Sacha Lefebvre Founder, Paid Ai Search · LinkedIn

#1 on Google. Not in ChatGPT. Here is why.

Republic of Tea ranks first on Google for “green matcha tea” - a keyword with over 100,000 monthly searches (as Image 1 shows). First organic result. 4.8-star rating. Hundreds of reviews. A legitimate, well-established brand.

Republic of Tea ranking first on Google for green matcha tea

I opened ChatGPT and asked: “recommend me some green matcha tea.” I ran that query three times in fresh sessions. Republic of Tea did not appear once.

ChatGPT recommended Encha, Matchaful, Ippodo, Kettl, and Matcha Konomi instead. As Image 3 shows, ChatGPT explained its picks: “brands like Encha, Matchaful, Ippodo, Kettl, or Matcha Konomi are likely to show up because they usually have some combination of Japanese sourcing, small-format packaging, good color/freshness, strong reviews, and clear intended use.”

ChatGPT explaining why it recommends Encha, Matchaful, Ippodo, Kettl, and Matcha Konomi

So I asked ChatGPT directly: why not Republic of Tea? The answer: “Origin is broad. Japan, not specific enough. I favor brands that specify Uji, Nishio, Kagoshima, cultivar, harvest, or producer.”

Republic of Tea’s product page lists “Japanese matcha from premium tencha leaves.” Their own matcha guide acknowledges that Japanese quality is judged by leaf color, harvest time, cultivar, growing region, and processing method (republicoftea.com). That detail does not appear on their product pages (as Image 2 shows).

Republic of Tea product page missing region, cultivar, and harvest detail

ChatGPT could not verify the claim. So it did not recommend them. This is the signal.

ChatGPT is not a search engine. It is an intermediary (OpenAI, “Introducing shopping research in ChatGPT”). Before it recommends a product to a shopper, it needs proof - the full chain connecting a claim to the evidence behind it. When a brand says this tea is good for energy, ChatGPT wants to know why. It looks for independent scientific studies, real reviews confirming the effect, and specific verifiable detail. The intermediary cannot recommend what it cannot verify.

OpenAI says ChatGPT shopping weighs availability, price, quality, seller status, and checkout support when deciding which merchants to surface (OpenAI, “Buy it in ChatGPT”). Merchants can provide a direct product feed so ChatGPT reflects more accurate information (OpenAI Help Center, “Shopping with ChatGPT Search”). But none of that matters if the intermediary cannot connect your claims to your proof.

Republic of Tea is not a bad brand. They are an under-specified brand. Google rewarded their authority. ChatGPT penalized their vagueness.

The same gap exists in most Shopify catalogs right now. Claims without proof chains. Origin stated as a country, not a region. Benefits asserted without studies or reviews to back them. Google and ChatGPT are optimizing for different things. Most brands are only built for one of them.

That is this week’s signal.

Sacha June 6, 2026 · Issue #9

View ChatGPT Ads services · If you want this signal four times a week, get the Wire.

Correction policy. If anything in this post is wrong, I’ll fix it inline and add a Correction line above this notice with the date, what was wrong, what’s correct, and the source. No silent edits. Sources for this issue are listed here.