Hi,

I've been involved in a major ecommerce site relaunch, redesign and backend and have been involved in several successful ones before. On the whole things have gone great. Of course there are some minor bugs to fix, rendering issues etc but my real concern is a significant decline in conversion rate BUT only for IE and Edge users. All other major browsers conv rate has increased.

Trawling through analytics and the site itself in those browsers there is no obvious sticking points e.g bounce rate has improved on all browser inc IE & Edge, no particular point in the funnel is suffering more than others, no particular page or landing page is doing much worse than others etc. it just seems like all the kpi's (except bounce) are down a little on the 2 microsoft browsers, which has a significant impact.

One of the things that changed on relaunch is that we moved to https and im wondering if that might be causing an issue with IE / Edge security or privacy settings or that functionality on the site that relies on cookies is being affected, though in testing we haven't come across anything.

I'd appreciate any advice or insight.

Thanks,

Andy

Hi,

Thanks for you advice.

I don't see what you see with Amazon, in fact I see something very strange - http://d.pr/i/1ik1T

The product reviews sit in their /product-review/ directory and these urls appear to be no index, no followed. They also have a canonical back to the original product page - why bother if the url is no indexed? If I've got that right it would seem Amazon don't get the full value of the millions of pages or product reviews they have?!

This from the Google url you referenced seems to sort out my problem thanks!

rel="next" and rel="prev" are orthogonal concepts to rel="canonical". You can include both declarations. For example, http://www.example.com/article?story=abc&page=2&sessionid=123 may contain:

Hi,

I help manage a large site that uses a lot of params for tracking, testing and to help deal with paginated content e.g. abc.com/productreview?page=2. The paginated review content correctly uses rel next and rel prev tags to ensure we get the value of all of the paginated review content that we have.

The volume of param exclusions I need to maintain in Google & Bing Webmaster tools is getting clunky and frustrating. I would like to use self referencing canonicals, which would make life a lot easier. Here's my issue:

• If I use canonicals on the review pages the paginated content urls would also use the same canonical e.g. /productreview?page=2 pointing to /productreview I believe I am going to lose the value of those reviews, even though they use the rel next rel prev tags. BTW airbnb do this - do they know something I don't, don't care about the paginated reviews, or are they doing it incorrectly, see http://d.pr/i/14mPU

Is my assertion above correct about losing the value of the paginated reviews if I use self referencing canonicals? Any thoughts on a solution to clearing up the param problem or do I have to live with it?

Thanks in advance,

Andy

Hi,

I've seen a big ranking drop for many major terms, for a particular site, just on Google. This happened Fri 20th or Sat 21st just gone. I don't see any news on an algorithm update over the weekend.I had changed many of the sites major page title protocols 2 weeks ago but a) I would have expected any negative effect before now and not all at once b) the protocols were carefully crafted to avoid traffic drops for major terms and c) i'm seeing traffic drops for keywords that still start at the beginning of the page title d) im seeing drops for some pages which are still using the OLD page titles. I had even tested the protocol on a number of pages in advance to ensure it wouldn't cause problems. As a bit of background - the title protocols were changed to make them more user friendly and less keyword heavy. CTR from search improved so was hoping for better not worse rankings!

Ideas, gratefully appreciated.Andy

Some points not mentioned (I think) yet...

The number 1 way to get links to a website is traffic...the more people that see the site, the more chance of it getting links, social shares etc. If PPC traffic & organic traffic (& other types of traffic) are not going to the same domain they're likely to get less links per domain, which reduces their chance or ranking well organically, which is contradictory to what you are wanting to achieve. If I were the client and I were getting links to non indexable pages, I wouldn't be happy. I would investigate if those PPC landing pages actually do have links/shares and if so use that to support your case i.e. those links could have been pointed at the domain you are trying to improve organic rankings for.

Remarketing via PPC is a huge opportunity via display and especially RLSA. If I was running their PPC it would really annoy me that the audience that I could remarket to is reduced  because the traffic is split over domains and not consolidated in 1.

Analytics is really messy. I'm a firm believer that any web reliant business needs to understand and use the data available to them in order to grow, especially if traffic/online revenue is significant. Splitting analytics (GA or otherwise) and KPI's over multiple sites, analytics accounts etc is not conducive to managing and benefiting from that data.

Sub domains - might not be relevant, based on why they want to use off domain pages but could you suggest they use sub domains instead of different domains? This way they can still promote/operate separate sites but you'd have the advantages of 1) capturing link equity 2) increasing your remarketing audience 3) having all your analytics in 1 place.

A friend has a large site and his dev team are about to launch with the sitewide addition of trailing slashes on urls, which are not on the original site. Both versions of the url work e.g. /example and /example/ and the dev team is reluctant to change back as it will take 3 days (developers hey?)

My SEO mind tells me this is a bad thing. Yes you could:

• 301 from non slash to slash but sitewide I don't like the idea because of the small loss in link juice x lots of pages.

• You could canonical from non slash to slash but the non slash is the original and canonicals are supposed to point to the original version of the content.

1. Any thoughts appreciated

2. Couldn't the devs use the htaccess file to force the / version to the non / version?

Thanks

Pages dont just get equity from external links of course. If a category page has 10 links to it the product pages linked to on that page benefit. The wholesale drop in rankings isn't because every page had an external link to it.

Thanks Alan, yes they have good external links to many pages. They retail a very niche product and have a lot of forum, review, social type links. It might be though if need be they just have to focus mostly on 301s for the pages with those links. As best practise I am in favour of 301'ing regardless of external links as the link equity gets messed up and causes ranking issues, as in this case, as well as sending a signal to the engines about the amount of wasted resource they will use crawling a site with 1000s of 404s.

Thanks Donna & Luis. Luis is right i'm looking for a way for this not to be a mammoth manual task for their developer.

Thanks, the regex is a good idea and might be part of the solution for some urls at least but there seems to be some discrepancies in logic between old and new product urls and some of the new product urls are actually still the same as the old (which of course is fine).

Thanks Luis, unfortunately neither 1 or 2 are ideal.

1. I don't think there is much logic in the change of url structure between old and new product urls which makes that idea impossible.

2. Thats going to be a last resort

Andy

A friend has relaunched a website but his web guys (he didn't consult me!) didn't do any 301s and now traffic unsurprisingly has tanked.

The old site and database no longer exists and there are now 2000+ 404's.

Any ideas how to do the 301s from old urls to new product urls WITHOUT it being a massive manual job?

Thanks. BTW in my original question I used the word spanner a couple times when I meant hammer which probably confused some people. I've edited it. I prefer hammers anyway

Hi,

SCENARIO:

A site has say 3 examples of the same product page but with different urls because that product fits into 3 different categories e.g.

/tools/hammer

/handtools/hammer

/specialoffers/hammer

and lets say the first 2 of those have the canonical pointing to /specialoffers/hammer YET that page is now redundant e.g. the webmaster decided to do away with the /specialoffers/ folder.

ASSUMPTIONS:

• That is going to seriously hamper the chances of the 2 remaining versions of the hammer page being able to rank as they have canonicals pointing to a url that no longer exists.
• The canonical tags should be changed to point to 1 of the remaining url versions.

As an added complication - lets say /specialoffers/hammer still exists, the url works, but just isn't navigable from the site.

Thoughts/feedback welcome!

Looks like its not just us - this is starting to get more attention https://www.seroundtable.com/google-cache-dated-19497.html

Im noticing a huge % of pages on my sites and those picked at random from Google searches, high traffic pages from big sites like booking.com have not been cached since 22nd Oct by Google. Anyone else noticed this or got insight on it?

Andy

It seems like Google confirmed this week in a G+ hangout that content in click to expand content e.g. 'read more' dropdown and tabbed content scenarios will be discounted.  The suggestion was if you have content it needs to be visible on page load. Here's more on it https://www.seroundtable.com/google-index-click-to-expand-19449.html and the actual hangout, circa 11 mins in https://plus.google.com/events/cjcubhctfdmckph433d00cro9as.

From a UX and usability point of view having a lot of content that was otherwise tabbed or in click to expand divs can be terrible, especially on mobile.

Does anyone have workable solutions or can think of examples of really great landing pages (i'm mostly thinking ecommerce) that also has a lot of visible content?

Thanks

Andy

I can see 90% of the pages I look at havent been cached since 22nd Oct, and I mean pages picked at random from hotels.com, booking.com and other big sites - is there a wider google cache issue going on?

I've gone through several large relaunches in recent years. I kept bullet point notes on what should be done to avoid ranking decline & wrote about it on my blog, hope it helps - http://www.andy-maclean.net/seo-website-relaunch-checklist/.

If results fell off a cliff, my 1st port of call would be checking the redirects were handled correctly, nothing untoward is going on with your robots.txt file and check GWT for any significant issues.

Andy

Seeing the same thing from July 8th onwards for both companies, and both use/have used Adroll.

Hi,

I've done a fair amount of research on international SEO including here on MOZ but was hoping some fellow Mozzers might have some practical examples of how they have got 1 domain to rank in multiple countries, ideally US & UK. Im possibly looking at getting a high authority domain which ranks great on US into the UK engines. I want to keep to the 1 domain to benefit from the high authority and for logistical reasons.

Thanks in advance,

Andy

Any update on this? This seems to have been in development for the best part of the year.

Hi,

I have rich snippet mark up showing in serps for some pages but not others. All pages test fine using Googles structured data testing tool. Whats really annoying is that they seem to appear for some pages but not others within the same directory / page format.

None of googles troubleshooting suggestions on the issue are a problem i.e.

• Does your markup follow our usage guidelines?
• Is your marked-up content hidden from users?
• Is your markup incorrect or misleading?
• Is your marked-up content representative of the main content of the page?
• Have you supplied enough information?
• Have you only recently updated your content?
• Does your markup include incorrect nesting?
• Reviews: Does your review use count instead of vote?

There are alot of instances when the same mark up is used twice e.g. on product x page in one directory and on product x page in a different directory (theres no dupe content). I wondered if that could be a reason but there are alot of instances when product x in directory a has the snippet when it doesnt in directory b. There doesnt seem to be an identifiable pattern as to why one page whould show the snippet and not another.

Any feedback appreciated. Happy to pm example pages.

Andy

Hi h2o2, maybe the phrasing wasnt clear but I think you misunderstood:

"For bigger sites especially I really wouldnt advise not using breadcrumbs."

301 redirect to new domain likely wouldnt help, you're bringing the tainted link juice with you, assuming links is the issue.

Most 'experts' think you wont recover from Penguin until the next Penguin update. Its unclear to me if this agency 'kept sending automated links' that you would have been cleared by any update if this process was still ongoing.

Bear in mind Penguin isnt just about links.

Changing domain might be a valid option but it does depend on the site itself, authority, number of spammy links you still have etc.

Not that its much comfort but id also say, remember Google isnt the only source of traffic.

Google themselves pointed me in the right direction and I figured out another solution with a tech friend.

Googles solution:

Surprisingly simple - I created advanced segments for visits to each product page e.g. for views of pages containing 'spanner in the url'. Then view traffic reports/ecommerce tab and bingo, conv rate for anyone who visited the site and looked at 1 or more spanner pages. Thankfully, and this is the key, analytics shows the visitor numbers for those that looked at 1 or more spanner pages so it avoids duplication.

Solution we came up with:

Create a session level custom variable which tracks when a visitor visits a product page i.e. 'spanner' page within a session and ends up purchasing. the custom variable is set so that is someone visits say a hammer and then a spanner page and buys a spanner, its the latter product i.e. spanner which is tracked for conv rate purposes. The exact setup of this is beyond me but it works in theory.

Hope that helps someone else with this vexing issue!

(On Mikes point about tracking page value instead, its a great one but conversion rate is an important metric for lots of reasons but not least as an indicator of sales performance. If AOV went up it would increase page value but conversion rate might be static or even falling).

Thanks mike. That makes sense but again there would be alot of opportunities for duplication e.g. the page value of spanner search results would be x, the page value of spanner product pages would be y and because of how people get around the site y would be included in x and vice versa. Ill check out that article.

David, thanks for replying again. That might work. Theres other issues here which might mean its just not possible e.g. to track conversion rate for a given product I think we can only do so if there is intent for that product e.g visit a spanner page, conduct a spanner search from home. All the visitors who e.g. type in the domain direct and bounce immediately (and the 1000s of other similar connotations) cant be included in conversion figures for a given product.

That doesnt work because a purchaser may have viewed 20 spanner pages. The conv figures in that scenario would vary wildly based on how many spanner page views there were and not give you a true conversion rate based on a unique visit.

Hi Ashley,

Thanks for the reply. That gives me the landing page that originated the purchase but not the conv rate?

Andy

Hi,

I need to identify the conversion rate for a product, lets call it a spanner.

I have 100s of spanner product urls and I ensured that the url protocols must include the product name e.g /red_spanner so its easy for me to work out the conversion rate in analytics for all my spanner pages as I just add 'spanner' to the landing page filter, hit the ecommerce tab and bingo.

What I cant figure out is how to work out the conversion rate for all spanner sales which includes alot of sales which didnt originate on spanner pages e.g. home page > search result > checkout. Theres 1000s of variations of this e.g. email > home > search > product page > checkout.

How can I work out the total conversion rate for spanners which needs to include: people landing on a spanner page and people who didnt arrive at a spanner page and did a search with eventually got them to the spanner pages.

Hopefully its not as complicated as I think!

Thanks in Advance!

Paul,

I would also add, as kevbudmonroe says, big site changes can create flux which can soon be recoverable, or better, on its own.

Im sure your scouring the analytics for clues but if you havent already look at your intelligence reports in Google Analytics - they might throw up something you didnt think of.

Good luck,

Andy

No problem Paul, hope it helps, let me know if you have anything to add the list, ill credit it back to you.

Heres a relaunching a site SEO checklist I did a while back which may help.

Google bolds 'with' when you search for w/ so it does recognise it but does w/ help you rank for a term with 'with' in is the bigger question.

Does anyone know of a list/resource of abbreviations recognised by Google?

For example does Google recognise w/ as with?

Andy

Hi,

The obvious thing to reply is 'different enough for what?' but id also ask whats the logic for doing that? Its the same words but in different order. What value does one bring to page a that the other doesnt to page b? Pages should have unique page titles, the fact the first 2 phrases are in a different order doesnt render them unique. Assuming they are for different but similar pages id use different phrases such as invoicing templates.

Andy

Like it, but id bear in mind though its widely believed that CTR is a ranking factor. Encouraging non click behaviour may negatively affect your positioning although I do widely see numbers in PPC ads actually increase CTR.

Using all applicable Adwords extensions. They can make such a difference to campaigns. Some of my client ads look like they're on steroids compared to the competition

Breadcrumbs are a great way for engines to understand how your site is structured and to pass link juice logically and evenly to the places you want it. For bigger sites especially I really wouldnt advise not using breadcrumbs.

Nofollow is no longer a way to block indexing of internal links. External yes, internal no.

Chad's answer about canonicals is a good one. Depending on how/why the duplicate pages are created/required example, sort orders or page 2, page 3, page 4 of the same page, you also have a couple of other options:

• Managing url paramaters in Google Webmaster Tools http://support.google.com/webmasters/bin/answer.py?hl=en&answer=1235687
• Googles rel next and rel prev tags http://googlewebmastercentral.blogspot.co.uk/2011/09/pagination-with-relnext-and-relprev.html

Matthew, I wrote a piece here a while back about what to make sure you do when relaunching/revamping a website to avoid ranking disaster. You may find it gives you some answers. I would also say, dont panic yet. A website redesign is a huge code change and flux can be expected initially.

No value other than a bit of competitive research.

I would say no, the H2 is purposefully for setting out something else on the page e.g. a H1 might be 'Dog Leads' and the H2 might be 'Leather Dog Leads'.  Also why waste the opportunity to help the page rank for something a bit different, in the above case 'leather dog leads'? I cant imagine why anyone would want to SEO a page for just 1 keyword.

To use your exact example above I would do something like:

How to install a car battery - Joe's Auto Store

Car battery installation

I dont think having the same h1 & h2 because it makes automation of your tags easier a great reason to do this. If that is your reason, then presumably it will happen right throughout the site and you will have the h1 & h2 replicating itself 100s of times which looks very spammy.

Hi,

Correlation doesnt always equal causation, what you reference is unlikely to be a cause of ranking decline inless something was done incorrectly e.g. have they rel prev'd your category page by mistake. (Even then im not sure that would cause an issue).

If you want to paste your site and an example url in here, ill gladly take a quick look when I have a minute.

Andy

I agree mostly with those above stating dont build a new web site.

Just to be different though - another option is a local sub domain. e.g. chicago.yourwebsite.com. This has advantages such as keyword in the domain name, it receives some link juice from the root domain and depending on your keywords you could get your root domain and sub domain in the same search results.

That said I have actually advised against this in the past as a sub folder on the root domain still has more chance of ranking well than the sub domain for lots of reasons I wont go into now.