Google AI Mode & ChatGPT Both Testing Bottom Citation Cards

Both Google and OpenAI moved quietly this week to change where citation cards appear in their AI-generated answers. According to Google AI Mode & ChatGPT Tests Bottom Card Citations, reported by Search Engine Roundtable editor Barry Schwartz on July 22, 2026, Google AI Mode is testing an anchor that scrolls users directly to citation cards at the bottom of the page, while ChatGPT is experimenting with a citation box that loads as an overlay when a user clicks. Two different interfaces. Same general direction.

The Google test was spotted by Sachin Patel, who shared a video on X showing the behavior in action. According to Patel, "Google appears to be testing a change to citation cards in AI Mode. When opening a free listing product, the citation cards move to the bottom of the page." The ChatGPT version was flagged by Brodie Clark via his SERP Alert account, noting that ChatGPT "seems to be experimenting with how it's linking out to sources from its output" and that the organic shopping experience now includes a new product panel that appears after clicking.

Both tests are still experimental. No official announcements have come from Google or OpenAI. But when two of the biggest AI answer engines make the same structural shift in the same week, it's worth paying attention.

The Details: What's Actually Changing

Right now, in the standard AI Mode experience, citation cards tend to appear near or alongside the content they reference. The new Google test breaks from that. Instead of citations sitting in-line, the interface anchors you down to the bottom of the answer page where the citation cards live. You're not opening a new tab. You're not getting a traditional link click. You're being scrolled within the same AI-generated result.

The ChatGPT version works differently. Rather than anchoring you anywhere, it loads a citation overlay box when you click. Think of it like a modal window surfacing the source information without fully navigating away from the AI answer. Brodie Clark also noted this same experimental behavior is showing up in ChatGPT's organic shopping experience, where a new display and product panel appears on click.

These are two distinct UX approaches solving the same problem: how do you give users access to sources without pulling them away from the AI answer? Google pushes you down the page. ChatGPT keeps you in place with an overlay. Neither approach sends you directly to the source the way a traditional blue link would.

Why It Matters for SEO and Content Strategy

Here's the question Schwartz raised directly in his article: will these interfaces lead to more or fewer clicks? That's not rhetorical. It's the right question. And right now, nobody has a clean answer.

Think about what this shift signals. Both Google and ChatGPT are optimizing for user retention inside the AI experience. Keeping someone anchored to the bottom of an AI answer page or holding them in an overlay means they're less likely to leave. That's good for the platform. It's a more complicated story for the publisher or brand hoping to earn that referral visit.

I've watched citation placement evolve steadily since AI Overviews first rolled out, and the direction is consistent: citations are moving from prominent, above-the-fold positions toward supporting roles. Getting cited is still valuable. But getting cited doesn't automatically mean getting clicked anymore. The gap between "mentioned as a source" and "visited as a destination" is widening.

For brands that have built content strategies around earning AI citations, this is a wake-up call about metrics. Citation count is a leading indicator. Referral traffic from those citations is a lagging one. If the UI changes make the jump from citation to actual site visit harder, your traffic numbers can drop even while your citation rate holds steady or improves.

There's also a shopping-specific angle worth flagging. Brodie Clark's observation that ChatGPT's organic shopping experience is also getting a new product panel display after clicking suggests these UX experiments aren't limited to informational queries. E-commerce and product-focused content is in the mix too.

What to Do Now

You can't control where Google or ChatGPT place their citation cards. But you can control how you respond to the trend. Here's where I'd focus right now:

1. Start separating citation tracking from referral traffic tracking in your reporting. If you're measuring AI visibility by traffic alone, you're going to miss the picture. Track where you're being cited across AI Mode, ChatGPT, and Perplexity separately from the downstream click behavior.
2. Audit your brand strength in AI answers. When someone sees your site name in a citation card at the bottom of an AI result, is your brand name compelling enough to earn the click even when the UX makes it harder? Weak brand recognition is a conversion problem hiding as a traffic problem.
3. Double down on content that earns placement in shopping-adjacent AI results if e-commerce is relevant to your clients. The ChatGPT product panel behavior Clark flagged is early-stage, but it's worth getting structured product data and content in order now.
4. Watch your search console and analytics for referral source shifts. If you see AI-driven referral traffic dip while your content hasn't changed, UI tests like these may be the cause. Don't reflexively blame content quality.
5. Talk to your clients proactively. If an agency client sees a dip in AI referral traffic and you haven't already framed this context for them, that's a painful conversation. Get ahead of it.

Background & Context: The Bigger Trend

This didn't come from nowhere. Since Google launched AI Overviews broadly and ChatGPT started surfacing web citations in its responses, the industry has been circling one core tension: AI answer engines want to satisfy the user's query inside the interface, while publishers and brands need that traffic to actually land on their site.

Early citation designs were relatively visible. As these products have matured, citations have been deprioritized visually, collapsed into footnotes, or tucked behind interactions like the ones being tested now. This is not a random design choice. It reflects a product philosophy that the AI answer is the destination.

The conference circuit has been debating this hard. At events like BrightonSEO and SMX, the conversation has shifted from "how do I rank in AI Overviews" to "how do I actually convert from AI citation visibility." These tests make that question more urgent, not less.

And yes, both Google and ChatGPT experimenting with the same general approach in the same week is noteworthy. Whether that's coincidence or competitive mirroring, the result is the same: the industry is converging on a model where citations exist but clicks require more friction.

If you're not already tracking your AI citation footprint separately from traditional rank tracking, now is the right time to start. AI visibility tracking is something we built into Aergos specifically because these two signals have to be measured differently. Citation presence and click-through are not the same number, and treating them as the same will lead you to wrong conclusions about what's working.

The bottom line: citations moving to the bottom of the page is a small UI change with a potentially large traffic implication. Stay close to your data as these tests expand.