak is a spamming technique, such as keyword stuffing.

Did I miss something? Why the discussion here sounds to me that all tweaks are a potential problem?

How I understand it is first you made a tweak, then Google check if the tweak is a spamming technique, if it is, this patent applies, if it isn't, then it does not apply. Am I wrong?

I didn't read the whole patent, so please let me know if I missed something.

[TypicalSurfer]

In the list of spamming techniques in Bill's article, keyword stuffing and meta tag stuffing got my attention. To me it didn't suggest that title tweaks, or tweaks of anything else is a problem unless your tweak is a spamming technique, such as keyword stuffing.

The "spamming techniques" described aren't definitive:

such as keyword stuffing, invisible text, tiny text, page redirects, META tags stuffing, and link-based manipulation.

and the catchall here:

No element, act, or instruction used in the present application should be construed as critical or essential to the invention unless explicitly described as such. Also, as used herein, the article "a" is intended to include one or more items. Where only one item is intended, the term "one" or similar language is used. Further, the phrase "based on" is intended to mean "based, at least in part, on" unless explicitly stated otherwise.

You need to think in terms of any change to a document that impacts its score.

This would certainly explain why forum posts from 2007 and old magazine articles can eat up a few slots on page one. Favoring the crusty stuff would be one obvious outcome.

You might be able to test this by logging into a forum and "rank modify" the ancient topic by replying to it and see if it causes a ranking change.

[edited by: TypicalSurfer at 1:40 pm (utc) on Aug 21, 2012]

[tedster]

The patent is full of language that uses the phrase "for instance". That seems to indicate that the techniques described may be used in other situations, not just the specifics being mentioned. Since the title tag is such a sensitive ranking factor, it seems logical to assume the patented techniques could be applied there.

Even more important, in past discussions here webmasters have reported seeing this kind of ranking fluctuation after they made title changes.