Taras Denisenko
UX Cases
Portfolio
June 2024–November 2025
This is a selected set of UX cases created for the product work at HUG. They highlight research, user flows, problem solving and interaction logic, showing practical UX thinking.
UX is not about pretty Dribbble shots. It is about solving real problems and driving revenue.
I. Semantic Data
Layer Setup
Some details have been removed due to NDA.
This case shows how cleaner metadata improves discoverability as AI agents overtake traditional search in content navigation (background: The Economist).
Problem
HUG is built for human reading, but people now rely on LLM results instead of visiting websites.
Our content is not machine readable, so AI systems cannot understand what HUG is,
cannot extract facts and cannot connect related information.
Tests with ChatGPT and Gemini confirmed that HUG is not recognised
or surfaced in AI generated answers.
Solution
Add a semantic layer that makes HUG content readable for AI systems
while keeping the existing UX unchanged.
JSON LD and schema.org will describe what each page represents,
define the key entities and show how they connect.
This gives LLMs a reliable structure instead of inferring meaning from plain text.
Planned Work
- Implement JSON LD across core HUG pages;
- Align metadata so all content follows one unified model;
- Map relationships between profiles, projects and key sections in a format LLMs can index;
- Apply ai.txt rules and connect the semantic layer to the future AI first endpoint.
Expected Results
- LLM visibility increasing from 15% to 55%;
- Around 25% lift in AI driven traffic from clearer metadata;
- HUG appearing in relevant AI generated queries instead of being ignored.
Real World Test
“Open Calls for digital artists 2025” — response: “2025 Virtual Exhibition + Artist Book” (opencallforartists.com), with the theme “Memory”, a deadline, and a link. HUG.ART wasn’t mentioned.
“What is HUG.ART?” — response: “A platform for artists” (general, lacks context). With structured data, Gemini would display HUG open calls along with the host, deadline, theme, and link.
User View
AI View (Current)
AI View (LLM Optimised)
With JSON-LD, Gemini would display HUG Open Calls in this structured way:
II. Reducing
Checkout Friction
Some details have been removed due to NDA.
This case demonstrates the ability to diagnose cross-platform funnel failures and remove structural blockers.
Problem
HUG campaign traffic sent to GET.ART was failing to convert: the homepage showed 6,692 views with only 50 seconds of engagement, search held ~10 seconds, the where-to-buy screen generated high exits, and the cart (/cp/cart.php, 2:59) forced users into mandatory sign-up.
These drop-offs reduced the effectiveness of HUG campaigns that promoted .ART domains and lowered overall registration yield.
Solution
Removed the mandatory sign-up wall inside GET.ART checkout, allowing users from HUG campaigns to pay first and create an account later.
Surfaced Google and Apple login at the top and reduced form friction to restore a low-barrier purchase flow.
Work Done
- Tracked the HUG → GET.ART funnel in GA4 and pinpointed drop-offs
- Identified the forced sign-up step as the main checkout blocker
- Switched to a pay-first flow with simple email capture
- Moved Google/Apple login to the top and reduced form friction
Results
- Cart drop-off decreased by 18%
- Checkout completion increased by 12%
- Google/Apple login usage increased by 27%
- Average purchase time reduced by 22%
- HUG saw a 9% uplift in successful registrations from campaign traffic once the broken domain-purchase path was fixed on GET.ART
Checkout Entry
Drop-Off Flow
GA4 Traffic
Engagement (sec)
Page Engagement
- /en: 6,692 views, avg. engagement 50s
- /en/search: ~10s engagement, weak intent signal
- /cp/cart.php: 2m 59s engagement, forced sign-up caused drop-off
Impact Summary
+
pp
Recovered HUG registrations from campaign traffic after fixing the domain–checkout path.
+12%
55%
43%
-18%
55%
37%
-22%
100%
78%
III. Search Relevance Upgrade
Some details have been removed due to NDA.
This case shows how backend refinement boosts search relevance and lowers friction without any UI changes.
Problem
Search on the HUG homepage wasn’t returning relevant results.
Analytics showed high drop-off during search, and only a small share of users continued to a profile.
Many users couldn’t find artists they already knew, leading to repeated attempts and dead-end queries.
Solution
Search was rebuilt to surface relevant results faster.
Elasticsearch was introduced to deliver immediate suggestions and stronger ranking.
Autocomplete was improved to guide queries, reduce typing effort and present clearer, meaningful options.
Work Done
- Reviewed GA behaviour (queries, drop-off, continuation rate)
- Analysed heatmaps and session recordings
- Refined autocomplete behaviour and selection states
- Tested the new flow with real users
- Prepared technical notes and edge-case rules for dev and QA
Results
- Search-driven actions increased from 15% to 55% (+40 pp)
- Continuation improved by 25 pp
- Profile traffic grew by 25%
- Fewer abandoned searches and clearer query paths
Legacy Search UI
- Exact-match only
- No fuzzy matching
- No synonyms
- No ranking (flat list)
- Low relevance
- Unhelpful suggestions
- Variant matches (“chole”, “cloe”)
- Typo-tolerant matching
- Synonym / keyword expansion
- Boosted artist profiles
- Prioritised active projects
- Fast, relevant suggestions
Search Flow Comparison
Checkout Fixes
Reduces effort and speeds up input
Boosts relevance of top results
Increases first-click success rate
Reduces empty and failed searches
Impact Summary
+
pp
+40 pp
55%
15%
+25 pp
40%
15%
+25%
125%
cap.
100%
IV. Open Call Recovery
Some details have been removed due to NDA.
This case shows how structured UX thinking can restore consistency and clarity in unstable environments.
Problem
A last-minute rule almost derailed our Open Call submissions.
Artists suddenly had to provide a .ART domain or an ID.ART profile to apply.
It surfaced without warning, appeared too late, and was phrased differently across competitions.
This caused confusion, more support tickets and fewer finished submissions.
Solution
Take a systemic view of the issue to understand why the rule caused disruption.
Reduce its impact by showing it earlier, making it clearer, and presenting it the same way across all Open Calls.
Use this outcome to define simple principles for how future mandatory rules should be introduced.
Work Done
- Audited all Open Calls to find inconsistent wording and missing context
- Gave users a short explanation of why a .ART domain or ID.ART profile is required
- Standardised how the rule is shown across Open Calls
- Created a framework for clear, consistent mandatory rules
- Added an early link to the promo page
Results
- Submission drop-off improved from 63% to 49% (−14 pp) after making the rule clear and visible earlier
- Support tickets related to the rule decreased by 31% during the following two weeks
- User confusion reports fell by 39% across feedback channels, based on Amplitude, Zendesk and survey data
Inconsistencies
Submission Flow Comparison
Timing vs Drop-off
Drop-off (%)
Impact Summary
-
pp
-14 pp
63%
49%
-31%
380
cap.
260
-39%
50%
31%
cap.
V. Cross-Product Promo Code Launch
Some details have been removed due to NDA.
This case shows how aligning promo code behaviour across products created an immediate revenue lift with minimal UX changes.
Problem
Registration conversion for .ART domains was 7.3%.
Google Analytics showed a 65% drop-off on GET.ART during domain entry.
Users did not understand how HUG and GET.ART were connected, and the integration delivered no visible value.
Solution
A simple cross-product mechanic was introduced:
a promo code claimed in HUG and applied on GET.ART.
The flow was designed mobile-first with clear steps and guidance.
A Figma prototype was aligned with backend and marketing.
Work Done
- Behaviour and drop-off analysis in Microsoft Clarity
- 20 user tests of the new flow
- Coordination with marketing + backend
- API requirements and implementation logic for promo codes
Results
- 45+ registrations in the first month
- ~3 new registrations per week
- Revenue reached USD 4,265 from 1,643 domains
- Conversion improved from 7.3% to 9.5%
- Drop-off reduced from 65% to 25%
Before (Old Flow)
Cross-Product Flow
After (New Flow)
Impact Summary
USD
9.5%
7.3%
65%
25%
The earlier HUG brand and visual foundation were created by the previous design team led by Creative Director Lindsey Gemmill. My contribution at .ART centred on UX updates, usability fixes, and iterative improvements during the post-acquisition phase.
Contact
rebotics@protonmail.com
© 2025 Taras Denisenko
Taras Denisenko
UX Cases
Portfolio
June 2024–November 2025
This is a selected set of UX cases created for the product work at HUG. They highlight research, user flows, problem solving and interaction logic, showing practical UX thinking.
UX is not about pretty Dribbble shots. It is about solving real problems and driving revenue.
I. Semantic Data Layer Setup
Some details have been removed due to NDA.
This case shows how cleaner metadata improves discoverability as AI agents overtake traditional search in content navigation (background: The Economist).
Problem
HUG is built for human reading, but people now rely on LLM results instead of visiting websites.
Our content is not machine readable, so AI systems cannot understand what HUG is,
cannot extract facts and cannot connect related information.
Tests with ChatGPT and Gemini confirmed that HUG is not recognised
or surfaced in AI generated answers.
Solution
Add a semantic layer that makes HUG content readable for AI systems
while keeping the existing UX unchanged.
JSON LD and schema.org will describe what each page represents,
define the key entities and show how they connect.
This gives LLMs a reliable structure instead of inferring meaning from plain text.
Planned Work
- Implement JSON LD across core HUG pages;
- Align metadata so all content follows one unified model;
- Map relationships between profiles, projects and key sections in a format LLMs can index;
- Apply ai.txt rules and connect the semantic layer to the future AI first endpoint.
Expected Results
- LLM visibility increasing from 15% to 55%;
- Around 25% lift in AI driven traffic from clearer metadata;
- HUG appearing in relevant AI generated queries instead of being ignored.
Real World Test
“Open Calls for digital artists 2025” — response: “2025 Virtual Exhibition + Artist Book” (opencallforartists.com), with the theme “Memory”, a deadline, and a link. HUG.ART wasn’t mentioned.
“What is HUG.ART?” — response: “A platform for artists” (general, lacks context). With structured data, Gemini would display HUG open calls along with the host, deadline, theme, and link.
User View
AI View (Current)
AI View (LLM Optimised)
With JSON-LD, Gemini would display HUG Open Calls in this structured way:
II. Reducing Checkout Friction
Some details have been removed due to NDA.
This case demonstrates the ability to diagnose cross-platform funnel failures and remove structural blockers.
Problem
HUG campaign traffic sent to GET.ART was failing to convert: the homepage showed 6,692 views with only 50 seconds of engagement, search held ~10 seconds, the where-to-buy screen generated high exits, and the cart (/cp/cart.php, 2:59) forced users into mandatory sign-up.
These drop-offs reduced the effectiveness of HUG campaigns that promoted .ART domains and lowered overall registration yield.
Solution
Removed the mandatory sign-up wall inside GET.ART checkout, allowing users from HUG campaigns to pay first and create an account later.
Surfaced Google and Apple login at the top and reduced form friction to restore a low-barrier purchase flow.
Work Done
- Tracked the HUG → GET.ART funnel in GA4 and pinpointed drop-offs
- Identified the forced sign-up step as the main checkout blocker
- Switched to a pay-first flow with simple email capture
- Moved Google/Apple login to the top and reduced form friction
Results
- Cart drop-off decreased by 18%
- Checkout completion increased by 12%
- Google/Apple login usage increased by 27%
- Average purchase time reduced by 22%
- HUG saw a 9% uplift in successful registrations from campaign traffic once the broken domain-purchase path was fixed on GET.ART
Checkout Entry
Drop-Off Flow
GA4 Traffic
Engagement (sec)
Page Engagement
- /en: 6,692 views, avg. engagement 50s
- /en/search: ~10s engagement, weak intent signal
- /cp/cart.php: 2m 59s engagement, forced sign-up caused drop-off
Impact Summary
+
pp
Recovered HUG registrations from campaign traffic after fixing the domain–checkout path.
+12%
55%
43%
-18%
55%
37%
-22%
100%
78%
III. Search Relevance Upgrade
Some details have been removed due to NDA.
This case shows how backend refinement boosts search relevance and lowers friction without any UI changes.
Problem
Search on the HUG homepage wasn’t returning relevant results.
Analytics showed high drop-off during search, and only a small share of users continued to a profile.
Many users couldn’t find artists they already knew, leading to repeated attempts and dead-end queries.
Solution
Search was rebuilt to surface relevant results faster.
Elasticsearch was introduced to deliver immediate suggestions and stronger ranking.
Autocomplete was improved to guide queries, reduce typing effort and present clearer, meaningful options.
Work Done
- Reviewed GA behaviour (queries, drop-off, continuation rate)
- Analysed heatmaps and session recordings
- Refined autocomplete behaviour and selection states
- Tested the new flow with real users
- Prepared technical notes and edge-case rules for dev and QA
Results
- Search-driven actions increased from 15% to 55% (+40 pp)
- Continuation improved by 25 pp
- Profile traffic grew by 25%
- Fewer abandoned searches and clearer query paths
Legacy Search UI
- Exact-match only
- No fuzzy matching
- No synonyms
- No ranking (flat list)
- Low relevance
- Unhelpful suggestions
- Variant matches (“chole”, “cloe”)
- Typo-tolerant matching
- Synonym / keyword expansion
- Boosted artist profiles
- Prioritised active projects
- Fast, relevant suggestions
Search Flow Comparison
Checkout Fixes
Reduces effort and speeds up input
Boosts relevance of top results
Increases first-click success rate
Reduces empty and failed searches
Impact Summary
+
pp
+40 pp
55%
15%
+25 pp
40%
15%
+25%
125%
c.
100%
IV. Open Call Recovery
Some details have been removed due to NDA.
This case shows how structured UX thinking can restore consistency and clarity in unstable environments.
Problem
A last-minute rule almost derailed our Open Call submissions.
Artists suddenly had to provide a .ART domain or an ID.ART profile to apply.
It surfaced without warning, appeared too late, and was phrased differently across competitions.
This caused confusion, more support tickets and fewer finished submissions.
Solution
Take a systemic view of the issue to understand why the rule caused disruption.
Reduce its impact by showing it earlier, making it clearer, and presenting it the same way across all Open Calls.
Use this outcome to define simple principles for how future mandatory rules should be introduced.
Work Done
- Audited all Open Calls to find inconsistent wording and missing context
- Gave users a short explanation of why a .ART domain or ID.ART profile is required
- Standardised how the rule is shown across Open Calls
- Created a framework for clear, consistent mandatory rules
- Added an early link to the promo page
Results
- Submission drop-off improved from 63% to 49% (−14 pp) after making the rule clear and visible earlier
- Support tickets related to the rule decreased by 31% during the following two weeks
- User confusion reports fell by 39% across feedback channels, based on Amplitude, Zendesk and survey data
Inconsistencies
Submission Flow Comparison
Timing vs Drop-off
Drop-off (%)
Impact Summary
-
pp
-14 pp
63%
49%
-31%
380
c.
260
-39%
50%
31%
V. Cross-Product Promo Code Launch
Some details have been removed due to NDA.
This case shows how aligning promo code behaviour across products created an immediate revenue lift with minimal UX changes.
Problem
Registration conversion for .ART domains was 7.3%.
Google Analytics showed a 65% drop-off on GET.ART during domain entry.
Users did not understand how HUG and GET.ART were connected, and the integration delivered no visible value.
Solution
A simple cross-product mechanic was introduced:
a promo code claimed in HUG and applied on GET.ART.
The flow was designed mobile-first with clear steps and guidance.
A Figma prototype was aligned with backend and marketing.
Work Done
- Behaviour and drop-off analysis in Microsoft Clarity
- 20 user tests of the new flow
- Coordination with marketing + backend
- API requirements and implementation logic for promo codes
Results
- 45+ registrations in the first month
- ~3 new registrations per week
- Revenue reached USD 4,265 from 1,643 domains
- Conversion improved from 7.3% to 9.5%
- Drop-off reduced from 65% to 25%
Before (Old Flow)
Cross-Product Flow
After (New Flow)
Impact Summary
USD
9.5%
7.3%
65%
25%
The earlier HUG brand and visual foundation were created by the previous design team led by Creative Director Lindsey Gemmill. My contribution at .ART centred on UX updates, usability fixes, and iterative improvements during the post-acquisition phase.
Contact
rebotics@protonmail.com
© 2025 Taras Denisenko
Taras Denisenko
UX Cases
Portfolio
June 2024–November 2025
This is a selected set of UX cases created for the product work at HUG. They highlight research, user flows, problem solving and interaction logic, showing practical UX thinking.
UX is not about pretty Dribbble shots. It is about solving real problems and driving revenue.
I. Semantic Data Layer Setup
Some details have been removed due to NDA.
This case shows how cleaner metadata improves discoverability as AI agents overtake traditional search in content navigation (background: The Economist).
Problem
HUG is built for human reading, but people now rely on LLM results instead of visiting websites.
Our content is not machine readable, so AI systems cannot understand what HUG is,
cannot extract facts and cannot connect related information.
Tests with ChatGPT and Gemini confirmed that HUG is not recognised
or surfaced in AI generated answers.
Solution
Add a semantic layer that makes HUG content readable for AI systems
while keeping the existing UX unchanged.
JSON LD and schema.org will describe what each page represents,
define the key entities and show how they connect.
This gives LLMs a reliable structure instead of inferring meaning from plain text.
Planned Work
- Implement JSON LD across core HUG pages;
- Align metadata so all content follows one unified model;
- Map relationships between profiles, projects and key sections in a format LLMs can index;
- Apply ai.txt rules and connect the semantic layer to the future AI first endpoint.
Expected Results
- LLM visibility increasing from 15% to 55%;
- Around 25% lift in AI driven traffic from clearer metadata;
- HUG appearing in relevant AI generated queries instead of being ignored.
Real World Test
“Open Calls for digital artists 2025” — response: “2025 Virtual Exhibition + Artist Book” (opencallforartists.com), with the theme “Memory”, a deadline, and a link. HUG.ART wasn’t mentioned.
“What is HUG.ART?” — response: “A platform for artists” (general, lacks context). With structured data, Gemini would display HUG open calls along with the host, deadline, theme, and link.
User View
AI View (Current)
AI View (LLM Optimised)
With JSON-LD, Gemini would display HUG Open Calls in this structured way:
II. Reducing Checkout Friction
Some details have been removed due to NDA.
This case demonstrates the ability to diagnose cross-platform funnel failures and remove structural blockers.
Problem
HUG campaign traffic sent to GET.ART was failing to convert: the homepage showed 6,692 views with only 50 seconds of engagement, search held ~10 seconds, the where-to-buy screen generated high exits, and the cart (/cp/cart.php, 2:59) forced users into mandatory sign-up.
These drop-offs reduced the effectiveness of HUG campaigns that promoted .ART domains and lowered overall registration yield.
Solution
Removed the mandatory sign-up wall inside GET.ART checkout, allowing users from HUG campaigns to pay first and create an account later.
Surfaced Google and Apple login at the top and reduced form friction to restore a low-barrier purchase flow.
Work Done
- Tracked the HUG → GET.ART funnel in GA4 and pinpointed drop-offs
- Identified the forced sign-up step as the main checkout blocker
- Switched to a pay-first flow with simple email capture
- Moved Google/Apple login to the top and reduced form friction
Results
- Cart drop-off decreased by 18%
- Checkout completion increased by 12%
- Google/Apple login usage increased by 27%
- Average purchase time reduced by 22%
- HUG saw a 9% uplift in successful registrations from campaign traffic once the broken domain-purchase path was fixed on GET.ART
Checkout Entry
Drop-Off Flow
GA4 Traffic
Engagement (sec)
Page Engagement
- /en: 6,692 views, avg. engagement 50s
- /en/search: ~10s engagement, weak intent signal
- /cp/cart.php: 2m 59s engagement, forced sign-up caused drop-off
Impact Summary
+
pp
Recovered HUG registrations from campaign traffic after fixing the domain–checkout path.
+12%
55%
43%
-18%
55%
37%
-22%
100%
78%
III. Search Relevance Upgrade
Some details have been removed due to NDA.
This case shows how backend refinement boosts search relevance and lowers friction without any UI changes.
Problem
Search on the HUG homepage wasn’t returning relevant results.
Analytics showed high drop-off during search, and only a small share of users continued to a profile.
Many users couldn’t find artists they already knew, leading to repeated attempts and dead-end queries.
Solution
Search was rebuilt to surface relevant results faster.
Elasticsearch was introduced to deliver immediate suggestions and stronger ranking.
Autocomplete was improved to guide queries, reduce typing effort and present clearer, meaningful options.
Work Done
- Reviewed GA behaviour (queries, drop-off, continuation rate)
- Analysed heatmaps and session recordings
- Refined autocomplete behaviour and selection states
- Tested the new flow with real users
- Prepared technical notes and edge-case rules for dev and QA
Results
- Search-driven actions increased from 15% to 55% (+40 pp)
- Continuation improved by 25 pp
- Profile traffic grew by 25%
- Fewer abandoned searches and clearer query paths
Legacy Search UI
- Exact-match only
- No fuzzy matching
- No synonyms
- No ranking (flat list)
- Low relevance
- Unhelpful suggestions
- Variant matches (“chole”, “cloe”)
- Typo-tolerant matching
- Synonym / keyword expansion
- Boosted artist profiles
- Prioritised active projects
- Fast, relevant suggestions
Search Flow Comparison
Checkout Fixes
Reduces effort and speeds up input
Boosts relevance of top results
Increases first-click success rate
Reduces empty and failed searches
Impact Summary
+
pp
+40 pp
55%
15%
+25 pp
40%
15%
+25%
125%
cap.
100%
IV. Open Call Recovery
Some details have been removed due to NDA.
This case shows how structured UX thinking can restore consistency and clarity in unstable environments.
Problem
A last-minute rule almost derailed our Open Call submissions.
Artists suddenly had to provide a .ART domain or an ID.ART profile to apply.
It surfaced without warning, appeared too late, and was phrased differently across competitions.
This caused confusion, more support tickets and fewer finished submissions.
Solution
Take a systemic view of the issue to understand why the rule caused disruption.
Reduce its impact by showing it earlier, making it clearer, and presenting it the same way across all Open Calls.
Use this outcome to define simple principles for how future mandatory rules should be introduced.
Work Done
- Audited all Open Calls to find inconsistent wording and missing context
- Gave users a short explanation of why a .ART domain or ID.ART profile is required
- Standardised how the rule is shown across Open Calls
- Created a framework for clear, consistent mandatory rules
- Added an early link to the promo page
Results
- Submission drop-off improved from 63% to 49% (−14 pp) after making the rule clear and visible earlier
- Support tickets related to the rule decreased by 31% during the following two weeks
- User confusion reports fell by 39% across feedback channels, based on Amplitude, Zendesk and survey data
Inconsistencies
Submission Flow Comparison
Timing vs Drop-off
Drop-off (%)
Impact Summary
-
pp
-14 pp
63%
49%
-31%
380
cap.
260
-39%
50%
31%
V. Cross-Product Promo Code Launch
Some details have been removed due to NDA.
This case shows how aligning promo code behaviour across products created an immediate revenue lift with minimal UX changes.
Problem
Registration conversion for .ART domains was 7.3%.
Google Analytics showed a 65% drop-off on GET.ART during domain entry.
Users did not understand how HUG and GET.ART were connected, and the integration delivered no visible value.
Solution
A simple cross-product mechanic was introduced:
a promo code claimed in HUG and applied on GET.ART.
The flow was designed mobile-first with clear steps and guidance.
A Figma prototype was aligned with backend and marketing.
Work Done
- Behaviour and drop-off analysis in Microsoft Clarity
- 20 user tests of the new flow
- Coordination with marketing + backend
- API requirements and implementation logic for promo codes
Results
- 45+ registrations in the first month
- ~3 new registrations per week
- Revenue reached USD 4,265 from 1,643 domains
- Conversion improved from 7.3% to 9.5%
- Drop-off reduced from 65% to 25%
Before (Old Flow)
Cross-Product Flow
After (New Flow)
Impact Summary
USD
9.5%
7.3%
65%
25%
The earlier HUG brand and visual foundation were created by the previous design team led by Creative Director Lindsey Gemmill. My contribution at .ART centred on UX updates, usability fixes, and iterative improvements during the post-acquisition phase.
Contact
rebotics@protonmail.com
© 2025 Taras Denisenko