Botched SEO Migration Case Study
This migration was accidently messed up prior to this test/case study/experiment. It all started with a personal site project of Dave’s having an old WP theme that needed updating, it going sideways and there being more issues than a tired and sane person not making $$ directly off of it cares to spend at 11:30PM on a weeknight. So Dave reverted it back to an older but not mobile or remotely updated WP Theme.
That is when the real fun and issues started! Dave also failed to realize that in doing so he flipped URLs from post name to full date then post name. Months went by and in looking at GSC he realized his mistake. So then the plan came together for this whole experiment as the site was such a mess at that point Dave didn’t care about the site and just wanted to fix it.
In this episode we use “test” lightly as it was not very scientific and has more holes than swiss cheese. The goal however was simply to take a site that had been accidently messed up and make some type of learning out of it as best as we (Dave) could.
So what was “tested”?
- What gets pages out of the index faster? 404 or 410 statuses?
- Does it matter if you use 302 or 301 status/redirects?
- What redirect type (301 or 302) gets URLs updated faster in the SERPs?
- In general the goal was to poke around at watching URLs fall out of the index.
Anything else “tested”?
Yes. Much. The number of moving pieces was super high thus destroying any sort of real test. That said it was an opportunity to try and monitor and fix a pretty dated and messed up site.
- Moved from Non-WWW to WWW
- Moved to a new Theme/CMS
- Moved to a new URL Structure
- Moved from HTTP to HTTPS
GSC Coverage Data
Below you can see the total number of pages, errors and valid pages. The site had been on HTTP so what you are seeing here is all the HTTPS data that was or is available at the time of this podcast episode.
GSC Coverage Errors
Note the months of little to no change and then overnight the number of errors in GSC jumped from a handful to almost 30 and then started approaching 40 within a month. This is for a site that had maybe 120 real pages.
GSC Coverage Valid Pages
The migration’s impact shows up quicker for Valid Pages as by 4/4 almost 40 pages (30%) were suddenly gone. Then a month later at the time when the Errors spiked finally the valid pages just started to drop and drop and drop. In reality the site went from 120 or so valid pages to 10 or so now.
Page Level Test Data
Below is the click and impression chart (via GSC) for a few keywords and some pages. Three of the pages are different URLs that over the past 6 months have been live (again, part of the problem for this site). The thing to note is that how long it took for the switch over but once it did it is clear that Google favored the page/site much more than it had in the past.
Months Later Old Pages Still Ranking
This screenshot was taken the last week of June 2019. It is Page 2 of the search engine results but it is clear that Google still thinks there are 45 pages and that despite most of these in the image 301 redirecting, 404/410ing some 3+ months later they are still in the index.
Botched SEO Migration Case Study Learnings
So the results and things that happened to this site are really all over the place. So many things were broken, fixed, and done weirdly to try and test things.
- Do QA. Much of the issues this site had were due to Dave not doing QA last year.
- Test SEO and crawl your site BEFORE you launch a site/theme/migration.
- Download and use Xenu Link Sleuth and/or Screaming Frog as they are both free for small sites.
- Set a reminder for 2-3 days out to look at Google Analytics & Google Search Console to check on your changes.
- Look at log files 2-5 days after to see if Google/Bing is crawling.
- If you use a freelancer, design firm, developer or anyone – always QA any changes they make.