Traffic Shaping / Throttling Prior to Deindexing

[dresden87]

I launched a few "test" sites back in April to see what could happen if I played everything as safe as possible. 3 sites in all. I hired well-known freelancers/experts/professors in various niches to produce great content for these sites. We worked the PR angle really aggressively and most of these sites ended up with between 10 and 15 backlinks each, pretty much 100% brandname/url anchors from great sources.. edu/gov resource pages, interviews on forbes/cnn.com, etc.

Anyway - the sites followed a very strange pattern overall.

At first everything was cool.. these were EMD domains and ranked just fine right out of the gate after the first deep spider. Then things got a little weird. After the initial links started landing the homepages started going away and random subpages started ranking in the 30-60 range for the homepage keywords.

Then everything went back to normal for about a week or 10 days, depending on the site. Then it got weird again :)

During the next couple of weeks there were lots of big drops.. keywords that were in the top 10-20 dropping down to 150 or past 200. It was really surprising the day-to-day drops of hundreds of spots.

Now, starting about 2 weeks ago I started seeing some weird traffic patterns.. the sites wouldn't get any traffic at all during the day. Say, 7am to 7pm, no traffic. Then overnight there'd be a ton of visitors for the target keywords from Google. US traffic too, not overseas. I was getting a lot of the &cd=2 (4, 12, etc) strings appended to the Google referral url, so that seems like a sure bet of testing being conducted.

Over the weekend all 3 sites were completed dropped from the index. None of them pull up via [domain.com], only site:domain.com. No alerts in WMT at all.

Anyone else had any experiences like this? I'm beyond frustrated with this.. I feel like we approached these by-the-book.

[seoskunk]

I had a couple of sites drop like a stone this weekend too. I am trying not to be concerned and plough on regardless I suggest you try the same. Nice to know I'm in good company, thanks for a great post!

[nisiwi]

Google dances are usual occurrences, however I've been seeing strange things in SERPS recently, odd index counts being returned that change as you navigate paginated results or even changes by the day.

Massive throttling of search traffic with strange peaks and troughs, mostly traffic falling off a cliff while at the same time some non existent content is being ranked ?

With the escalating number of threads here and elsewhere discussing this their appears to be a distant lack of consistency right now with current updates, so the dust is well and truly in the air still.

Its a bit like walking on broken glass at the minute, so I'm inclined to stop walking and watch things unfold until we have something concrete and can clear the glass for a more comfortable journey.

The fact that google is returning results when using site:domain.com at least means that content is indexed and considered worthy of staying in the index. However, as far as answering the question about why its not ranking....no clue right now.....keep reading and listening and going over what you've done just to cover any probable or glaring mistakes.

Things are bound to become clearer eventually....sooner rather than later I hope.

[Bewenched]

I concur with the strangeness this weekend... our traffic dropped and then was flat and remained flat until today. I was starting to think that there was something wrong at our server hosting company.. thinking that maybe they had put some type of limit on us, but after a call with them it was definitely in the traffic being sent.

But like I said, the traffic came back today with about a 20% jump... hope it sticks.

[Sgt_Kickaxe]

You've got to analyze the traffic in order to make any sense of the overall condition of a site. Which pages? Traffic Source? Link profile of those pages? Web search or image search? etc.

Your pages may be a book to you but they are a bunch of individual pages to Google, they don't all rise and fall together regardless of what the book does.

[nisiwi]

Unless the shelf the book is on is broken ;-)

However, I would agree with Sgt_Kickaxe that assuming your SERPs results are stable and you know what's actually happening consistently then its time to work out the problem.

If your site is indexed but not showing up for searches, you might have some kind of penalty, especially when you say that doing a "yourdomain.ext" and a "www.yourdomain.ext" search in Google, didn't return your site as any site that hasn't been hit with a Google penalty should appear first in the results for that exact match search.

It's likely that its been Penalized by a Google Search Quality Engineer. If your site has violated the Google Webmaster guidelines but still has some redeeming value, your site may only be affected for some period of time and the penalty will then lift, it may not. So its time to take a look at what may have caused it;

Bad Neighborhoods - Linking out to gambling, #*$!, hacks, torrents, and similar sites are seen as bad.

Buying links - Iffy

Selling links - More iffy.

Sneaky redirects - Redirecting visitors but not SE bots.

Hiding content - Serving different links or content to SE's and human visitors.

Unnatural link patterns - Your back-links triggered an algorithmic flag. Typically due to keyword stuffing. By far the most common cause of algorithmic Google penalties. Its estimated that if more than 90% of your links are for a few highly targeted keywords, it'll trigger this.

Your only option is to fix whatever you think Google caught you doing and you'll have to try and work that out unfortunately. You then either wait for a few months and hope it comes back OR submit your site to Google for reconsideration which could back fire on you if you are still doing something dodgy that wasn't caught during a manual review.

If you've not done anything iffy then taking it up with the big G is probably worth it, once you've given your overall strategy a good going over.

[Robert Charlton]

We worked the PR angle really aggressively

This makes it sound like enough of the links might have been manipulated to raise a spam flag at Google. I'd be hesitant, though, to be quite as precise as nisiwi is being about what's going on. I'd look at precisely what you did to get your links, though.

[nisiwi]

I think being cautious about any solution makes a lot of sense Robert especially considering the state of things.

Just an opinion as to any probable causes so that you have at least some kind path to travel to find a solution.

I did mention "Unnatural link patterns" which as you mention could equate to "We worked the PR angle really aggressively".....

Depends on how aggressively though and what was done and whether this is actually responsible for causing the problem....who knows without more data or analysis, just guessing.

[dresden87]

Thanks for the responses everyone!

When I say we worked the PR angle.. I mean we used HARO/Profnet and Cision databases in order to get interviews and coverage in the media. We didn't control the placement (or anchor) on these links, nor did we pay for them. Most of them are linked to us using the brand name or domain.ext.

I understand the risks associated with doing dodgy things.. but I've been doing this a loooong time.. nothing dodgy on these sites (can't say the same for others, but those all got caught in the google zoo :)

[nisiwi]

Hear what your saying dresden87. Like I said the google game at the moment is not clear cut at all. Lots of stuff getting caught in their dredging net from what I'm seen from black to white. Try taking it up with the G team or wait it out.

[indyank]

Is there any recent example here of launch of a new site that went on to do well in the SERPS? It would be great to read their detailed story on how they went about it.

Google has massively increased the number of changes they introduce for the past couple of years. Considering this, I don't think it would be that easy to tell the correct reasons for the OP's question, though he has given a detailed account of his new launch. People are only going to guess their reasons here, as several known and unknown factors are at play. Even the popularity of people behind the site/startup is a success factor these days.

Moreover, I think any meaningful study of SERP rankings can be done only after a min. period of 6 months to 1 year from launch. So if I were you, I will only focus on studying users on the site and focus on improving their experience.

[indyank]

and most of these sites ended up with between 10 and 15 backlinks each, pretty much 100% brandname/url anchors from great sources.. edu/gov resource pages, interviews on forbes/cnn.com, etc.

these were EMD domains

I do see a problem there if the brand name happens to be a keyword.

[nisiwi]

If that is a probable cause indyank then anybody with a company name and URL that contains keywords explaining what they do may be getting penalised for no reason.....that would be very concerning...Airport Taxis Ltd or Promotional Merchandise City or similar....can you imagine.

dresden87 have you done any analysis of backlinks to see if you've been a victim of the so called "negative SEO" effect by a competitor (its a very scary thought if this is actually possible).

[indyank]

What I am saying is it is easier to trip in that scenario. There is a way to use keywords in domain name.

For example, I would say one important reason why wpmu considered to be a site with reasonable authority got trapped was because of their brandname anchors having "wordpress" in it. Google would have assigned it to the actual wordpress site and this being a popular name, it also happens to be a keyword. But i will have to alert you that this is my own judgement and I haven't read anyone stating this as a reason.

[dresden87]

@nisiwi

I have, and there is no evidence of that on any of these sites.

[seoskunk]

Maybe give it a week as there def seems to be some changes afoot. Maybe something will change in next few days

[bwnbwn]

Question before launching these EMD domains how were they used, how did you come to own them, and how fast were all these powerful links added?

[Robert Charlton]

Are these EMD domains by any chance interlinked, and/or do they have common inbound linking sources?

Also, how about hosting? Are the sites on the same server?

[Hope_Fowl]

Do you own the brands for those "brandname" URLs? If you're reviewing/selling Black & Decker products, maybe Google has decided some other site is more authoritative for that brand.

[dresden87]

@bwnbwn - I picked 2 up at auction and 1 fresh registration. They were added over the course of a few months.

@RobertCharlton - Not interlinked at all, no common inbound sources. All 3 are hosted at separate hosting companies on dedicated IPs. I've checked, the IPs are not blacklisted.

@Hope_Fowl - When I said brandname I meant my 'OWN' brand, so if the url is dresden87.com then the branded anchor would be Dresden87 or Dresden87.com

[spreporter]

Delete the pages, start all over again with new domains (with top and unique content), put just one link somewhere in a small PR site and wait a few months to add a few more backlinks...... With this natural strategy I have seen many new sites( few months to one year old ) ranking for hot keywords in my niche.....my 2 cent....

[Robert Charlton]

spreporter - Extremely interesting comment, as it resonates with an observation posted by jmccormac in this recent discussion.

Search engines need time & other signals to confirm a site is "fantastic"
http://www.webmasterworld.com/google/4467831.htm [webmasterworld.com]

jmcc noted...

A new site has a very specific link profile and link acquisition rate. The process of a new site gaining new inbound links is, where there is no outside SEO campaign involved, more like a process of accretion in that links are gradually added an the site goes through a few bursts of new inbound links.

Putting those observations together with your comment, it's possible that even if the links themselves were natural, the rate of acquisition was seen as not natural, particularly because it apparently wasn't sustainable.

There is a component in Google's historical data patent that looks at "spikiness" in link growth patterns as potential evidence of too much coordination, and thus as possible spam.

[scooterdude]

I am not a Google defender, however I reasonably certain that they rarely DEINDEX a site completely,

So perhaps you could confirm that all the contributors to the sites were fully aware that they were contributing to the sites,

That the subject matter of the sites would be one adsense would be happy to run with

[Web_speed]

Google has massively increased the number of changes they introduce for the past couple of years. Considering this, I don't think it would be that easy to tell the correct reasons for the OP's question, though he has given a detailed account of his new launch.

IMO, call it what you will, I've seen enough of this to call it by its real name, "shuffling". Pure random shuffling designed to keep everyone guessing, bleeding and ad spending... good luck.

[tedster]

call it by its real name, "shuffling". Pure random shuffling designed to keep everyone guessing, bleeding and ad spending

I agree with the word "shuffling", but not with the design that you are assuming. I think the design is statistical sampling: rapid testing of user response to various ranking schemes. And yes, it changes so fast you can only guess at the logic behind it, but I think there IS logic rather than randomness.

[Web_speed]

@tedster

The wired, once in every 48 hours (now), out-of-whack index changes all started once the FTC began sniffing around.

I have no doubt we've seen nothing yet...the "shuffling" will probably continue and will get more and more aggressive as time go by... it is not about quality anymore, it is not about quality for a few years now.

Sorry but the only logic i can see is in their two digits growth numbers quarter after quarter while the "index" is such a mess and almost everyone else are bleeding...

[spreporter]

"it is not about quality anymore" yes and no, but, the top sites have a few common things,1 historical data, 2 trust rank, 3 clean profile, 4 high CTR that can be also achieved through long time adwords adds(i monitor big brands in a certain industry that now ranking at top spots in organic results even for 1 word keyword because of daily top spots in adwords)
Links work like always worked for the above old sites, for new ones I will rather stick with my above opinion. As about quality (design, clear coding, no html errors) that comes as second or even no option.

[jmccormac]

@dresden87 @spreporter @Robert Charlton The link acquistion rate of a new site does follow a particular pattern but that pattern is something search engine operators are more likely to see than SEO people or webmasters. The basic natural link development/acquisition pattern (as seen by a search engine operator)for a new site without an SEO campaign runs like this (the main post is in the supporters section of the site here):

Stage 0: Site Launch
Minimal inbound links. Mainly outbound analytics and useful sites links.

Stage 1: Friends and Family
Inbound links from the websites of friends (often strongly linked in the webmaster's social network) and family.

Stage 2: Businesses and acquaintances
Inbound links from non-competing businesses in the same area (members of industry or chamber of commerce).

Stage 3: Customers and blogs
Inbound recommendation links from people who have used the site (discussion forums or blogs) or found the business useful.

Stage 4: Directories and other resources
Inbound links from trustworthy directories and people who have favourited the site as a useful link.

This is roughly the natural link acquisition pattern without an SEO campaign behind it however the timeline varies. Some sites can take months or years to acquire inbound links. When, as a search engine operator, you see a blizzard of inbound links to a relatively new site it looks rather odd and unless it is a driven from a mention on a major media site or authoritative site, it is typical of an SEO campaign or worse.

The worse case is a meat bot campaign with links coming in from sites, typically blogs, that don't have anything in common other than they (often) allow automatic posting of comments with no editorial approval or curation. That means that the inbound links would be sitting right in there with, potentially, a set of drugs/pron/warez links. I'm not sure how Google handles this kind of thing but it is a very risky thing for a search engine operator to allow this kind of pollution into their index so what can happen is that these open comment blogs can get the comments deindexed once identified and if the site is really problematic (ie an abandoned/derelict blog where there is little activity after the obligatory "hello world" post followed by the "ain't we great?" post) then the site will be deepsixed. That could mean that any outbound links from such abandoned sites are potentially suspect. Again it gets back to the blizzard of new links - most genuine sites (unless they are directories, large forums or data sites) do not have large numbers (hundreds of thousands) of outbound links.

Links seem to have their own Social Network. People naturally link to either the index page or the particular page in which they have an interest. The words used to link can often be as simple as the site or domain name. SEO people tend to use keywords for anchor text but natural links (as opposed to bought or artificial links) may not so there's a possibility of greater variation of anchor text and terms. That cliche about all politics being local can apply to many websites too in that if they are more than brochureware sites, they tend to have a localised or identifiable set of users that can be associated with a niche or a geographical area.

It does take time for a search engine (again not sure how Google handles this exactly) to establish a baseline for how a site is updated and used so SERPs and positioning should vary during this period. But if the PR angle has been played heavily then there is a possibility of meat bot links and and identical anchor text links that could have flagged the site for review. The most important thing to remember about a "by-the-book" approach is that search engine operators may be using a different book. :)

Regards...jmcc

[Robert Charlton]

My emphasis and question mark added...

...high CTR that can be also achieved through long time adwords ads (i monitor big brands in a certain industry that now ranking at top spots in organic results even for 1 word keyword because (?) of daily top spots in adwords)

..."ranking"? ..."because"? There may well be all sorts of correlations. Some studies have suggested a symbiotic effect between top organic and AdWords CTR, but I don't think we can simply extend this automatically to a ranking cause.

We might be able to explain ad position, eg, with ongoing searcher satisfaction with the brand, or perhaps it's due to excessive ad spend. We might similarly be able to correlate organic rankings with exposure in all channels and thus with traffic derived from that exposure... but I think it's safe to say that without user engagement and satisfaction, exposure by itself may not translate into better rankings. I don't want us to take this discussion off topic, though. How do you relate these factors to the question of why these particular sites might have dropped?

With regard to the shuffling, I'm very curious about the traffic patterns the OP observed. They may suggest another aspect of the Google algo evaluation process, and they may or may not relate to the patterns of traffic shaping and throttling that we've discussed as "traffic shaping"....

Zombie Traffic and Traffic Shaping - Analysis
http://www.webmasterworld.com/google/4437835.htm [webmasterworld.com]

There's no question in my mind that constant shuffling we see in many forms is related to high speed statistical sampling, and I very much agree with tedster that there is some very heavy duty logic behind it. Beyond the inevitable latency that's going to occur in a multi-database system of Google's size, it's in no way random.

[Robert Charlton]

jmccormac - Thanks very much for posting that.

(the main post is in the supporters section of the site here)

For those who are WebmasterWorld supporters, the more extended post from about a month ago is available in this thread in the supporter's Google SEO forum...

What is a natural link profile?
http://www.webmasterworld.com/opengoogle/4452832.htm [webmasterworld.com]

For more detail on the idea of "spikiness" in the Google historical data patent that I was referring to, see this archived discussion from 2005...

New Google Patent Details Many Google Techniques
http://www.webmasterworld.com/forum30/28814.htm [webmasterworld.com]

From the patent...

While a spiky rate of growth in the number of back links may be a factor used by search engine 125 to score documents, it may also signal an attempt to spam search engine....

Just to throw out some thoughts for consideration... were the patterns of link growth to the sites in question far enough outside normal patterns to have been flagged as "spiky", thus prompting further tests?

And were there any patterns in the "shuffling" or traffic observed that might suggest what Google was trying to measure? Any stats on visitor engagement with the site during the times of the test? Was Google trying to evaluate, eg, whether the notoriety the site was receiving was based on wide appeal or on heavy duty promotion?

I feel like we approached these by-the-book.

Possibly, the definition of "by-the-book" could be changing.