Corporate B2B Websites
We Audited 35 Leading Ukrainian Brands: Where Businesses Lose Ad Spend on Mobile Speed, 5–83 MB Pages, and ChatGPT Blindness
Executive Summary: What Real 28-Day User Telemetry Revealed Across 35 Ukrainian Brands
Instead of reviewing banking monopolies or massive marketplaces with 200-person engineering departments, we ran Site Roast across 35 recognizable Ukrainian commercial brands spanning five high-acquisition-cost verticals: real estate developers, private medical clinics, Ukrainian D2C fashion labels, fitness/hospitality/automotive networks, and EdTech/B2B exporters.
Crucially, we refused to rely on volatile synthetic lab guesses alone. For 34 of the 35 companies, we pulled Google’s 28-day real-user Chrome UX (CrUX) field telemetry—measuring what actual smartphone visitors experience in production—while all 35 of 35 websites completed full HTML and AI audits. The average site scored 82/100 (median 83/100), yet 51% of audited pages (18 of 35) force smartphones to download 5 to 83 MB on initial load, 37% (13 of 35) miss the 2.5s Core Web Vitals threshold for real humans, 71% (25 of 35) lack an `/llms.txt` file for ChatGPT and Perplexity, and 26% (9 of 35) render broken link previews in Telegram. Below is the complete data-backed breakdown of where paid traffic leaks and how the top 5 brands win.
Ranking 5 Ukrainian Verticals: Where Websites Hold Up and Where Expensive Leads Drop Off
Grouping all 35 brands by industry exposed a clear structural divide: Ukrainian fashion labels and tech-focused EdTech companies ship the cleanest, fastest mobile experiences, while real estate developers—who pay the highest cost per lead in search and social ads—trail the field due to bloated 20+ MB assets and slow initial rendering.
1. Industry Benchmark Matrix Across All 5 Verticals
The table below summarizes verified averages across all 35 audited domains grouped by industry:
| Industry Vertical | Scanned HTML | Avg Score | Avg Mobile LCP (28d) | Avg Page Weight | Missing `llms.txt` | Broken Telegram OG | Category Leader |
|---|---|---|---|---|---|---|---|
| Ukrainian Fashion & Lifestyle | 7 of 7 | 86 / 100 | 1.8 s | 4.5 MB | 5 of 7 | 2 of 7 | GUNIA Project (98/100) |
| EdTech, B2B & Manufacturers | 7 of 7 | 85 / 100 | 2.2 s | 13.4 MB | 5 of 7 | 1 of 7 | Ajax Systems (95/100) |
| Private Medical Clinics | 7 of 7 | 81 / 100 | 2.4 s | 5.5 MB | 6 of 7 | 3 of 7 | Medicover (93/100) |
| Hospitality, Fitness & Automotive | 7 of 7 | 80 / 100 | 2.3 s | 19.7 MB | 4 of 7 | 2 of 7 | Winner Automotive (92/100) |
| Real Estate Developers | 7 of 7 | 77 / 100 | 3.7 s | 23.9 MB | 5 of 7 | 1 of 7 | RIEL (90/100) |
2. The High-CPL Paradox: Architecting Lead Capture Without Bloating the Hero
In real estate and B2B services, a single qualified inquiry from paid search can cost dozens of dollars. Yet developers in our audit averaged 23.9 MB per homepage and 3.7 seconds of real-user waiting time before the main hero renders—50% slower than Google’s 2.5s safe zone.
When we engineer turnkey corporate B2B websites and client acquisition systems, we keep heavy media decoupled from the above-the-fold conversion path: prospects immediately see a sharp value proposition, primary CTA, and key proof points without waiting on bloated background video. Our corporate build for WashService demonstrates this focused first-screen structure:

The Gap Between a Founder’s MacBook and 4G Mobile Traffic: Half of Sites Weigh 5–83 MB
Leadership teams almost always approve website builds on high-resolution laptops connected to fast office fiber. Meanwhile, 75–85% of paid social and search traffic arrives on smartphones over uneven 4G networks.
1. Payload Extremes & Real-User Delays Across 35 Domains
Field telemetry highlighted two costly technical bottlenecks:
37% (13 of 35 live domains) exceed the 2.5s LCP threshold for real smartphone visitors: notable slowdowns appeared on `Avalon` (6.6 s), `SAGA Development` (5.6 s lab 4G), `Biosphere` (4.8 s), `Oxford Medical` (4.3 s), and `Intergal-Bud` (4.2 s).
51% (18 of 35) ship 5 MB to 83 MB on a single page view: nine brands crossed 15 MB—led by `Emily Resort` (83.0 MB), `Avalon` (67.9 MB), `Laba` (55.4 MB), `SAGA Development` (26.3 MB), `Intergal-Bud` (24.4 MB), `APOLLO NEXT` (23.7 MB), `Kovalska` (19.6 MB), `KAN Development` (16.2 MB), and `SAT` (15.6 MB).
Note how switching to 28-day CrUX field metrics prevented false alarmism on pages with video backgrounds like `Laba` (`laba.ua`): synthetic mobile simulators previously penalized background media loops unrealistically, whereas real-user data confirms their hero renders in 2.5 s—yet visitors still pay a 55.4 MB mobile data penalty per load.
2. Engineering Lightweight Mobile Flows for One-Handed Users: Good Day Pharmacy Case Study
High-volume consumer platforms stay fast on 4G by pairing lightweight WebP media with exposed search and bottom-weighted navigation inside the 375px Thumb Zone. For the national Good Day Pharmacy mobile overhaul, we structured home search, category filtering, 1-tap pack-size toggles, and local pharmacy pickup booking into clean, thumb-ready screens:




ChatGPT & Perplexity Blindness: Why 71% of Brands Miss Organic AI Recommendations
Buyers increasingly ask ChatGPT, Perplexity, and Gemini to shortlist clinics, developers, or B2B vendors before ever clicking a traditional search ad. When AI crawlers cannot parse structured positioning or hit server errors, recommendations flow straight to better-structured competitors.
The Hidden Monthly Cost of a 24 MB Legacy Website on a $6,000 Paid Acquisition Budget
- Paid traffic conversion rate: 1.1% (mobile wait-time drop-off + vague headline)
- Qualified leads from $6,000 spend: 55 inquiries at $109.09 CPL
- AI search citation readiness: Missing llms.txt & Schema.org markup
- Net profit after ad spend: +$600/mo (ad costs consume margin)
- Paid traffic conversion rate: 2.8% (instant mobile render + interactive quiz)
- Qualified leads from $6,000 spend: 140 inquiries at $42.86 CPL (2.5x cheaper)
- AI answer & messenger readiness: Clean llms.txt, JSON-LD & shareable Telegram cards
- Net profit after ad spend: +$10,800/mo (+$10,200/mo monthly uplift)
Fixing mobile speed bottlenecks, vague hero copy, and AI-crawler blind spots on a flat $6,000 monthly ad budget cuts acquisition costs by 2.5x and pays back the engineering investment in two weeks.
1. AI Crawler Audit Findings: 25 of 35 Missing llms.txt & Zero Intentional Bot Blocks
Checking `/llms.txt` availability, `robots.txt` policies, and `Schema.org` structured data split our cohort sharply:
71% (25 of 35 scanned brands) have no `/llms.txt` file, and 29% (10 of 35) expose zero Schema.org JSON-LD markup. Private medical clinics lag the furthest with just 1 of 7 (`Oxford Medical`) publishing `llms.txt`, while real estate, fashion, and EdTech/B2B each have 2 of 7.
10 proactive brands already published `llms.txt`: `Mate academy`, `GUNIA Project`, `Oxford Medical`, `KAN Development`, `APOLLO NEXT`, `blago`, `VOVK`, `Emily Resort`, `Sport Life`, and `SAT`.
0 of 35 companies intentionally block AI bots in `robots.txt`: none of the audited brands restrict `GPTBot` or `ClaudeBot` on purpose, confirming that AI search invisibility stems from missing `/llms.txt` files and structured `Schema.org` data rather than firewall blocks.
2. Passing the 5-Second Test for Both Humans and LLMs: AImReply Case Study
Both busy executives and LLM summarizers rely on the primary `H1` and immediate product context to decide what a company actually offers. Notice how our SaaS landing page for AImReply removes ambiguity in five seconds with a concrete headline, instant email input, and live in-product visual proof:

Broken Telegram Link Cards, Missing H1s, and the Top 5 Highest-Scoring Sites
Conversion friction also shows up in everyday sharing workflows and initial lead capture. One quarter of the audited brands lose visual credibility whenever a link is pasted into client chats.
1. Messenger Previews (26% Broken) & Interactive Lead Capture Over Passive Forms
Exactly 9 of 35 (26%) audited brands fail basic Open Graph checks, producing blank or text-only link bubbles when shared in Telegram or LinkedIn—including household names such as `Dobrobut`, `GUNIA Project`, `SOLMAR`, `Emily Resort`, `G.Bar`, `Laba`, `Into-Sana`, `Smart Medical Center`, and `Avalon`. Another 10 of 35 (29%) either omit their `H1` tag completely or split it across scattered navigation blocks.
Replacing static 'Leave your phone' boxes with structured interactive estimators transforms cold clicks into qualified conversations, as shown in our WashService investor quiz calculator:

2. Spotlight on the Top 5 Highest-Scoring Ukrainian Websites (All Rendering in 1.2–1.4s)
Credit goes to the five engineering and marketing teams that led our entire 35-brand benchmark—every single one delivers 1.2–1.4s real-user mobile LCP:
GUNIA Project (`guniaproject.com`) — 98 / 100 (`1.2s`, `3.0 MB`): Top overall score with `100` in speed, AI visibility (`llms.txt` live), trust, and stack—needing only Open Graph polish for messengers.
Ajax Systems (`ajax.systems`) — 95 / 100 (`1.2s`, `3.1 MB`): Fast, resilient international architecture; adding `/llms.txt` would push it near a perfect 100.
Medicover (`medicover.ua`) — 93 / 100 (`1.4s`, `1.3 MB`): Lightest and quickest patient-facing site across all audited medical clinics.
Etnodim (`etnodim.ua`) — 92 / 100 (`1.4s`, `4.3 MB`): Zero critical issues and crisp catalog speed on mobile.
Winner Automotive (`winnerauto.ua`) — 92 / 100 (`1.4s`, `4.6 MB`): Benchmark performance and trust signals in the automotive retail vertical.
Strong showings also came from Oberig Clinic (`91/100`, `1.3s`), RIEL (`90/100`, `1.0 MB`), Mate academy (`88/100`, `2.0s`, `llms.txt` live), and HARVEST (`88/100`, `1.1s`).
Complete 35-Brand Benchmark Table & 30-Second Self-Audit Protocol
Review the full dataset below to see how recognizable competitors in your vertical perform on real-user mobile speed, page payload, `llms.txt`, and Telegram preview health.
1. Complete Results Table Across All 35 Ukrainian Companies
Metrics list overall score (`/100`), mobile LCP (`28d real users` vs `lab`), homepage weight, `llms.txt` presence, and Telegram Open Graph status:
| # | Brand | Vertical | Score | Mobile LCP | Weight | `llms.txt` | Telegram OG |
|---|---|---|---|---|---|---|---|
| 1 | KAN Development | Real Estate | 86 | 1.9 s (28d) | 16.2 MB | Yes | OK |
| 2 | Intergal-Bud | Real Estate | 68 | 4.2 s (28d) | 24.4 MB | No | OK |
| 3 | RIEL | Real Estate | 90 | 2.4 s (28d) | 1.0 MB | No | OK |
| 4 | SAGA Development | Real Estate | 63 | 5.6 s (lab) | 26.3 MB | No | OK |
| 5 | Kovalska | Real Estate | 84 | 1.6 s (28d) | 19.6 MB | No | OK |
| 6 | blago | Real Estate | 80 | 3.4 s (28d) | 11.9 MB | Yes | OK |
| 7 | Avalon | Real Estate | 69 | 6.6 s (28d) | 67.9 MB | No | Broken |
| 8 | Dobrobut | Medical Clinics | 81 | 2.2 s (28d) | 4.0 MB | No | Broken |
| 9 | Oberig Clinic | Medical Clinics | 91 | 1.3 s (28d) | 2.6 MB | No | OK |
| 10 | Into-Sana | Medical Clinics | 68 | 2.7 s (28d) | 10.7 MB | No | Broken |
| 11 | Oxford Medical | Medical Clinics | 87 | 4.3 s (28d) | 5.3 MB | Yes | OK |
| 12 | Smart Medical Center | Medical Clinics | 65 | 3.0 s (28d) | 9.7 MB | No | Broken |
| 13 | Garvis | Medical Clinics | 81 | 1.9 s (28d) | 4.8 MB | No | OK |
| 14 | Medicover | Medical Clinics | 93 | 1.4 s (28d) | 1.3 MB | No | OK |
| 15 | MustHave | Fashion & Lifestyle | 76 | 1.9 s (28d) | 11.1 MB | No | OK |
| 16 | VOVK | Fashion & Lifestyle | 83 | 1.4 s (28d) | 3.2 MB | Yes | OK |
| 17 | brabrabra | Fashion & Lifestyle | 81 | 2.9 s (28d) | 4.6 MB | No | OK |
| 18 | GUNIA Project | Fashion & Lifestyle | 98 | 1.2 s (28d) | 3.0 MB | Yes | Broken |
| 19 | Etnodim | Fashion & Lifestyle | 92 | 1.4 s (28d) | 4.3 MB | No | OK |
| 20 | SOLMAR | Fashion & Lifestyle | 84 | 2.8 s (28d) | 3.5 MB | No | Broken |
| 21 | HARVEST | Fashion & Lifestyle | 88 | 1.1 s (28d) | 1.5 MB | No | OK |
| 22 | Emily Resort | Hospitality & Services | 79 | 2.3 s (28d) | 83.0 MB | Yes | Broken |
| 23 | Bukovel | Hospitality & Services | 83 | 3.4 s (28d) | 6.8 MB | No | OK |
| 24 | Sport Life | Hospitality & Services | 78 | 2.9 s (28d) | 5.5 MB | Yes | OK |
| 25 | APOLLO NEXT | Hospitality & Services | 78 | 3.0 s (28d) | 23.7 MB | Yes | OK |
| 26 | G.Bar | Hospitality & Services | 63 | 2.0 s (28d) | 10.4 MB | No | Broken |
| 27 | VIDI | Hospitality & Services | 85 | 1.0 s (28d) | 4.2 MB | No | OK |
| 28 | Winner Automotive | Hospitality & Services | 92 | 1.4 s (28d) | 4.6 MB | No | OK |
| 29 | Laba | EdTech & B2B | 79 | 2.5 s (28d) | 55.4 MB | No | Broken |
| 30 | Projector | EdTech & B2B | 85 | 1.6 s (28d) | 3.5 MB | No | OK |
| 31 | Mate academy | EdTech & B2B | 88 | 2.0 s (28d) | 2.3 MB | Yes | OK |
| 32 | Kormotech | EdTech & B2B | 76 | 1.8 s (28d) | 10.6 MB | No | OK |
| 33 | Biosphere | EdTech & B2B | 84 | 4.8 s (28d) | 3.0 MB | No | OK |
| 34 | SAT | EdTech & B2B | 87 | 1.6 s (28d) | 15.6 MB | Yes | OK |
| 35 | Ajax Systems | EdTech & B2B | 95 | 1.2 s (28d) | 3.1 MB | No | OK |
2. Quick Tweaks vs. Turnkey Rebuild: Benchmark Your Domain Now
If your site already converts reliably and simply lacks `/llms.txt`, clean `og:image` tags, or WebP compression, you only need a short engineering sprint. However, if mobile users wait 4+ seconds to download 20+ MB while reading a vague hero headline, minor styling patches won’t save your acquisition budget.
Run a free 30-second website audit in Site Roast right now to inspect your 28-day CrUX speed, ChatGPT readiness, live Telegram preview, and first-screen offer.
Key Takeaways
Want to see where your website scores alongside these 35 brands?
51% (18 of 35) of audited Ukrainian brands ship 5–83 MB per homepage load, and 37% (13 of 35) fail the 2.5s real-user mobile speed threshold.
71% (25 of 35) still lack an /llms.txt file for ChatGPT and Perplexity, leaving organic AI recommendations to faster competitors.
26% (9 of 35) display broken link cards when shared in Telegram, and 29% (10 of 35) omit or misconfigure their primary H1 headline.
Combining sub-2s mobile rendering, clear first-screen offers, interactive lead capture, and AI readiness slashes CPL by 2–2.5x within the first month.


