Study · September 2026
383 flower delivery websites: what stops them from getting orders
We checked 383 local flower delivery sites in 20 Russian cities. Most orders slip away on the phone: about two thirds show no price on the first screen, more than 8 in 10 have buttons too small for a thumb, and 37% share links without a bouquet photo. Scans ran July 28 to September 28, 2026, with a manual recheck on September 29.
Vlad, VladSEO · published Sept 29, 2026
- 383
- local shop websites
- 20
- Russian cities, Moscow to Vladivostok
- 40
- sites rechecked by hand on a phone
- 114
- requests to load the home page, median
How we checked
From July 28 to September 28, 2026 we searched for local flower delivery shops in 20 large Russian cities. Marketplaces and delivery apps were dropped at the start. A scanner read each home page and a few inner pages, then opened the site in real Chrome at phone size, 390×844, to measure page height, button sizes, where the first price sits and how many requests load.
On September 29 we took 40 random sites and checked them again with our own code, independent of the scanner, and judged the first-screen price by eye from a screenshot of each. Server speed was measured for all 383 sites from a server in Moscow: three runs, median.
This isn't a random sample of the market. These are shops that came up in search for flower delivery in their city, so the right wording is "out of 383 local shop websites in 20 cities".
From 468 sites to 383
- 468domains found in search
- −27home page didn't open
- −43dropped: chains, wholesale, no catalog, duplicates, parked domains
- −15only a bot wall or an empty page
- 383sites in the study
How far to trust the scanner
- 1For each check we counted how often the scanner agreed with the manual result across the 40 sites. The bar for publishing was 85%.
- 2Preview image, request count, page length, H1 and the answer to a missing page agreed on 95-97.5% of sites. Those shares are published as is.
- 3Where the scanner erred in one direction, we recalculated the share from the manual check and rounded it: it found prices on the first screen that weren't visible, and missed WebP that was there. Those rows are marked with an asterisk.
The main findings in one table
Shares are counted from the sites where the check worked: 373 for code checks, 372 for the phone pass.
| Mistake | Share of sites | How it costs orders | How to check on a phone |
|---|---|---|---|
| No product price on the first phone screen | ≈ 2/3* | How it costs ordersNo price in sight, so the visitor goes to compare where prices show right away. | How to check on a phoneOpen the home page and don't scroll. Can you see a single bouquet price? |
| Most buttons and links under 44 px | 85% | How it costs ordersThumbs land on the wrong thing, people get annoyed and leave. | How to check on a phoneTap the menu items and the Buy button with your thumb. Did you hit them first try? |
| Home page longer than 14 screens | 48% | How it costs ordersContacts, reviews and delivery terms at the bottom barely get seen. | How to check on a phoneScroll to the end and count the swipes. |
| More than 120 requests on load | 45% | How it costs ordersOn a weak connection the page takes long to assemble, the first screen stays blank. | How to check on a phoneTurn off Wi-Fi, open the site and time how long until you see bouquets. |
| No preview image (og:image) | 37% | How it costs ordersBouquets get shared in messengers, and a link without a photo looks like a bare address. | How to check on a phoneSend yourself the home page link on WhatsApp. Does a photo show up? |
| Photos without WebP or AVIF | ≈ 60%* | How it costs ordersHeavy photos load slowly on mobile data, some shoppers leave before the bouquets appear. | How to check on a phoneRun the home page through PageSpeed Insights and look for the modern image formats item. |
| No H1 on the home page | 18% | How it costs ordersSearch engines can't tell as easily that this is flower delivery in your city. | How to check on a phoneIn Chrome on Android, type view-source: before the address and search for "<h1". |
| Both www and non-www addresses open | 16% | How it costs ordersOne home page lives at two addresses and search splits the weight. | How to check on a phoneType the address with and without www. One of them should switch on its own. |
| Server starts answering after 0.8 s | 11% | How it costs ordersWhile the server thinks, the shopper stares at a white screen. | How to check on a phoneOpen the site on mobile data. White screen for over a second: talk to your host. |
| No Yandex Metrica analytics | ≈ 1 in 10* | How it costs ordersThe owner can't see where orders come from or where they're lost. | How to check on a phoneOpen your analytics and find yesterday's orders by source. Can't do it in a minute? You don't really have analytics. |
| A missing page answers 200 instead of 404 | 2% | How it costs ordersSearch spends its crawl on junk addresses instead of the catalog. | How to check on a phoneAdd /check-123 to your address. You should see a "not found" page. |
* share from the manual check of 40 sites, rounded
Speed: the weight of the page
Measured from Moscow, the servers answer fast: median time to first byte 0.25 s, full home page response 0.3 s. Only 11% start answering after 0.8 s, and only 3% take longer than 2.5 s to deliver the HTML.
The time goes into everything that loads after the HTML: a median of 114 requests on the phone, more than 120 on 45% of sites, and photos without WebP on about 60%. We measured from a data center; on a mobile network it's slower.
What the scanner can't see
The scanner doesn't go through the cart, checkout or payment, and that's where a shop loses people who already decided to buy. In our manual reviews we order from a phone up to the payment page and find things the table doesn't show, like "in stock" in the search markup while the page only takes pre-orders. We don't publish shares for these: automated detection was wrong too often.
What we chose not to publish
Some scanner numbers failed the manual check. We dropped them, even though they sound more dramatic.
- 1Slow home pages as measured by the scanner. It ran from Mexico, and the round trip across the ocean added seconds.
- 2"Layout jumps while loading". Missing image sizes are common, but they matched real layout shift poorly.
- 3"A permanent crossed-out price". On the 40 hand-checked sites we never found one on half of the home page prices.
- 4Ratings without reviews, "in stock" on pre-order items, missing business markup: too many false positives to publish a share.
Check your own site in 10 minutes
All you need is a phone on mobile data. Look at it like a customer seeing your site for the first time.
- 1
First screen
Open the home page and don't scroll. Can you see a bouquet price and an order button?
- 2
Thumb
Tap five menu items and the Buy button with your thumb. One miss means the buttons are too small.
- 3
Shared link
Send yourself the home page and one bouquet. Both should show a photo and a name.
- 4
Mobile data
Turn off Wi-Fi, reload and time how long until you see bouquets.
- 5
Page length
Scroll to the end. More than 14 swipes means delivery terms and reviews should move up.
- 6
Checkout
Order for tomorrow at a set time: date, time, address, card message. Stop at the payment page.
- 7
Addresses
Type the address with and without www, then add /check-123. The first should redirect, the second should say "not found".
- 8
Analytics
Find yesterday's orders by source. If that takes more than a minute, your ad and site decisions are guesses.
What this looks like in our own shop
We run a flower delivery shop in Krasnodar. In the summer of 2026 we rebuilt its product pages, sped up the site and fixed checkout on phones. Completed site orders for September 1-22 went from 147 a year ago to 183, up 24%, and phone conversion from 2.5% to 3.8%. Google traffic hasn't recovered from the May 2026 update: 26% fewer visitors from Google for September 1-15 than a year before.
Questions
Which mistakes should I fix first?
The ones on the first phone screen: price, order button and button size. Then the preview image for messengers, which is one evening's work. Speed and technical items come next.
Can I cite this data?
Yes, with a link to this page. Please word it as "out of 383 local flower delivery sites in 20 Russian cities, VladSEO, September 2026".
Is my shop in the study?
Only Russian shops are. We don't publish or share shop names or addresses. For a review of your own site, send us the address. It's free.
Limits
- 1Not a random sample: shops found through search, some of them looked up as review candidates.
- 2Scans ran over two months, the manual check on one day. Some sites changed in between.
- 3The manual sample is 40 sites, with a margin of about ±15 points on a share.
- 4Sites behind bot protection or blocking foreign IPs are underrepresented.
- 5Speed was measured from a Moscow data center, not from a phone.
- 6All sites are Russian. Treat the shares as a Russian benchmark.
Free, 2-3 days
Want to know which of these your site has?
Send us the address. The review is free, takes 2-3 days and comes with measurements and phone screenshots.
- Every check from the table, run on your site
- Checkout on a phone up to the payment page
- What to fix first and what your developer can do alone
We reply in writing within a couple of hours. No calls needed and no sales script.
Numbers you can check
- ≈400
- orders a month in the flower shop I run
- 36.2%
- of orders in Sept 2026 came from returning customers, up from 28.1% in May
- 10 years
- that site in search, full history on the table
The same case shows a 51% Google drop in May 2026. Most people hide that. I show it and write what I am doing about it.