Are Local Florists Finally Winning, Thanks to AI?

I asked ChatGPT, Gemini and Claude where to buy flowers in four US cities. In 36 of 36 answers, a local florist came first.

By Nick, co-owner of Kvetka Flower, a floral studio in Portland, Oregon

Key findings

• 36 of 36: in every answer to a search with a city name, ChatGPT, Gemini and Claude ranked a local florist first.
• 72% of recommendations for city searches went to local florists. Order gatherers like FTD, Teleflora and 1-800-Flowers got 14%.
• The wording matters: "florist [city]" brought 95% local florists, while "same day flower delivery [city]" brought 58%, with order gatherers taking 31%.
• 0% local florists when the search had no city. Without a location, AI falls back on national brands.
• AI shoppers buy more often: at our shop, visitors from ChatGPT converted at 4.8%, compared with 3.2% from Google search.

Based on 208 recommendations for 14 real searches in Portland, Austin, Boise and Nashville, collected September 25, 2026. Full data (CSV).

Why I looked into this

This summer, a new source of orders showed up in our store analytics: ChatGPT. In July we got one order from it, in August four, in September six. That's roughly one in every nine orders we took that month. According to Shopify, shoppers who came to us from AI assistants spent about $1,250 in the last 30 days. $1,090 of that came from ChatGPT alone, up 77% from the 30 days before. Separate lines for Google AI Mode, Gemini and Meta AI only showed up in the report a few days ago. And clicks from regular AI Overviews in Google are most likely still counted as plain Google traffic, so the real number is higher.

For a small studio, that's a real channel, and it's growing faster than anything else we have. So I wanted to understand how AI assistants decide who to recommend when someone is looking for flowers. Especially compared with big delivery services like 1-800-Flowers and Teleflora, which have spent years pulling customers away from local florists through ads and search results.

So I went where people argue about this all the time. On Reddit, in SEO and small business subreddits, threads asking "who does AI actually recommend?" come up constantly. After reading through them, I ran a small experiment of my own: I gave three AI assistants real search queries about flowers in four cities and went through all 208 recommendations they came back with.

Spoiler: local florists win clearly when someone is looking for flowers in their own city. The order gatherers are left with same-day queries and searches with no city at all, and even there it's complicated. I don't think they're getting out of that corner.

Who order gatherers are, and why I don't like them

If you've ever ordered flowers from a site like 1-800-Flowers, FTD or Teleflora, chances are that site didn't make your bouquet. It took the order, kept a cut and passed it to a local florist near the recipient. Under the classic split, out of every $100 the florist who actually arranges and delivers the flowers gets about $73, and then still pays the network fees on top of that. The customer pays full price, and noticeably less of it ends up in the flowers.

What bothers me isn't the idea of a middleman. It's that a few big players with huge ad budgets ran this market for years. But even for them, the model has stopped working. In May 2026, the CEO of 1-800-Flowers told investors on an earnings call: "we were buying transactions for $40 and making $20 margin on each transaction." If the customer doesn't come back, "you are just wasting dollars." Since 2022 the company's revenue has fallen by a third, and its stock is down 93% from the peak. FTD went through bankruptcy back in 2019. My opinion: it's only going to get harder for them from here.

If the biggest player in the market admits a $40 click doesn't pay off, a small studio has no business in that auction. The winner is whoever customers find on their own. That's why I wanted to know where AI is sending them now.

AI search went mainstream in two years

My six orders a month are a drop in the bucket compared with what's happening overall. In November 2023, ChatGPT had 100 million weekly users. By February 2026, it had 900 million. The Gemini app has more than a billion monthly users. Google AI Mode grew from 100 million to more than a billion in a year, and 2.5 billion people a month see the AI answers in regular Google results (AI Overviews).

For stores, something else matters more: people aren't just searching through AI, they're buying. According to Adobe, AI traffic to US retail sites has grown 14x since fall 2024. In March 2025 those visitors were 38% less likely to buy than everyone else. By May 2026 they were 54% more likely. Someone who arrives from an AI assistant usually already knows what they want: they asked a question, got an answer and came for something specific.

We see the same thing at our small scale. Over the last 90 days, visitors from ChatGPT bought 4.8% of the time, visitors from regular Google search 3.2% of the time, and the site average was about 2.5%. It's a small sample, but it points the same way as Adobe's data. More than half of those visitors land straight on our homepage: AI recommends us as a shop, not as a specific bouquet.

So the question of who AI recommends isn't theoretical anymore. For the flower business, it's a question of where customers will go over the next few years.

How I tested it

I wanted the test to look as much as possible like the way a regular person searches. So I didn't make up the queries. I took them from Google Keyword Planner. These are the phrases Americans actually type when they want to order flowers.

That gave me 14 queries. Three for each of four cities: "flower delivery," "florist" and "same day flower delivery" plus the city name. And two of the most popular queries with no city at all: "flower delivery" and "same day flower delivery." Each of those is searched somewhere between 100,000 and a million times a month, so it wouldn't have been fair to leave them out.

Here's how I picked the cities:

• Portland: our home, a market I know from the inside.
• Austin: we're moving there soon, and I was curious how AI sees that market.
• Boise: a smaller city.
• Nashville: a big, fast-growing city in the South.

If the result holds in four very different places, it's a pattern, not a fluke.

I asked each query to three assistants: ChatGPT, Gemini and Claude. Every chat was temporary, with no history and no personalization, and web search turned on. The prompt was an ordinary human request: here's my search, recommend five places to order flowers and tell me why. It said nothing about local florists or order gatherers. All answers were collected on September 25, 2026, in the evening Pacific time.

That came to 208 recommendations. I checked every one by hand: opened the website, looked at the address, figured out who was actually behind the name. Then I sorted them into four groups:

1. Local florists: a real shop or studio in that city that makes and delivers the bouquets itself.
2. Order gatherers and wire services: FTD, Teleflora, 1-800-Flowers, From You Flowers and the like. They take the order and pass it to a local florist for a cut.
3. Nationwide online delivery: Bouqs, UrbanStems and similar services that ship flowers in a box.
4. Marketplaces and retail: BloomNation, DoorDash and so on.

A word about Portland. It's my city, so Kvetka could show up in the answers. I didn't tweak anything, didn't ask the AI to mention us and didn't throw out any inconvenient answers. And yes, we're there. Kvetka appeared in 7 of the 9 Portland answers, more than any other florist in the city. For "same day flower delivery portland oregon," ChatGPT and Claude both put us in first place. I'm really proud of that. We don't have the ad budget of a national chain. We just spent a few years building a decent website, collecting reviews and doing honest work. Apparently that's exactly what AI knows how to notice.

Before this, I did a trial run: 12 long, conversational questions in Gemini, AI Mode, Perplexity and Claude. Local florists got 89% of the recommendations there. But the questions were too much "mine," not the way real people usually ask. So I treat the second run, built on Keyword Planner queries, as the main one, and every number from here on comes from it.

It's important to understand that this is a snapshot of one evening, not months of observation. AI answers change from one try to the next and depend on things like the time of day. In the evening, for example, shops that are still taking orders for today have an edge. I'll say more about the limitations at the end of the article. But as you'll see, the overall picture came out very clear.

What the assistants said

First, the main result. For queries with a city name, the assistants gave me 179 recommendations, and 72% of them were local florists. The rest:

• 14% order gatherers;
• 10% nationwide online delivery;
• 4% marketplaces and retail.

What struck me even more was who came first. There were 36 such answers, and in all 36 the first recommendation was a local florist. Not once did an order gatherer, a national service or a marketplace take the top spot. Local florists sat at position 2.5 on average, everyone else at 4.3. So the gatherers are in the answers, but somewhere near the bottom of the list, as a backup option.

The result depended most of all on how the query was worded:

• "florist [city]": 95% local. You're looking for a florist, and AI takes you literally: here are the florists in your city.
• "flower delivery [city]": 63% local. This is where services that ship flowers in a box nationwide, like Bouqs and UrbanStems, show up. They got 22%.
• "same day flower delivery [city]": 58% local and 31% order gatherers. This is the only place where gatherers still hold noticeable ground. Same-day delivery has always been their main promise: "order from us and we'll find a florist who can deliver today." For now, AI repeats that logic. But even here, more than half of the recommendations went to local shops.

Among the non-local names in city answers, Teleflora and Bouqs came up most often, 11 times each.

City by city, the picture was almost identical. Portland, Austin and Nashville were 73% local, Boise 68%. So this isn't about one special city. It's a general pattern. Boise is telling for another reason: in regular Google results for one of these queries, 9 of the 10 links went to order gatherers. The assistants still put local shops first. AI isn't just retelling the first page of search results. It figures out who actually makes bouquets in that city.

There are differences between the assistants, but they all lean the same way. ChatGPT recommended local florists 83% of the time, Gemini 67% and Claude 66%.

Now the other side. For the two queries with no city ("flower delivery" and "same day flower delivery"), not a single local florist made the list. Zero. Only FTD, 1-800-Flowers, Teleflora, Bouqs, UrbanStems and the like. I don't see that as an AI mistake. It doesn't know where you are, so it gives a default answer: the big national brands. But it's important to understand. These are exactly the two queries people search hundreds of thousands of times a month, and here the gatherers are still fully in the game.

So the takeaway is simple: as soon as a city appears in the query, local florists win. Without a city, whoever has the loudest brand wins.

But I'm fairly sure this result is temporary. We tested in temporary chats where the assistant knew nothing about the person. I was sitting in Portland, and even so, not one assistant filled in the city on its own. Real life is different. People use AI through their own accounts, and the assistant has memory and chat history. It knows what city you live in, and often where your mom lives when you want to send her flowers. If the conversation includes the recipient's address, or even just the city, then "flower delivery" effectively turns into "flower delivery portland oregon." And there, as you can see above, the picture is completely different.

I couldn't test this honestly. That would take several real accounts with different histories in different cities, which is a whole different scale of research. So for now it's my hypothesis, not a finding. But if it's right, even the biggest no-city queries will gradually shift to local florists, simply because AI will know better and better who you're sending flowers to and where.

What I'd tell other florists

I'm not a consultant or a marketer. I'm a florist who has spent the last two years building our website, writing our copy and figuring out SEO in the evenings after deliveries. So this isn't a playbook. It's what I see in my own shop and in this study.

First: you don't need to do anything special "for AI." When I read Reddit, people are always hunting for some secret trick: special files, magic prompts, services that promise to "get you into ChatGPT" for a monthly fee. I didn't do any of that. I think AI works on the same logic Google once built its search on: the more honest signals that you're real and good, the higher you rank. The difference is that AI seems better than Google at telling a real shop from a middleman site. You could see that very clearly in Boise.

Second: say on your website where you are and who you serve. It sounds obvious, but I see plenty of florist websites that mention their city once, in the footer. The results show that the city in the query decides everything. So it should be everywhere on your site where it's natural: in headings, in bouquet descriptions, on your delivery page, in the neighborhoods you deliver to. We have a separate page for same-day flower delivery in Portland. That's the page ChatGPT linked to in its answer, not our homepage.

Third: same-day delivery. This is the only place where order gatherers are still strong in AI answers, at 31%. Because it's their main promise. If you deliver same day, say it loudly and specifically: how late you take orders and how late you deliver. We take same-day orders until 6 PM and deliver until 10 PM. I think that's one reason we showed up so often that evening. AI is looking for who can still help someone right now.

Fourth: profiles and reviews, and not just on Google. I used to think Google Business Profile was enough for local search. That's no longer true. Gemini and AI Mode, both from Google, rely on Google Maps and Google reviews. ChatGPT, on the other hand, signed a deal with Yelp in summer 2026 and uses Yelp's reviews, ratings and photos in its local answers. I know a lot of florists who gave up on Yelp long ago. I'd go back: fill out your profile, add photos and ask happy customers to leave a review there too. A real address, fresh reviews and correct hours on every platform. All of it is boring, and all of it works.

Fifth: look at where your visitors are already coming from. Most florists I talk to don't know whether they get any traffic from AI. They most likely already do. In Shopify you can see it by referral source (chatgpt.com, gemini and so on) and in the channel reports. When I first looked, it was a revelation. That's also where I saw that ChatGPT visitors often land on our wedding pages, but nobody books from there yet. That's next on my list.

And finally, about the order gatherers. I don't think they'll disappear tomorrow. But for the first time in a long while, a small shop has a channel where the winner isn't whoever spent the most on ads. For now. There are almost no ads in AI answers yet, and if they show up, the gatherers with their budgets will be first in line. That's why I think now is the best time to get your website and profiles in order. Not for some trick, but so that when AI looks for a florist in your city, it finds you easily.

If you just want to order flowers

My advice to shoppers is short: don't overthink it.

For years, ordering flowers online looked the same everywhere. You go to a website, click "Birthday" or "Anniversary," scroll through identical bouquets and pick the one that bothers you least. The site knows nothing about you or the person you're sending flowers to.

With AI, you don't have to do that anymore. You can just say what you want. That your mom loves peonies and can't stand lilies. That this is an apology, not a celebration. That your friend just moved to Portland and feels lonely. That your budget is $100 and it needs to arrive tonight. Ask about something new, something there was never anywhere to ask before. The more of real life you put into your request, the better the answer.

What this study doesn't prove

I want to be honest with anyone who cites these numbers, so here are the study's limitations.

It's a snapshot of one evening. All answers were collected on September 25, 2026. Ask the same questions tomorrow morning and the answers will be a little different. AI answers fresh every time and takes into account things like who is still taking orders today.

The sample is small. 14 queries, 4 cities, 3 assistants and 208 recommendations. That's enough to see the overall picture, but it's not an academic study of thousands of queries.

The assistants knew nothing about me. I deliberately used temporary chats with no history or memory. That makes the answers cleaner, but real users are in a different situation. I described above how that might change the picture. I can't test it yet.

I did the classification myself. I checked every business by hand, and some cases are borderline. Kremp, for example, is technically a real flower shop in Pennsylvania, but for other cities it works as an order gatherer. BloomNation passes orders to local florists but is still a marketplace that takes a commission. In borderline cases, I looked at who actually makes the bouquet in the recipient's city.

I'm not a neutral party. I'm a florist, I have opinions about order gatherers, and I don't hide them. That's exactly why I'm publishing all the data: the queries, the answers and the classification. Anyone can check it and repeat the study.

The sales data is from one shop. The numbers on AI orders and conversion are Kvetka's, 167 visits from ChatGPT over 90 days. It's real data, but it's one shop, not the whole market.

Full data: download all 208 recommendations (CSV), with the query, assistant, rank, business, website and my classification for each one.

In closing

Over the last 30 years, the number of flower shops in the US has fallen by more than half. There are many reasons, and order gatherers are only one of them. But for a long time the rules were simple: whoever pays the most for ads gets found.

For the first time, I'm seeing those rules change. When someone asks AI about a florist in their city, the assistant almost always names a real shop, and it puts a local florist first 100% of the time. And those people go on to buy, more readily than people who come from regular search.

I don't think it will stay this way forever. As soon as ads show up in AI answers, the big companies will arrive with their budgets. But until that happens, small shops have a window. And I think it's only going to get harder for the order gatherers. Their business is built on standing between the customer and the florist, and AI keeps getting better at connecting the two directly.

Mary and I started Kvetka with just the two of us, no investors and no ad budget. Today AI recommends us to people looking for flowers in Portland on a Friday evening. I can't think of better proof that honest work is finally starting to pay off.

Sources

• 1-800-Flowers Q3 fiscal 2026 earnings call, May 7, 2026: transcript via Investing.com
• The Wall Street Journal: 1-800-Flowers Moves To Invest In Its Brand After Overrelying On "Buying Clicks"
• CNBC: Flower delivery company FTD files for bankruptcy
• Search Engine Land: OpenAI: ChatGPT now has 900 million weekly active users
• Google: Gemini app hits 1 billion monthly active users
• Google: Alphabet Q2 2026 earnings call, Sundar Pichai remarks
• The Motley Fool: AI Overviews reach over 2.5 billion monthly users
• Digital Commerce 360: Adobe: AI-referred traffic to retail sites doubles in a year
• Search Engine Land: ChatGPT gains access to Yelp reviews, ratings, and photos
• U.S. Census Bureau: County Business Patterns by Industry: 2022
• Stock prices: MacroTrends, StockAnalysis.com
• Search volumes: Google Keyword Planner, United States, September 2025 to August 2026