About nine months ago I was facing a sizeable site migration project at my employer at the time, and I was tasked with mapping out the best way of handling this migration to minimise the impact on our search engine rankings.
I wrote a post summarising my research into the SEO aspects of a site migration, and I feel the time has come to look back at the migration and the lessons we learned from it.
The site migration was a two-step process – we updated the design and we added new sections with fresh content. We decided to follow the recommendations outlined in my site migration blog post pretty much to the letter:
Content: We phased in the new content one batch of pages at a time. We put a couple of new pages live, linked to them from the homepage, and waited for them to be indexed & cached. Then we put the next batch of new pages online.
Design: The design changes were implemented gradually as well. The old and new designs weren’t radically different, it was more a tweaked & modernised version of the old design, so we felt it would be fine to have the old and new designs co-exist on the site for a while.
We first did a Google Website Optimiser A/B test to make sure the new design yielded at the very least a similar conversion rate. When this was confirmed, we migrated pages to the new design one at a time. The URLs all remained the same so we didn’t have to do any 301-redirects.
We used the Duplicate Content tool to ensure the HTML code and content of our key pages with high SERP rankings matched at least 90% in the old and new designs, so we wouldn’t get hit with a ranking penalty when we put the new version up.
When a page was updated with the new design, we waited for it to be indexed & cached in Google and checked how its SERP rankings were affected.
The end result was a site with a fresh design and new sections added, with minimal impact on SERP rankings. We did see some fluctuations in rankings but these fell well within the normal daily and weekly ranking variations.
We also noted that the new content started ranking fairly soon for relevant keywords, despite no direct links coming in to those pages. This is most likely due to the incoming link value generated across the rest of the site, spilling over to the new content.
It was a long and labour-intensive process, and in hindsight I’m not sure it would have impacted the rankings massively if we just switched the site over in one go. But as organic search generates a significant portion of the sites traffic and revenue, it was definitely better to be safe.