Disney+ shows up on Google with a URL that ends in utm_source=chatgpt.com. The first reflex says: botched canonical. I think something else happened.
I google “Disney+ Login”, hover over the first result, and down in the status bar it says: disneyplus.com/identity/login?utm_source=chatgpt.com. On Google, right at the top, with ChatGPT’s tracking parameter attached. That’s a nice little accident, and you can learn quite a lot from it about how Google really works. But before anyone goes looking for culprits at Disney, it’s worth looking at the mechanics behind it.
The first reflex
And now to you, dear SEO colleagues. “Somebody at Disney botched the canonical, rookie mistake, blah blah blah.” Yeah, yeah. Could be. I don’t know the site from the inside and I don’t know what its canonical said that day. But: nobody at Disney made up that parameter. ChatGPT adds utm_source=chatgpt.com itself to links it hands out, for example in citations and source references. So the linked website can see in its analytics where the visit came from. Well meant, really.
What probably happened
My guess, and from the outside it can’t be more than that: ChatGPT cites Disney+, dutifully attaches its parameter, and that exact URL is now out in the world. People click it, copy it from the address bar, share it, link to it. Google finds it, sees that it keeps showing up out there, and picks this variant. Even if the canonical on the page “says” something else. A login page is the perfect candidate for this: nobody reads its URL, people just use it. Google takes note of the canonical and does its own thing anyway.
The canonical is a suggestion
That’s the point that gets lost in all the Disney gloating: the canonical tag isn’t an order. Google’s own documentation calls it “a strong signal”, and that wording is deliberate. There are other signals next to it, and Google weighs them:
- redirects, according to Google also a strong signal
- the canonical in the page’s head
- which URL is linked internally
- which URL is in the sitemap, according to Google a weak signal
- which variant is linked and visited out there
If enough of these point to the UTM variant, it wins. I’ve seen plenty of cleanly set canonicals that Google simply ignored. Wild, actually, if you think it through: everyone who links to your site with parameters gets a say in your indexing. You build the page, and other people vote on the address it shows up under on Google.
And this is where it gets more interesting to me than the question of who was asleep at Disney. ChatGPT has become a link source in its own right. Every answer with a source reference throws URLs into the world, in the shape ChatGPT gives them. If you only think of AI visibility as being cited, you miss the side effect: the citation comes back with its own label attached and may end up in Google’s index.
How to check which URL Google picked
For your own site this takes two minutes. Google Search Console has the URL Inspection tool: paste the address at the top, and under page indexing it shows you the canonical URL you declared and, next to it, the one Google selected. If the two don’t match, you have your own little Disney moment. You’ll find these cases collected in the page indexing report, among the duplicates where Google chose a different canonical than you did.
Stack your signals
Ok, the good news: signals can be stacked. A single one isn’t always enough; together they pull in the same direction.
- Never link internally with UTM parameters, not from menus, teasers or the footer either.
- Only the clean URL goes into the sitemap.
- Set the canonical on every page anyway, pointing to itself.
- For campaign links you hand out yourself, think briefly about how often they’ll get passed on.
Sure: you could also strip the parameters with a server-side redirect. I usually advise against it, because then your analytics no longer sees where the visit came from, and that’s exactly why the things are there in the first place. In Drupal, by the way, the Metatag module sets the canonical; how it holds up against the others is in my comparison of Drupal SEO modules.
A side note about myself, because otherwise this would be dishonest: every link from this blog to Traffalyzer, my SEO tool, carries UTM parameters itself. On purpose, otherwise I can’t see which article sends people over. The link you just passed included. So I get a say in the indexing of a site too, just my own. With a handful of links I’m not worried; the mechanics are the same as with Disney, though.
Disney will survive, the login page works no matter what’s attached at the end. Meanwhile, go open the URL Inspection for your homepage ;)
Comments