There’s a new traffic source growing in your analytics right now, and until recently GA4’s default reports hid it inside the referral bucket. Visitors arriving from ChatGPT, Perplexity, Gemini and Copilot click through when an assistant cites you. The numbers are small on most sites. The trend line and the visitor quality still deserve their own measurement.

Why it matters: a shopper who asked an assistant “which sunscreen won’t leave a white cast” and got your product in the answer has already done most of their research. For a D2C brand, that’s the visitor you want more of, so you need to know how many arrive, where they land and whether they buy.

Below: the GA4 setup first, then what actually helps a D2C store get cited, including the parts nobody can prove yet.

Does GA4 Track ChatGPT and Perplexity Traffic on Its Own Now?

Partly. Google has added an AI Assistant channel to GA4’s Default Channel Group. When GA4 recognises the referrer as an AI assistant, it sets the medium to ai-assistant and the session lands in that channel, which sits above Referral in Google’s rule order. Google’s default channel group documentation names ChatGPT, Gemini, DeepSeek, Copilot and Grok as examples. It doesn’t name Perplexity.

So why build anything yourself?

• Perplexity isn’t on Google’s list. For a D2C brand it’s often the second-biggest AI source, and Google doesn’t say whether it’s covered. A custom regex that includes perplexity.ai is how you make sure it’s counted.
• You control the definition. Google hasn’t published its full list. Your own regex means you decide what counts, can check it, and can add new assistants as they appear. Custom channel groups can also be applied to your reports retroactively, so check your older months in both views.
• AI Overviews and AI Mode are excluded. Those clicks still arrive as google / organic. Nothing in GA4 separates them, and your custom group won’t either.

Run both, and compare them. Your custom group is the one you control.

How to Track Perplexity and ChatGPT Traffic in GA4, Step by Step

You need Editor access on the GA4 property. No code, no Tag Manager. A standard property allows two custom channel groups, so if you’ve already used both, edit one of them instead of creating a third.

The regex for step 3. It covers the assistants that send measurable referral traffic today; add anything new you spot in step 1.

(.*\.)?(chatgpt\.com|chat\.openai\.com|perplexity\.ai|gemini\.google\.com|copilot\.microsoft\.com|claude\.ai|chat\.deepseek\.com|grok\.com|meta\.ai|you\.com)

The (.*\.)? at the start catches subdomains such as www.perplexity.ai. Don’t shorten it to plain words like gemini|copilot: you’ll pull in unrelated sources that happen to contain them.

Why Some AI Clicks Still Show as Direct or Unassigned

Whatever your AI Assistants channel shows is a floor. A chunk of real AI-driven visits will never carry a clean signal, for a few reasons:

• Copy and paste. Someone copies your URL or brand name out of an answer and types it into a new tab. No referrer, so GA4 calls it Direct.
• Apps and stripped referrers. Links opened from mobile apps, or links marked noreferrer, often arrive with no referrer at all.
• ChatGPT’s UTM tag without a medium. Since mid-2025 ChatGPT adds utm_source=chatgpt.com to many of the links it cites, but no utm_medium. In the old default grouping that lands in Unassigned. Your source regex catches it either way.
• Google’s own AI answers. Clicks from AI Overviews and AI Mode come through as Google organic. Search Console reports them inside normal Search data, mixed with everything else.

The cheapest patch is a “How did you hear about us?” question after checkout, with ChatGPT and Perplexity as options. Imperfect, but it tells you something GA4 can’t.

Reading the Numbers Honestly

How Do You Get Traffic From Perplexity to a D2C Website?

You get it by being one of the sources Perplexity reads and trusts when a shopper asks a product question. Perplexity runs a live web search for most answers and cites the pages it used, and ChatGPT search works much the same way. Here’s what I’d check first on a D2C store.

1. Let the AI crawlers in

Perplexity indexes with PerplexityBot, which follows robots.txt. ChatGPT search uses OAI-SearchBot. If either is blocked, you can’t be cited from your own site. Check robots.txt, then check your CDN or firewall: some bot-protection settings block AI crawlers by default, and store owners often don’t know it’s switched on. You can block training crawlers like GPTBot separately without touching the search ones.

2. Put the product facts in plain HTML

Price, availability, size, ingredients or materials, who it’s for, shipping and returns. If those only appear after JavaScript runs, or live inside an image, assume an assistant can’t use them. Add Product schema with offers, GTIN where you have one, and aggregateRating built from real reviews. It won’t guarantee a citation. It does remove one reason to skip you.

3. Publish the comparison content shoppers ask about

People ask Perplexity “X vs Y”, “best X for oily skin”, “is X worth it”. Pages that answer those questions directly, with a clear verdict near the top and honest trade-offs (including where a competitor fits better), are the ones that get pulled into answers. A collection page with forty product tiles and two lines of copy usually isn’t.

4. Get talked about where Perplexity looks

In my research on which brands ChatGPT, Perplexity and Gemini recommend, the pattern was clear: assistants repeat consensus. They lean on review roundups, editorial “best of” lists, Reddit threads, marketplace listings and YouTube reviews. If your brand isn’t on those, your own site has to work much harder. Check which sources Perplexity cites for your category’s top questions, then earn a place on them.

5. Know where Perplexity Shopping stands

Perplexity runs a free merchant program for its shopping results, but it’s aimed at businesses that sell and ship to the US. If you’re an Indian D2C brand selling only in India, that route isn’t open yet, and your citations will come from ordinary answers instead.

Clicks and mentions are different things. To see whether ChatGPT names your brand at all, even when nobody clicks, use the free AI visibility tracker I’ve published. For the wider discipline, see generative engine optimisation.

What to Do With the Numbers Once You Have Them

Once the measurement exists, it changes decisions. The landing page report feeds your content plan. The conversion data justifies AI visibility work to whoever owns the budget. And the baseline lets you judge whether citation work moves anything at all. Measurement first, then optimisation has something to answer to.

For most D2C brands, AI referrals are a small add-on to an organic channel that still has to be built properly. If you’re paying for every visitor today, this D2C case study shows how a paid-only brand grew organic traffic 218% and revenue 5x in 18 months, which is the foundation AI citations sit on.

Frequently Asked Questions