Google's goto change is a non-event for AI Search Visibility

On Wednesday 26th August Google confirmed it's changed how links work in its search results.

Click a result now and you don't go straight to the website. You go through a Google link first, google.com/goto, and Google passes you on from there. You'll probably never notice as a searcher. The only real difference is that you can't see where a link goes before you click it.

Google's explanation was short: "We have a long history of deploying technical measures against evolving forms of abuse, and we regularly take steps to protect our services and users."

For the tools that track Google rankings, this is a genuine headache.

Why rank tracking is having a bad week

Rank tracking tools work by loading a Google results page and reading the links on it. The links are the data. That's the whole job.

Now those links are disguised, and they can't be decoded without following them. So instead of loading one page and reading it, a tool has to chase every single link to find out where it actually goes. Derek Perkins at Nozzle worked out that a five-page ranking report now takes between 500 and 1,000 requests, where it used to take one.

That's on top of Google removing the num=100 setting last September, which had already made these tools fetch ten pages where one used to do.

So if your rankings look odd over the next few weeks, check with your tool provider before you assume anything's actually moved.

We've been asked a few times since Wednesday whether AI search tracking is about to have the same problem. It isn't, and the reason is quite a nice illustration of what these platforms actually are.

AI search tracking works in a completely different way

Tracking rankings means reading a Google page.

Tracking AI search means asking a question and reading the answer. You send a prompt to ChatGPT, Claude, Perplexity, Gemini or Copilot, and you record what comes back. Whether Google has disguised the links on its own results pages has nothing to do with any of that. There's no Google page in the middle to disguise.

There is one place it does show up. Google's own AI answers, AI Overviews and AI Mode, appear on a Google results page, so the sources listed underneath them are Google links and they get disguised like everything else.

But look at what that actually affects. Not the answer. Not whether you were mentioned. Not how you were described. Only the addresses of the sources sitting beneath the answer, which now need decoding before a platform can tell you which website got the credit. DataForSEO, one of the big data providers behind these tools, had it fixed two days after the change went live.

So it's a job for the data providers, and it's already done. Nothing for you to do.

The asking part isn't a dark art. Anyone can do it.

Here's something the industry doesn't say often enough.

You can do the data collection yourself, today, for free. Open a private browsing window, go to ChatGPT, and work through the questions your customers actually ask. Do the same in Claude, in Perplexity, in Gemini. Note down whether you show up.

We'd recommend every marketing team does this once, because it's genuinely eye-opening. You'll see your category described in words you didn't choose. You'll see two or three competitors you'd stopped worrying about. And you'll usually find some website you've never heard of doing more to shape your reputation than your own homepage does.

What you'll have at the end of it is a handful of screenshots from one morning, from one person, on one laptop. And AI answers change. Ask the same question twice and you can get two different answers with two different sources. One go isn't a reading, it's an anecdote.

What a platform is actually for

The platform isn't there to do the asking. It's there to make sense of what comes back.

So you're mentioned. Lovely. How often, out of how many attempts? On which engines, and which one's ignoring you? When you are described, is it accurate, and is it flattering? Who's winning the answer when you're not in it? Which competitors keep appearing, and are they there more than you? Is any of this moving in the right direction, or are you looking at a blip?

Then the layer underneath. How much traffic are you getting from AI tools, and how much of that is actual people rather than bots? Which pages should you work on first, once you line your Search Console data up against how you're performing in AI answers? Which websites are being cited instead of you, and which ones would be worth getting onto? That one matters more than most people expect. Across the AI answers we've looked at, 92% of the sources cited are pages the brand doesn't own, so most of your visibility problem is sitting on somebody else's website.

And then the bit that turns numbers into a plan. What do you actually do about the gaps? In what order, and by when? How do you show progress to a board that still thinks in visits? What does your customer journey look like across AI search, from someone's first vague question right through to comparing you against two rivals? Where do people drop away? And how are your existing customers talking about you in the places these engines are reading?

You don't get any of that from a browser tab. You get it from asking the same questions every day, to every engine, and having someone who knows what the pattern means. It's the job platforms like Answyn exist to do.

What to do about it

If you're already tracking your AI search visibility, this change asks nothing of you. No settings to check, no history to caveat, no phone call to make.

If you're checking your visibility by opening ChatGPT now and again, that's a perfectly reasonable place to start. You're just working from an anecdote, and you'll make better calls from a proper picture.

The bigger point is worth holding on to. Search measurement is splitting into two different jobs. One of them depends on Google letting you read its pages, and Google has made that much harder twice in the last year. The other depends on asking good questions and knowing what to do with the answers, and nobody can take that away from you.

We know which one we'd rather build on.

Frequently asked questions

Does Google's goto change affect AI search visibility tracking?

Almost not at all. AI visibility platforms work by asking questions to answer engines and recording the replies, not by reading links off a Google results page. The one exception is the sources listed under Google's own AI Overviews and AI Mode, which are Google links and so get disguised like the rest. The data providers behind these tools had that decoded within days.

Why does it hit rank tracking so much harder?

Because for rank tracking, the links are the data. A rank tracker's entire job is to load a Google results page and read the addresses on it. Now every one of those has to be chased individually, which turns a single request into hundreds.

Do I need to do anything?

No. If you're already tracking your AI search visibility, nothing about this change asks anything of you. If you also use a rank tracking tool, it's worth checking with your provider before you read anything into a sudden move in your rankings.

Can I track my AI search visibility myself?

Yes, and it's worth doing once. Open a private browsing window, ask ChatGPT, Claude, Perplexity and Gemini the questions your customers ask, and note whether you appear. What you can't do by hand is repeat it daily across every engine, which is what you need before the numbers mean anything.

What should I look for in an AI visibility platform?

The asking is the easy part, so judge platforms on what they do with the answers. You want to see how often you're mentioned and on which engines, how you're described, which competitors keep appearing instead of you, which websites are being cited, how much of your traffic is coming from AI tools, and what your customer journey looks like from first question to final comparison. Answyn is built around exactly those questions if you want a sense of what good looks like.