Type into a search box and you will probably write something like "decaf green tea". Three words. Years of search engines taught us to cut our questions down to the fewest words that might work.
Now listen to how people talk to an AI assistant. Something like this, which is an example and not a real customer:
"I'm looking for a decaf green tea for my mum. She likes lighter flavours, caffeine keeps her awake, she has two cups a day, she's on a tight budget, and organic would be nice if it doesn't cost much more."
That is the same person wanting the same thing. The difference is that the second version tells you who it is for, why caffeine matters, how much she drinks, what she can spend and what would make it just right. Every one of those details is something a good answer can use. In the short version, all of it has been thrown away before anyone could help.
I call the second kind of request Premium Intent.
Why it needed a name
I started using the term in June 2026, while I was rebuilding teas.co.uk. My working notes from 11 June put it simply: "Premium intent is not a keyword."
The marketing trade already has words that sound close. "High intent" and "buyer intent" describe how ready someone is to buy. "Long tail" describes searches that are rare. Neither describes what I was looking at. A person can be ready to buy and still tell you almost nothing. Premium Intent is about how much the person has told you: their situation, the feeling behind it, the timing, the limits, the budget, the things they would rather avoid, and what a good result would look like to them.
"Premium" describes the detail in the request. It is not a fee, and it is not a posh customer. A pensioner asking for a cheap, gentle evening tea is making a Premium Intent request. So is a dad trying to get his son a birthday present he will actually use.
I am not claiming nobody has ever put those two words side by side. They have: Ian Jindal wrote about "premium intent" in April 2026, in a piece on airport shopping, in a different sense from mine. I am saying that Premium Intent as a named idea, defined like this, and built into a shop as something an AI can call, started here.
What it changes for a small shop
A customer who arrives with Premium Intent has already done most of the work. They have explained what they want. My job is no longer to persuade them about tea in general. It is to find the one or two products that genuinely fit and say why.
When that match is good, I would expect a few things to follow: the sale is more likely to happen, it is less likely to come back, and the customer is more likely to ask the same way next time. The assistant that sent them is more likely to come back to a source that served them well. I want to be honest that this is a belief I am testing, not a figure I can show you yet.
It also changes who can win. A request like the one for my imaginary mum is full of conditions, and every condition is a fact that has to be true. Decaf. Light. Affordable. Ideally organic. An advert can push a product to the top of a list. It cannot make a caffeinated tea decaffeinated. A specialist who knows their products properly, and writes those facts down clearly, can be the right answer even next to the biggest shop in the world. The more detailed the request, the more depth matters.
How it shaped teas.co.uk
Premium Intent is the reason the shop is built the way it is.
Every product sits on one checked set of facts that people, search engines and AI assistants all read the same way, including price per cup, pack size and caffeine status. We record "decaffeinated" and "caffeine free" as separate things, because they are, and a Premium Intent request often turns on exactly that difference.
The shop also has a service built to receive the whole request. Our Ask API tells any AI that calls it that it supports Premium Intent. Our public instructions for AI systems say it "performs best given the full spoken request": context, emotional stakes, constraints, budget and what success looks like, "not a keyword summary", in about 40 to 60 words. In return, we promise to acknowledge the emotional weight of a request, ask a clarifying question if we need one, and give honest advice on where to spend and where to stop. A test of the live Ask API first recorded that support on 19 June 2026, and it is still published today.
"Where to stop" matters to me. If someone asks for something we do not stock, or a pack size we do not sell, the honest answer is to say so. A shop that drags every request towards whatever it happens to have in stock is wasting the detail the customer gave it.
Why it is the reason we opened the door to AI
Most of the web is busy blocking AI crawlers. I went the other way, and Premium Intent is why. These detailed requests are being asked of assistants, every day, in full sentences. If an assistant cannot read my shop, it cannot answer them from my shop, and the person goes to whoever it could read. So teas.co.uk is open to every AI that behaves well, and protected against the ones that do not. There is more on that in Most of the web is closing the door to AI. We opened ours.
Where the name is on record
- June 2026: I start using the term while rebuilding teas.co.uk. My notes of 11 June say "Premium intent is not a keyword."
- 19 June 2026: a test of the live teas.co.uk Ask API shows it declaring Premium Intent support, with the guidance "Send the full human intent, not a keyword summary."
- 14 November 2026: the shop's agents.json, its public notes for AI systems, still asks callers for the full spoken request, "not a keyword summary".
The full record, with sources, is kept on my Tea Wiki entry, under Premium Intent: the dated record.
What I hope happens next
I would like every specialist with deep knowledge, whether that is tea, tyres or hedges, to start building for Premium Intent rather than keywords. Write down what you know properly, say what each thing is not good for, and make it easy for an assistant to read. The people asking are already telling us exactly what they want. The least we can do is be ready to listen.
What others have found
Outside research and reporting that bears on this, with links to the original sources.
- Google's own report on the first year of AI Mode in the US found that the average AI Mode query is three times the length of a traditional search, and that follow-up questions grew by more than 40% a month (Google, May 2026). Search Engine Journal covered the findings in Google data shows AI search users moved past keywords and AI Mode queries are 3X longer.
- Among the most common first words in AI Mode searches is "I": people are telling the search engine about their own situation, not typing a product category (Search Engine Journal, from Google's data).
- In the same Google data, planning queries grew 80% faster than AI Mode queries overall, and comparison questions beginning with "which" grew 40% faster (Search Engine Journal).
- Google says AI Mode has passed a billion monthly users, with queries more than doubling every quarter since launch (Google; see also Search Engine Journal on the first year of AI Mode usage figures).
- Adobe reported that traffic to US retail websites from generative AI tools rose 693.4% in the 2025 holiday season compared with the year before, and that those visitors converted 31% more often than visitors from other sources (Adobe, January 2026).
Read more
- Premium Intent: the Tea Wiki entry.
- Three Premium Intent days: three imagined requests, followed all the way through.
- The decaf fruit tea problem: why every condition in a request is a fact that has to be true.
- Truth entity: the owned, checkable record that makes a Premium Intent answer possible.
- Under the Hood: what the shop measures, and where the evidence stops.
- Meet the founder: why I built teas.co.uk
More from the news
- Most of the web is closing the door to AI. We opened ours.
- Three days that have not happened yet, and the kind of request we built for
- The decaf fruit tea problem
Lee Samuel Tucker
Founder, teas.co.uk
Tunbridge Wells, Kent
