Restructuring Your Website and How to Minimize Traffic Loss

In February 2011, a client I was working with ran global usability test in cooperation with Spotless Interactive in London, Hamburg and Oslo. One year later and the feedback from the usability tests are being implemented, which includes improvements to the booking engine, navigation menu, a url rewrite and a new home page.

With a total of 21 websites in 15 languages, launching a new global web structure without losing a significant portion of traffic was always going to be a difficult task. Especially as more than 50% of total visits are from search engines. This blog post will show you two examples of how to launch a new website:

• Website 1 launches and loses 35% of organic traffic
• Website 2 launches and loses 4% organic traffic

By following the process described below, you can feel confident in knowing your new website launch will be smooth.

Minimizing risk and measures we took

With the support of Norwegian inbound marketer Sverre Bech-Sjøthun, we were able to create a step-by-step plan to implement before, during and after the launch.

Before the launch

We started by setting up a project in Basecamp with the web developers and to ensure all stakeholders were aligned.

Having just recently upgraded to the newest version available of the CMS, we found that the steps taken to minimize the traffic loss during the website launch would also be a good time to address to increased crawl errors reported since the upgrade.

By upgrading the CMS for more than twenty websites, the number of crawl increased tenfold to more than 4,500 errors globally.

Example of errors reported for one of the sites

404 page visits increased during the same period

The first step was to approach each site separately and by using Xenu Link Sleuth we are able to reduce the number of broken links. We repeated this process for each of the sites. Across the entire web structure, fixing broken links was a lot of work and requires a dedicated person. Understanding the importance of addressing these issues is the only way it can be prioritized above everyday tasks.

We then started working on the URL mapping:

• Using Open Site Explorer we ran a report for the top 500 linked to pages
• Using Google Webmaster Tools we ran a report for the top 10 Links to Your Site
• Using Microsoft Excel, we mapped out the site and new URL structure

Managing the top 500 linked to pages in a time consuming process but highly important when launching a new website. The process took half a day per website and was the most demanding of all steps taken.

Once the URLs were mapped, we then created an XML sitemap based on the live version of the website.

The step-by-step process for launching the new website included:

• Map URLs and redirects
• Submit XML sitemap
• Fix crawl errors in Webmaster Tools
• Monitor web traffic in Google Analytics

During the Launch

Once the new site went live, the 301 redirects were implemented and the XML sitemap submitted to Google Webmaster Tools. Traffic was monitored in Google Analytics and errors monitored in SEOmoz.

How to launch a new website redesign

The following illustrates how one team gave SEO a high priority and how another team didn’t. Here are the results.

Website 1

You will always run into problems that you did not see when launching a new website. The process plan was created and all stakeholders were aligned. However, as a team we did not execute the plan when launching this website and implementing correct redirects and on-site SEO were not prioritized.

The day we went live with website 1 (14th June, 2012):

• No XML sitemap was added to webmaster tools
• Not all 301 redirects were implemented (more than 50% missing)
• 302s pages were sending traffic to a soft 404 page (not a 404 HTTP status)
• Missing meta tags including page titles and meta descriptions causing duplicate content

One issue was that 50% of the redirects were not implemented. The issue being we did not know which 50%. Using SEO automatic bulk URL checker we manually checked each URL and HTTP status code. A second issue was that Google was indexing the test server resulting in duplicate content - more than 276 pages were indexed.

By not implementing the process plan, we lost a lot of organic traffic. In fact, organic brand traffic decreased by 45% compared with the previous week and year on year organic traffic was down by 49%.

Organic traffic is now down -34% comparing the previous month (easy to identify the launch date)

Impressions down -18% comparing the previous month

Website 1 is the client’s most visited website within the global web structure. The above charts from Google Analytics have been seen by senior management and addressing these issues has now been prioritized. It’s not too late, but there is no doubt that by losing 34% of organic traffic a considerable amount of sales have also been lost.

We are now in the process of updating XML sitemaps, implementing any outstanding redirects and fixing crawl errors on site.

Website 2

For website 2, we had a lot more control and I was allowed to be hands on with the process.

The day we went live with website 2 (26th June, 2012):

• Uploaded the XML sitemap immediately after launch
• Mapped out all URLs of the site, which included a URL rewrite with user-friendly URLs
• 301 redirects were implemented and tested
• Monitored the web traffic for both referrals and organic traffic

Once live, I blogged about the new launch, tweeted the launch to more than 2,000 followers, informed all stakeholders internally and had the news published on the company intranet. The day following the launch, we sent out a newsletter to 1,600 subscribers that included a tip to "check out the newly launched website!". The tip can also be found in my email signature.

Organic traffic is now down -4% comparing the previous month (barely visible)

Impression share trend continues as before

Here is a list of actions taken for the website 2 launch.

Actions summary:

• Create report of top 500 linked to pages from Open Site Explorer
• Map URLs from old site to new site with redirects
• When launching new site, implement redirects
• Submit XML sitemap to webmaster tools
• Test new top 20 linked to pages for correct 301 implementation
• Attract new site links through blogging and social media shares
• Send out newsletter and inform customer base
• Promote launch in company email signature
• Monitor traffic in google analytics
• Monitor and fix crawl errors in webmaster tools
• Submit new XML sitemap (two weeks post launch)

Concluding summary

With the website 1, we only followed the 50% of the plan and we lost 34% of organic traffic. With website 2, we followed the plan exactly as it should have been, we constantly monitored traffic in web analytics and tested both referral links and 301 redirects – a valuable lesson in having a plan and sticking to it.

If you are about to launch a new website, have your SEO consultant on-site or hire an expert during this process and involve your web developers throughout the launch. Make sure this is prioritized within the organization and not left to those who do not understand the importance of SEO. No one can afford to lose 34% organic traffic.

Steven Macdonald 

Source: http://www.seomoz.org

Read more ►

How to Free Your E-Commerce Site from Google's Panda

On Feb. 25, 2011, Google released Panda to wreak havoc on the web. While it may have been designed to take out content farms, it also took out scores of quality e-commerce sites. What do content farms and e-commerce sites have in common? Lots of pages. Many with zero or very few links. And on e-commerce sites with hundreds or thousands of products, the product pages may have a low quantity of content, making them appear as duplicate, low quality, or shallow to the Panda, thus a target for massive devaluation.

My e-commerce site was hit by Panda, causing a 60% drop in traffic overnight. But I was able to escape after many months of testing content and design changes. In this post, I'll explain how we beat the Panda, and what you can do to get your site out if you've been hit.

The key to freeing your e-commerce site from Panda lies at the bottom of a post Google provided as guidance to Pandalized sites:

One other specific piece of guidance we've offered is that low-quality content on some parts of a website can impact the whole site’s rankings, and thus removing low quality pages, merging or improving the content of individual shallow pages into more useful pages, or moving low quality pages to a different domain could eventually help the rankings of your higher-quality content.

Panda doesn't like what it thinks are "low quality" pages, and that includes "shallow pages". Many larger e-commerce sites, and likely all of those that were hit by Panda, have a high number of product pages with either duplicate bits of descriptions or short descriptions, leading to the shallow pages label. In order to escape from the Panda devaluation, you'll need to do something about that. Here are a few possible solutions:

Adding Content To Product Pages

If your site has a relatively small number of products, or if each product is unique enough to support entirely different descriptions and information, you may be able to thicken up the pages with unique, useful information. Product reviews can also serve the same purpose, but if your site is already hit by Panda you may not have the customers to leave enough reviews to make a difference. Additionally, some product types are such that customers are unlikely to leave reviews.

If you can add unique and useful information to each of your product pages, you should do so both to satisfy the Panda and your customers. It's a win-win.

Using Variations To Decrease Product Pages

Some e-commerce sites have large numbers of products with slight variations. For example, if you're selling t-shirts you may have one design in 5 different sizes and 10 different colors. If you've got 20 designs, you've got 1,000 unique products. However, it would be impossible to write 1,000 unique descriptions. At best, you'll be able to write one for each design, or a total of 20. If your e-commerce site is set up so that each of the product variations has a single page, Panda isn't going to like that. You've either got near 1,000 pages that look like duplicates, or you've got near 1,000 pages that look VERY shallow.

Many shopping carts allow for products to have variations, such that in the above situation you can have 20 product pages where a user can select size and color variations for each design. Switching to such a structure will probably cause the Panda to leave you alone and make shopping easier for your customers.

Removing Poor Performing Products

If your products aren't sufficiently unique to add substantial content to each one, and they also don't lend themselves to consolidation through selectable variations, you might consider deleting any that haven't sold well historically. Panda doesn't like too many pages. So if you've got pages that have never produced income, it's time to remove them from your site.

Getting Rid of All Product Pages

This is a bold step, but the one we were forced to take in order to recover. A great many of our products are very similar. They're variations of each other. But due to the limitations of our shopping cart combined with shipping issues, where each variation had different shipping costs that couldn't be programed into the variations, it was the only viable choice we were left with.

In this option, you redesign your site so that products displayed on category pages are no longer clickable, removing links to all product pages. The information that was displayed on product pages gets moved to your category pages. Not only does this eliminate your product pages, which make up the vast majority of your site, but it also adds content to your category pages. Rather than having an "add to cart" or "buy now" button on the product page, it's integrated into the category page right next to the product.

Making this move reduced our page count by nearly 90%. Our category pages became thicker, and we no longer had any shallow pages. A side benefit of this method is that customers have to make fewer clicks to purchase a product. And if your customers tend to purchase multiple products with each order, they avoid having to go from category page to product page, back to the category page, and into another product page. They can simply purchase a number of products with single clicks.

Noindexing Product Pages

If you do get rid of all links to your product pages but your cart is still generating them, you'll want to add a "noindex, follow" tag to each of them. This can also be a solution for e-commerce sites where all traffic enters on category level pages rather than product pages. If you know your customers are searching for phrases that you target on your category pages, and not specifically searching for the products you sell, you can simply noindex all of your product pages with no loss in traffic.

If all of your products are in a specific folder, I'd recommend also disallowing that folder from Googlebot in your robots.txt file, and filing a removal request in Google Webmaster Tools, in order to make sure the pages are taken out of the index.

Other Considerations: Pagination & Search Results Pages

In addition to issues with singular product pages, your e-commerce site may have duplicate content issues or a very large number of similar pages in the index due to your on-site search and sorting features. Googlebot will fill in your search form and index your search results pages, potentially leading to thousands of similar pages in the index. Make sure your search results pages have a rel="noindex, follow" tag or a rel="canonical" tag to take care of this. Similarly, if your product pages have a variety of sorting options (price, best selling, etc.), you should make sure the rel="canonical" tag points to the default page as the canonical version. Otherwise, each product page may exist in Google's index in each variation.

MAXMORITZ

Source: http://www.seobook.com

Read more ►