Everyone’s worst fear with site migration is a massive drop in traffic. And there are some examples of those worst fears being realised on prominent websites.
For example, in January 2012, Levi tried to internationalise their site by changing the taxonomy to include a “language” code.
The problem was, they didn’t follow up with an adequate redirection strategy and their visibility (measured by SearchMetrics, above) plunged from 6,000 to just a little more than 4,300 in the space of a week, and declined further for another month, dipping below the 4,000 mark, before starting to recover.
And then there’s Ebay.com, who in consolidated pages – and consequently “deleted” sections of their website without adequately redirecting URLs - revealed that their visibility almost halved in two weeks.
But moving a site to a new website can be accomplished without sacrificing traffic or revenue. And a well-managed move may even produce significant growth because of the bringing together of multiple assets.
That’s not to say that there will be no loss of visibility: there will almost certainly be some in the first few weeks where there is high volatility as the search engine spiders discover new URLs and update its index.
But successful site migrations show minimal visibility loss during the first few weeks followed by visibility growth, dependent on the type of migration. This explains the dilemma felt by webmasters contemplating a site move or merger.
Late last year, a large B2B company came to us to help them combine five websites into one as part of a plan to consolidate several constituent online businesses. For compliance reasons, a sixth commercial finance website within the group could not be moved.
For us at TMI, we felt that a small B2B operation is less able to manage a minor drop in traffic than say Levis or Ebay, so the procedure had to be watertight.
We began with a thorough analysis of what would be required from the agency and its client, and what each party could expect at every stage of the process..
As part of this, we created a client-friendly “Pre-migration Analysis” report, detailing the needs and procedures, written in beginner’s language – with developer-level appendices for the back-room boys and girls who have to put things into operation – to make the migration as painless as possible. All client-facing reports should be written in plain language to help them buy into the project. If they can buy in, they will spend what’s required.
While migrations should generally be planned months ahead with months to conduct the process, for company reasons, the client needed us to be ready in a little less than 15 working days. They also planned to move their websites to a new hosting company as part of the transfer.
The next step was to collect all the information we could about the sites from Google Analytics, Google Search Console, the GoogleIndex, third-partytracking tools such as AHRefs and SEMRush, backlink checkers including Link Research Tools and even direct crawls of the site using Screaming Frog and SearchMetrics. By the end, we had a complete understanding of what made the site tick, to a level of detail which made the client blush.
Deciding which pages should be migrated, amalgamated and updated is a game of two halves. There are obvious matches such as contact pages and T&Cs, and other uniques like product pages which directly transfer to the new site. But beyond these, experience comes to bear.
We prioritised updates by using external data such as backlinked pages and landing page traffic and used a content gap analysis of all the sites to find holes in coverage. The resulting spreadsheet of URLs to be created or be redirected, was then used to find patterns in taxonomy. Spotting patterns means you can apply logical redirects for some transfers, which reduces server load and improves page speeds.
Eventually, we had a complete inventory of pages before (and after) the migration, ready to be released into the wild. Every URL with some sort of prominence – either from appearing in the SERPs for a particular keyword phrase or which had backlinks pointing to it – was assigned a weighted, planned 301 redirect to a new page on the new site.
But the process also gave us a list of pages needed to fill gaps in the existing sites’ coverage, pages which could be merged, and pages that would be allowed to die (via 404 server instruction). These redirects were then applied on development sites, ready for the final switch over.
As a result of all of this, our expectation was that there would be little or no loss of visibility for any of our client’s websites.
On the appointed day, we began by physically testing all redirects and seeing how they resolved. Those which returned an error had their redirect checked and updated.
On launch day ...
- We applied the 301 redirects based on our mapping document.
- Updated all canonical tagson your sites so they pointed to the new site URLs.
- Updated all internal linkslikewise
- Reached out to important link donors, and ask them (nicely) to update their links.
- Updated the XML Sitemap, and submit to Google Search Console (and Bing’s Webmaster Tools).
In subsequent days, GSC revealed other more subtle defects, including “soft 404” errors where odd pages are redirected to the site homepage. In some instances, where content was no longer available, we applied new 404 instructions to encourage URLs to drop naturally out of the index, or 410 instructions to enforce deindexation.
Pages on the old sites were also coded up as NOINDEX which meant they cannot turn up as search results. Because of the varying gaps between search engine spiders returning to individual URLs – the “crawl interval”. In fact, some legacy pages were still appearing in searches weeks later, but they are continuing to dwindle in number, as the bots catch up with less important pages.
Ideally, the old sites should redirect to the new one on a page-by-page basis, indefinitely. If those redirects are lost, all of the inbound links earned by the old site will also be lost. While it’s possible that Google will attribute links pointed at the old sites to the new one, even without the redirect, it's not worth the risk.
Now the dust has settled, can the migration be considered a success, measured by the scale of lost traffic and SEO authority?
The bottom line is that two months on, Google Analytics and Search Console show traffic to the new website has not just maintained the levels going to the original sites, it has increased 55%, while the site now ranks well for relevant keywords never seen before.
We’ve also noticed an increase in indexation – and traffic – to the finance site that couldn’t be migrated, most likely the effect of the spring clean we gave it – new content, meta data, etc. – and the increased authority of the main site being passed on through backlinks.
Analytics data aside – as good as that is – the client has also reported a 16% increase in telephone enquiries in the weeks following the migration, including for markets they had not previously engaged with.
Work continues. We’re adding more useful interesting content to both sites and reaching out for new backlinks, and we’re targeting subjects that the client wants to highlight in the next few months.
And we’re now using this migration as a model for future work. We already have one underway at the moment and another is in the pipeline.
Max Brockbank is the head of SEO at The Media Image.