HTTP HTTPS Migration Gone Wrong - Please Help!
-
We have a large (25,000 Products) ecommerce website, and we did an HTTP=>HTTPS migration on 3/14/17, and our rankings went in the tank, but they are slowly coming back. We initially lost 80% of our organic traffic. We are currently down about 50%.
Here are some of the issues. In retrospect, we may have been too aggressive in the move.
- We didn't post our old sitemaps on the new site until about 5 days into the move.
- We created a new HTTPS property in search console.
- Our redirects were 302, not 301
- We also had some other redirect issues
- We changed our URL taxonomy from http://www.oursite.com/category-name.html to https://www.oursite.com/category-name (removed the .html)
- We changed our filters plugin. Proper canonicals were used, but the filters can generate N! canonical pages. I added some parameters (and posted to Search Console) and noindex for pages with multiple filter choices to cut down on our crawl budget yesterday.
Here are some observations:
- Google is crawling like crazy. Since the move, 120,000+ pages per day. These are clearly the filtered pages, but they do have canonicals.
- Our old sitemaps got error messages "Roboted Out". When we test URLs in Google's robots.txt tester, they test fine. Very Odd.
- At this point, in search console
a. HTTPS Property has 23,000 pages indexed
b. HTTP Property has 7800 pages indexed
c. The crawl of our old category sitemap (852 categories) is still pending, and it was posted and submitted on Friday 3/17 - Our average daily organic traffic in search console before the move was +/-5,800 clicks. The most recent Search Console had HTTP: 645 Clicks HTTPS: 2000 clicks.
- Our rank tracker shows a massive drop over 2 days, bottoming out, and then some recovery over the next 3 days.
- HTTP site is showing 500,000 backlinks. HTTPS is showing 23,000 backilinks.
I am planning on resubmitting the old sitemaps today in an attempt to remap our redirects to 301s.
Is this typical? Any ideas?
-
our dev team has requested to 302 initially and then 301 as the final solution. Could this cause a problem for the temporary 302?
-
Looks like you've done a sterling job GWMSEO. Well done!
I deal with a lot of small sites where the decision to move from http to https was made for them... by Shopify (many without them even realising). Some of these sites take months to recover. So, to see recovery within a few weeks is a good result.
Once question I have for you: how have you resolved the backlinks issue (your observation #6)?
What I have observed is that ahrefs doesn't associate the old (http) backlinks with the new version (https). I believe this then impacts backlink profile and subsequently MOZ Domain Authority, which then impacts rankings and traffic.
301 redirects ensure you don't lost the traffic, but it doesn't move the backlinks from http to https.
Nobody has been able to confirm or deny my suspicions yet. So:
- Do you (or anyone in the MOZ community) know if these lost backlinks DOES impact Domain Authority?
- Is there any way to fix it, other than go through all the old backlinks and manually change to https?
Murray
-
Thanks to all for the help. Our rankings are climbing back. We should be back to status quo in +/-2-3 weeks after the move.
-
Nice, good job! I would double check with your CDN provider to make sure implementation was done correctly according to their process. Unless you're saying you discontinued use of the CDN when you switched?
I wouldn't panic, just make sure your team knows that you can't control the rate at which Google re-indexes the website and that it's still early in the process to tell if there is an issue somewhere. Let Google do their thing and then once your traffic and rankings seem more regular, reevaluate. At that point I would add HTTP/2 support if possible and measure the impact from that because that provides some additional benefits such as a boost to site speed.
-
Thanks for the punch list. Our rankings turned around some when we submitted the old http sitemap.
- Images only on HTTP in CDN. HTTPS, No CDN.
- Yes
- Yes. (I was a stickler on this one)
- Yes
- Yes, Yes, Yes.
- Yes
- Yes, we fixed that.
- Yes. Initially our https redirects were 302 (Not on me, LOL.). We resubmitted our old (http) sitemap file today to crawl so that Google can pick up the change.
- Yes
- Yes
- No disavow file. Our backlink profile is remarkably squeaky clean. TF-57, CF-47
Bonus: Yes.
-
1. Are you using a CDN?
2. Did you update all your internal links to https?
3. Did you update all of your canonical tags?
4. Did you update all of your hreflang tags?
5. Are you using plugins/modules from a third party? Are they secure? Do they have documentation or a rep you can contact about migrating to https?
6. Some CMS's have specific settings that need to be altered when migrating - make sure those were done correctly.
7. Use screaming frog to check for any external scripts, and ensure they're calling https.
8. Did you update your old redirects?
9. Did you update your robots.txt file to include the new https sitemap?
10. Did you enable HSTS?
11. Do you have a disavow file? Did you update it for HTTPS?Bonus:
Did you update all of your other paid campaigns, analytics, etc. to reflect the migration? -
OK, that makes a big difference. Before you resubmit any sitemaps, make sure your technical implementation is perfect. Fix all sitemaps, make sure all URLs are properly 301'd, and make sure you are not generating any type of mixed-content security errors. Make sure all sitemaps are going to the new URLs. Most important, when a user clicks on an old URL (e.g., from search results or an old backlink), they need to be 301 redirected to the same page on HTTPS (not the home page). Then resubmit your sitemaps.
Your transition is still very 'young,' and it will likely take 6-8 weeks before Google completely replaces the old site with the new site in it's algorithms and search results. You should expect a temporary drop in traffic during this transition, but assuming you have done everything correctly, things should normalize after that.
-
Tustind:
Sorry - we did it a week ago - 3/14/17.
We just changed the 302s to 301s yesterday. I believe we need to resubmit our old sitemaps for google to index the 301s.
-
So you made this migration over a year ago now? From my knowledge and experience doing a domain change, which has a lot in common with a https change, we experienced about 2 months of depressed traffic, after which point our traffic normalized and then continued on a growth trajectory. So I suspect that your biggest problem was not properly implementing the 301 redirects. You basically told Google that you were only making a temporary change, which could explain why after an entire year your site has still not been fully picked up. When did you change the 302s to 301s?
Got a burning SEO question?
Subscribe to Moz Pro to gain full access to Q&A, answer questions, and ask your own.
Browse Questions
Explore more categories
-
Moz Tools
Chat with the community about the Moz tools.
-
SEO Tactics
Discuss the SEO process with fellow marketers
-
Community
Discuss industry events, jobs, and news!
-
Digital Marketing
Chat about tactics outside of SEO
-
Research & Trends
Dive into research and trends in the search industry.
-
Support
Connect on product support and feature requests.
Related Questions
-
Preserving rankings after migration
I have acquired new a client who wants to migrate to a new platform. He also wants a 100% guarantee we maintain he Google search rankings. How realistic is this? Anyone got a checklist they can recommend?
Intermediate & Advanced SEO | | muzzmoz0 -
Http to https what should one look out for
Hi, we are moving our site from http to https and would like to know what should one look out for. What are the things that can go wrong, and what is the best way to go about it. Regards Tai
Intermediate & Advanced SEO | | Taiger0 -
Video SERP Help
Hello Friends,
Intermediate & Advanced SEO | | KINQDOM
I try to appear on search results of property related search terms with my property videos. Here is a sample property video
http://www.antalyahomes.com/videositemap.asp May you please check it and tell me what I do wrong? Thanks in advance for your time.0 -
Dealing with non-canonical http vs https?
We're working on a complete rebuild of a client's site. The existing version of the site is in WordPress and I've noticed that the site is accessible via http and https. The new version of the site will have mostly or entirely different URLs. It seems that both http and https versions of a page will resolve, but all of the rel-canonical tags I've seen point to the https version. Sometimes image tags and stylesheets are https, sometimes they aren't. There are both http and https pages in Google's index. Having looked at other community posts about http/https, I've gathered the following: http/https is like two different domains. http and https versions need to be verified in Google Webmaster Tools separately. Set up the preferred domain properly. Rel-canonicals and internal links should have matching protocols. My thought is that we will do a .htaccess that redirects old URLs regardless of the protocol to new pages at one protocol. I would probably let the .css and image files from the current site 404. When we develop and launch the new site, does it make sense for everything to be forced to https? Are there any particular SEO issues that I should be aware of for a scenario like this? Thanks!
Intermediate & Advanced SEO | | GOODSIR0 -
Complex URL Migration
Hi There, I have three separate questions which are all related. Some brief back ground. My client has an adventure tourism company that takes predominantly North American customers on adventure tours to three separate destinations: New Zealand, South America and the Himalayas. They previously had these sites on their own URL's. These URL's had the destination in the URL (eg: sitenewzealand.com). 2 of the three URL's had good age and lots of incoming links. This time last year a new web company was bought in and convinced them to pull all three sites onto a single domain and to put the sites under sub folders (eg: site.com/new-zealand). The built a brand new site for them on a Joomla platform. Unfortunately the new sites have not performed and halved the previous call to action rates. Organic traffic was not adversely affected with this change, however it hasn't grown either. I have been overhauling these new sites with a project team and we have managed to keep the new design but make usability/marketing changes that have the conversion rate nearly back to where it originally was and we have managed to keep the new design (and the CMS) in place. We have recently made programmatic changes to the joomla system to push the separate destination sites back onto their original URL's. My first question is around whether technically this was a good idea. Question 1 Does our logic below add up or is it flawed logic? The reasons we decided to migrate the sites back onto their old URL's were: We have assumed that with the majority of searches containing the actual destination (eg: "New Zealand") that all other things being equal it is likely to attract a higher click through rate on the domain www.sitenewzealand.com than for www.site.com/new-zealand. Having the "newzealand" in the actual URL would provide a rankings boost for target keyword phrases containing "new zealand" in them. We also wanted to create the consumer perception that we are specialists in each of the destinations which we service rather than having a single site which positions us as a "multi-destination" global travel company. Two of the old sites had solid incoming links and there has been very little new links acquired for the domain used for the past 12 months. It was also assumed that with the sites on their own domains that the theme for each site would be completely destination specific rather than having the single site with multiple destinations on it diluting this destination theme relevance. It is assumed that this would also help us to rank better for the destination specific search phrases (which account for 95% of all target keyword phrases). The downsides of this approach were that we were splitting out content onto three sites instead of one with a presumed associated drop in authority overall. The other major one was the actual disruption that a relatively complex domain migration could cause. Opinions on the logic we adopted for deciding to split these domains out would be highly appreciated. Question 2 We migrated the folder based destination specific sites back onto their old domains at the start of March. We were careful to thoroughly prepare the htaccess file to ensure we covered off all the new redirects needed and to directly redirect the old redirects to the new pages. The structure of each site and the content remained the same across the destination specific folders (eg: site.com/new-zealand/hiking became sitenewzealand.com/hiking). To achieve this splitting out of sites and the ability to keep the single instance of Joomla we wrote custom code to dynamically rewrite the URL's. This worked as designed. Unfortunately however, Joomla had a component which was dynamically creating the google site maps and as this had not had any code changes it got all confused and started feeding up a heap of URL's which never previously existed. This resulted in each site having 1000 - 2000 404's. It took us three weeks to work this out and to put a fix into place. This has now been done and we are down to zero 404's for each site in GWT and we have proper google site maps submitted (all done 3 days ago). In the meantime our organic rankings and traffic began to decline after around 5 days (after the migration) and after 10 days had dropped down to around 300 daily visitors from around 700 daily visitors. It has remained at that level for the past 2 weeks with no sign of any recovery. Now that we have fixed the 404's and have accurate site maps into google, how long do you think it will take to start to see an upwards trend again and how long it is likely to take to get to similar levels of organic traffic compared to pre-migration levels? (if at all). Question 3 The owner of the company is understandably nervous about the overall situation. He is wishing right now that we had never made the migration. If we decided to roll back to what we previously had are we likely to cause further recovery delays and would it come back to what we previously had in a reasonably quick time frame? A huge thanks to everyone for reading what is quite a technical and lengthy post and a big thank you in advance for any answers. Kind Regards
Intermediate & Advanced SEO | | activenz
Conrad0 -
Site migration from non canonicalized site
Hi Mozzers - I'm working on a site migration from a non-canonicalized site - I am wondering about the best way to deal with that - should I ask them to canonicalize prior to migration? Many thanks.
Intermediate & Advanced SEO | | McTaggart0 -
Help needed for a domain
I have a small translation agency in Brazil (this website), totally dependent on SEM. We are in business since 2007, and we were on top position for many relevant keywords until the middle of 2011, when the ranking for the most important keywords started dropping. In that time, we believed that we needed to redesign the old static website and replace it by a new modern one, with fresh content and with weekly updates, which we did, and it's now hosted on Squarespace. I took care to keep the old links working with 301 redirections. When we made the transfer from the static site to Squarespace (Mar/2012, see the attachment), the ranking dropping became even more serious. Today, we have less than 50 unique visitors per day, in a total desperate situation! To make things worse, we received an alert from Google on 23/September/2012 talking about unnatural inbound links, but Google said that "As a result, for this specific incident we are taking very targeted action on the unnatural links instead of your site as a whole", so we thought we didn't need to worry about. Google was correct, I worked many hours to register our website in web directories, I thought there would be no problem since I was doing this manually. My conclusions are: Something happened prior to Mar/2012 that was making us losing territory. I just don't know what! The migration to Squarespace was a huge mistake. I lost control over the html, and squarespace doesn't do a good job optimizing the pages for SEO. We also were also blasted by Penguin on September, but I believe this is not the main cause of the drop. We were already running very badly at this time. My actions are: a) I generated a DTOX report and I'm trying to clean up the links marked as toxic. That's a hard work! After that I will submit a reconsideration request. b) I'm working on the site: Improving internal link building for relevant keywords Recently I removed a "tag cloud" which I believe was hurting my SEO. Also, I did some redirections that were missing. c) I trying to generate new content to improve link building to my site. d) I'm also considering to stop putting all my coins on this domain, and maybe start a fresh new one. Yes, I'm desperate! 🙂 I would appreciate a lot to hear from you guys, expert people! Thanks a lot, MWcEdPa.png?1
Intermediate & Advanced SEO | | rodrigofreitas0 -
Domain migration strategy
Imagine you have a large site on an aged and authoritative domain. For commercial reasons the site has to be moved to a new domain, and in the process is going to be revamped significantly. Not an ideal starting scenario obviously to be biting off so much all at once, but unavoidable. The plan is to run the new site in beta for about 4 weeks, giving users the opportunity to play with it and provide feedback. After that there will be a hard cut over with all URLs permanently redirected to the new domain. The hard cut over is necessary due to business continuity reasons, and real complexity in trying to maintain complex UI and client reporting over multiple domains. Of course we'll endeavour to mitigate the impact of the change by telling G about the change in WMC and ensuring we monitor crawl errors etc etc. My question is whether we should allow the new site to be indexed during the beta period? My gut feeling is yes for the following reasons: It's only 4 weeks and until such time as we start redirecting the old site the new domain won't have much whuffie so there's next to no chance the site will ranking for anything much. Give Googlebot a headstart on indexing a lot of URLs so they won't all be new when we cut over the redirects Is that sound reasoning? Is the duplication during that 4 week beta period likely to have some negative impact that I am underestimating?
Intermediate & Advanced SEO | | Charlie_Coxhead0