Moving my business from SaveYourself.ca to PainScience.com in late 2014 was a cock-up of epic proportions — the worst setback I’ve suffered as an entrepreneur. I lost about half my search rank, traffic and revenue, less than two years after I’d finally reached goals I’d set a decade before.1 The site has still not recovered…not even partially. As of the one-year anniversary of the move, nothing has improved.
This was a problem with Google search results (as opposed reader reaction to the brand change2). I lost organic search traffic from Google, and a lot of it. Domain moves are not supposed to go this badly — not if you know a 301 from a 420. When I pulled the trigger, I believed it would be no big deal, and maybe even really good. (Ha! Such optimism!) But I was also paranoid, because I have so often been burned by things that aren’t “supposed” to go badly. In fact, I was so paranoid that I had avoided the project for years.
Even though the risk was supposedly low, the stakes were definitely high.
Paranoid preparation for the move
I prepared for weeks. I studied best practices for moving domains, and I heeded them. I am no SEO expert, but I’m technically savvy and I had full control over every key factor. My move was standard and straightforward.3 I checked many key ideas with more educated friends, and then — in a costly over-abundance of caution — I hired an SEO consultant I probably didn’t really need just to review everything again. He’d done many domain moves for big clients, sites much more complex than mine, and he checked my work and chided me for worrying.
“It’ll be a breeze,” he told me. “Domain moves are easy. I’ve never seen one go wrong.”
And then we watched mine go wrong.
It kinda felt like being killed
After all that preparation and procrastination, I finally pushed the button on November 27, 2014. A week later, over the space of a few terrifying days, about 80% of my organic traffic went buh-bye, and my income with it. My consultant agreed the dip was alarmingly steep, but it was temporary and the customers would be back soon. But the nightmarish days of the crisis ticked by and it only got worse: 85% loss of traffic, then 90% for Christmas. Merry ho-ho-holy shit!
The little publishing business I’d spent a decade on — my surprising success story, my idealistic legacy, my retirement plan — was floating face down in the water.
It felt like being killed myself.
I was preparing for a sudden career change when the rank-o-meter twitched. The traffic started to come back. Which felt like coming up for air after a close call with drowning. But it climbed back to only about half of what I’d had originally, and then no more. So by early January I had half my business back…
And that’s where it’s been ever since.
A freak SEO accident?
I have never confirmed any cause for this disaster. Was it a freak accident, one for the X-files of search engine optimization? Or was it was actually normal-ish for a domain move, and I was simply misinformed and blinded by wishful thinking, in spite of my paranoia?
Neither of those explanations is comfortable, and I still don’t know which is true. Maybe it’s a bit of both. But a Google glitch still seems plausible to me. Here’s why, in a nutshell:
Content mostly ranks independenly of where it lives. The Google PageRank of any given page on a website is the output of an infamously secret and complicated formula. But however it works, the name of the website is not critical: it doesn’t matter where the page is,4 it only matters what’s on it and how other websites “vote” for it by linking to it. When you move, you just need to advertise the change of address, so that Google knows that any links to OldWebsite.com should now count as “votes” for the same thing on NewWebsite.com.
Rank doesn’t “transfer,” per se, to a new domain. It doesn’t move, so it can’t get lost along the way. The content ranks. It gets re-ranked in the new location — and the output of the formula is basically the same, because it’s the same content with the same links to it. As long as Google knows that links to OldWebsites.com now belong to NewWebsite.com. And Google most certainly did know that.
So there’s no reasonable reason there should have been anything more than a temporary, minor problem. Or so I thought.
Or maybe it was normal after all
Before the move, it seemed to me like the Internet agreed that domain moves are fairly safe. Not perfectly safe — 10-20% losses for up to a few months are considered normal — but certainly not routinely disastrous. I had seen a few horror stories, but they were all rendered much less horrifying in predictable ways:
- they were inspired by the shock of temporary serious rank losses
- obvious, amateurish mistakes (yeah, it’s going to go badly if you redirect every page on the old site to the new home page)
- unusual variables or complexity beyond the control of the hapless webmaster
I never found a horror story that seemed like a solid case study of a move that went wrong for no apparent reason. And I still haven’t seen any such stories.
But I have since found some experts who do believe domain moves are tricksy and hazardous. No one has actually explained it, but clearly some SEO folks aren’t surprised by it.
I recently stumbled on this classic post on Moz.com about domain moves, with many ominous comments about how unpredictable and disastrous they can be, and even some particulars.5 If I’d seen this before the move, I’m not sure I would have done it! Even though I used Moz.com’s best practices guide for domain moves, I had never noticed that post, or those comments, or any others like them elsewhere online. But there it is in pixels.
Since publishing this post, other examples have been brought to my attention. Sistrix tweeted: “You’re in no way alone out there,” and linked to this chilling case study of a bad domain move.
Nevertheless, simple Google glitchiness remains plausible…
Google bugs! Bugs galore! So many bugs!
I do not know if Google is to blame. But I do know bugs when I see them. And I have seen Google glitching. During the crisis, I spent several weeks doing very little but studying Google and how it works, and something became absolutely clear: Google is definitely buggy. At one point I wrote to my consultant:
Is it just me, or are there a lot of Google bugs?
He wrote back immediately:
SO MANY BUGS.
There are many examples of Google bugs, but here’s one of my favourites, one of the biggest facepalms (technically not a “bug,” just a missing feature, but still a good example):
Google Webmaster Tools gives publishers a bit of a dashboard for their websites, and one of its tools is a move tool, so you can send an unambiguous message to Google that your content on Domain A has moved to Domain B, and they tell you to do that, and so I did. Straightforward, right?
I might as well have taken a damn sugar pill: Google later copped to ignoring notifications of moves to SSL (“https”) sites. A technology they had strongly endorsed the previous summer.6 But months after the announcement, there was still zero support for SSL sites in GWT, and my move notification went into the ether. Whatever benefit there should be from move notifications in GWT…I didn’t get it.
This was only one of many similar examples. By early 2015, I had an overwhelming impression that Google is a much sloppier technology than I ever imagined. Bing seemed much cleaner and more stable by comparison.
Bing had no problem with the move
Bing handled the transition exactly like Google should have. My rank on Bing never budged: it just went from indexing SaveYourself.ca URLs to PainScience.com URLs. It was agonizing to behold such good behaviour from a search engine with about 5% of the clout of Google.
Just because Bing does a thing doesn’t mean Google will too.7 But if I had done anything seriously wrong, it would likely have been a problem on Bing. Obviously the move was basically technically sound, and what happened to my Google ranks was exclusively an issue with Google.
No happy endings yet
This is the second major disaster I’ve had with Google. In early 2011, SaveYourself.ca suffered collateral damage from the first (infamous) Google “Panda” update. I lost about half my traffic and revenue then too, and it was traumatic then too, and nothing changed for a year then too. But the needle moved in early 2012, and soon the site was twice as busy as it had been before Panda, and then it doubled again in 2013.
Unfortunately, there was a lot more traffic and money at stake this time.
2018 update: good news and bad
Most pages on PainScience.com eventually recovered, and quite a few have even gone on to new highs. That’s the good news.
The bad news is that it took over a year for any page to recover, and about two years for most of the rest of the recovery, and some of my most profitable content never did recover. Even though my overall traffic is now actually much higher than it was before the move, my income remains significantly lower, because of the continuing failure of specific, important pages to recover.
The disastrous effects of the domain move remain an unsolved mystery in my mind. I don’t even have a best guess as to what went wrong. Most of the possible explanations have long since been discarded.
I always knew I’d eventually need to do about $300/day in sales, sustained for at least five years, to break even (a rough calculation). I hit $100/day in early 2010: enough to justify making PainScience.com my full-time job: adequate cash flow. I got to $200/day in late 2012, which was enough to call it a success and start to feel quite secure, but still not enough to pay for the years of investment. I reached $300/day in early 2013, and then $400/day by early 2014. I was on-schedule! ↩
Some readers really do prefer SaveYourself.ca, but PainScience.com was the clear winner after a lot of consideration, for many reasons: SaveYourself.ca had vaguely religious connotations, smacked of over-promising, and was doomed to always rank best only in Canada. ↩
There were two possible minor curve balls: moving from a .ca (for Canada) to .com, and moving from HTTP to HTTPS. And there’s no real reason to suspect that either of these things would cause much trouble. ↩
There is one important known exception: a domain can be penalized by Google. If it previously had gross spammy content on it, and was penalized by Google, it is not a good place to set up shop. But I verified very carefully in advance that PainScience.com was penalty-free. ↩
They did something rather extraordinary, and announced publicly that they would reward SSL sites (better security, good for everyone!) with a little bit of extra rank. So I went into the move assuming that going from the non-SSL SaveYourself.ca (“http”) to the SSL PainScience.com (“https”) would be a good thing! So silly! ↩
Basically because Google has problems deriving from popularity that Bing can only dream of having. Specifically, Google is forced to play very complicated games of cat-and-mouse with spammers and hackers and black-hat-SEOs. ↩