Welcome to WebmasterWorld Guest from 18.104.22.168
In June 2007 Site A lost about 25% of its' traffic from Google after the major update around that time. It looked like an OOP as we lost rankings for exact phrases while retaining rankings for non-exact terms. Before we figured it out things got worse in October 2007 when major pages from the site were being removed or -950'd from the results on a daily basis. At the time we had about 30 links in the footer to internal article pages and the -950 thread on here suggested that could be culprit so I removed them. Immediately (the next day!) our pages started to return to their former positions and even our pages that had been suffering since June 2007 could be found again for exact phrases. Within two weeks traffic returned to it's pre-June 2007 levels. Problem solved so we thought.
In December 2007 I hired an SEO to check over all my sites for any potential problems. He felt Site A looked okay but recommended we improve our trust rank with some good quality links, so we choose some phrases to target and set about it. He did have a concern that linking from two of our other sites may be a problem (Site B and C which I'll come onto later) but wasn't certain that removing the links would do more good than harm so we left that alone.
In Feb 2008 we began to notice our rankings for the pages we had focused link building on gradually falling from their page 4 positions and dropping a page each week or so. This didn't have any noticeable effect on traffic but was a concerning sign, especially as all other rankings remained exactly where they had been for years.
In March 2008 we had to change hosts from our Canadian hosts and decided on a US company. All our sites have .co.uk domains and are aimed at the UK - the sites were hosted in the UK on shared servers until 2005 when we moved them to two dedicated servers in Canada and all continued to go from strength to strength for years so we had no reason to believe server location mattered. As before, all domains were on separate, non-consecutive IP's.
During April 2008 traffic dropped by 30% and rankings on the pages targeted for link building had dropped to effectively nowhere so at the start of May 2008 I called a halt to the link building. By mid June traffic had returned almost to normal (about 10% below) and the link building pages recovered to about page 10.
The peak season for Site A is November and December. During these months in 2008 our traffic lifted as normal but was constantly 25% lower than in 2007. I also noticed that for certain long'ish tail search terms our pages were bouncing in and out from high page 1 positions (which we'd held for years) to nowhere in 2-3 day cycles. The keyword density for the bouncing phrases is quite high but no higher than other phrases which are not suffering. The site content is mainly product information so certain words can be inadvertently repeated alot in some categories.
In Jan 2009 traffic has remained 25% lower than normal for this time of year and certain pages for certain phrases are still bouncing in and out.
A few other things to mention......Site A grew from 200-500 products (one page for each) from 2003 to 2006, and continued growing steadily to what is now around 2,000 products. So watered down PR could be an issue.
Site A has a menu of about 50-60 text links, all with relevent anchor text. I've read about mega menus and the potential problems but I'm not sure if ours is big enough to fall into this category. It's always been like that.
Interlinking is the other potential cause I haven't gone into in detail yet. Site A is our main ecommerce site selling products in a niche. Site B is an affiliate site (set up before Site A) featuring all sorts of products but one of it's eight sections is aimed at the same niche as Site A and contains our products and those of one of our competitors, deep linking to the product detail pages and category pages. Links to our competitors products are through an affiliate agency, links to our products are direct to Site A. Within this section of Site B we also link to various relevent sites on every page (including Site A).
We also have Site C which is another affiliate site but purely in the same niche as Site A (set up before Site A). It contains our products and our competitors and links through in the same way as Site B.
Site B and C link to Site A, but Site A does not link back.
The only other thing to say about Site B is it has always been my experimentation site and consequently does get a rough ride in Google (lots of chopping and changing, old pages ditched, lots of new pages added occasionally). Most of the content on Site B is provided by a third party price comparison site and we have a web directory where you can advertise for free in return for a link to us, or pay to be listed. All outbound links have been 'nofollow' for 6 months, except those to Site A.
I spend ALOT of time reading threads on this site so I know I've mentioned loads of potential causes (US hosted UK sites, interlinking, mega menus, duplicate content, link building, over optimisation, and so on). I've even considered the possibility of DNS hijacking which seems to match with a consistent 25% reduction in traffic. My sites are potentially guilty of them all but I'm trying to rationalise this into what's realistic.
I've always tried to keep Site A 'clean' as it is an ecommerce site and our main source of income. Everything (except the footer links - hands up to that one) has been done with the visitor in mind. Over the years that approach, and some basic optimisation, have got us a long way.
Site B is another story. That has always been an SEO testing ground and traffic fluctuates massively. Google has shown encouraging signs that suggest it could fly if we can clean it up (it's good now but there is a lot of historical shrapnel laying about).
Site A is the priority though. Are any of these issues causing a problem, is it a bit of everything, or is everything fine and I'm just being paranoid?
We have tried to figure out why, and can come up with no single logical reason. There is more competition now, so part could be a case of not us dropping so much as others coming up more (?).
But some things just don't make any sense. One section of a few pages for some pretty uncommon repair parts - some of the pages are at #1 for that part number, but others that are nearly identical but for description and # are being beat by Adsense/scraper sites.
We also have a couple of other non-ecommerce sites that do pretty well and link to site A. And they still rank quite often in the top 5 on page 1.
So I am beginning to wonder if it is just ecommerce sites that are taking this kind of hit.
I read so many threads on here where people ask what, to me, are fairly basic questions and get masses of responses from people who seem to know what they are talking about (and alot who don't!).
This is a complicated scenario and a long post, admittedly, but I thought at least one person might have a view, especially as it encompasses numerous issues that are normally only considered in isolation on here.
Come on. Have all the SEO experts on here gone shy all of a sudden?
Confirmational answers often end up in discussions of grey areas (such as "should I shrink my menu from 100 links" will get two or more lines of thought).
Yours is a case study. Hard to unpick. And frankly some members here would be expecting a hefty fee to answer you. You might avaoid the fee and get a better analysis by paying for membership and posting in the site review [webmasterworld.com] forum.
Are any of these issues causing a problem, is it a bit of everything,
Simple answer...maybe yes, maybe no!
The fact is that you have recognised possible issues therefore if you know of these then Google certainly does.
I assume your target market is the UK? My first recommendation would be UK servers, underestimating Google's attempts at geo-targetting these days is perilous.
Interlinking should not be a problem so long as there is little duplication. I have many sites linked together, they are all about the same widget subject, albeit from different countries, and Google has always ranked them fairly although I have to admit some of them more slowly than others.
Your menu is not a long one by any definition.
I had one UK site last year (10 years old) that suddenly started bobbing about all over the place and I'll be honest I have absolutely no idea what caused it other than faulty spidering by Google. By this I mean the site and all links should have been shown as example-exmaple.com yet Google insisted it was exampleexample.com.
In the end I rebuilt its entire link structure by hardcoding every URL to ensure example-example.com, once this was done it recovered very quickly and is doing better now than ever.
Another interesting point to note is that BEFORE the problem the site had 70% US traffic whereas now it has a minimum 50% UK traffic and US traffic is down to 25%.
Without seeing your sites it will be quite difficult to suggest any improvements.
Thanks also HuskyPup. Yes, the UK is the target market but it's a .co.uk domain and I have stated the geographical location in my Google Webmaster Tools account so I hoped that would mitigate and potential server location issues. Shame, hosting is much cheaper in the US, even at today's exchange rates.
Nice to know a 60 link menu is not considered 'mega'. That's one off the list!
My .com's on UK servers always rank better for UK searches whether that be the Googles .com or .co.uk site (It is currently ranked 6th out of 170 million sites for the biggest searched word and the plural version one as well).
I have not tested USA sites on USA server and USA searching but UK searches for UK sites on UK servers has always been higher up the list.
You say your products are growing steadily? Is it a gradual few a week or months of nothing and then a bulk going on? Slow and steady is always better than sudden big surges.
I used to test on another version of the domain like you but I never linked to the main site in case the test site got penalised and that rubbed off on the main site. This sounds like it might be a reason to me (especially if you mirror even a small amount and that mirroring is picked up).
Finally you admit there were some footer linking issues that may have caused it. Fixed those first and go from there.
You will find many having had no bad experiences and then those, me included, who have had extremely strange ones.
I am in the UK and all my domains are registered with the same information in the UK however 20+ of my US server .coms Google has for the last year decided they are in South Africa. My .us domains on the same server it has decided are in the US and are never ever seen on any results outside of US Google.com SERPs, not even supposed UK Google.com SERPs.
I have seen only this last month a UK registered US hosted .com only served in the US results. We have actually duplicated this site on a .co.uk hosted in the UK and wham, it's there in the Google.co.uk SERPs. The .com is going to be 301'd this week.
Another colleague with a 12 year old UK registered .com moved his site to a US server last year and has completely disappeared from the Google UK SERPS.
Google seems to have no problem in identifying my .in and .cn site as being from their respective countries.
I have had no problems, so far, with any of my UK hosted .coms and the UK Serps.
Shame, hosting is much cheaper in the US
Hmmm...well, you get what you pay for and if you're running a serious ecommerce site aimed at the UK then you need every optimisation trick available. Incidentally, without starting a "my host's cheaper than your host argument", what do you consider as "much cheaper"?
Product/page growth is around 10-25 a week on average (which equates to about 1% of the overall page count), with some products being discontinued and the old pages 301'd to the main category page or an alternative product if possible. I don't think we're adding pages too quickly but we have added alot since the site was at its' peak in 2007. My concern is we may have watered down PR too much for the incoming links we have (which hasn't changed much since 2007).
Site B is not a test site, it's real, but I tend to experiment with different SEO approaches there. All the products on ecommerce Site A are also featured on Site B as part of the much wider range of products it promotes. Is it enough to cause a problem though.
HuskyPup, it's a .co.uk domain on a US server however we do have the .com domain as well. We don't push it and all traffic comes to the .co.uk. The .com is not redirected to the .co.uk however (perhaps I should change that!).
I've found US hosting companies much better value (a quarter of the cost of a UK host for the equivalent server) and the service from the company we're with is better than any UK hosting company I've ever used.
I have the same as you, Site A is my main .com site and Site B is my co.uk live/test/play site. It has different carts, different SEO etc compared to the main one but I never link to the .com in case the co.uk ever gets a negative hit.
As for the $$ all I would say is for the sake of a few extra pound a month you'll get a UK site on a UK server and I would imagine that as long as everything else on the site is clear and blackhat free you'll see a increase in the SERP's.
As you saw from my previous post I haven't gone down the USA server route for a few years and my rankings have never been as strong. Not 100% evidence I know but it's enough for me to keep things as they are :-)
If you have a mirrored site on .com, either redirect it or prevent indexing. I'd redirect.
PR dillution- if your % increase in URLs is not matched by a similar % increase in links, you will slowly be dilluted.
Actually, that might be it. Your backlink profile might be going stale if you have not acquired new links in over a year, possibly approaching 2 years.
Also, I've re-read my post and I might have come across a bit harsh. Sorry about that.
I'm certain duplicate content is an issue for these sites but I'm not completely sure how Google handles this. My understanding is that small scale duplicate content within a site or across two or three sites owned by one person is simply filtered to show one result (the most relevent, highest ranking page) but the sites suffer no negative effects otherwise. Traffic is effectively shared between them. That's my understanding anyway.
I noticed when Site A had the -950 problem that pages from Site B and/or C took the positions vacated by Site A. I have also recently noticed pages from Site A and B alternating for 4th position on one term (like Google sees them as equally relevent but from the same source, so rotates them).
If I NOINDEX what I consider duplicate content on the weaker sites will Site A perform much better or simply pick up the traffic the other sites lose?