From Disarray to Structure:
How IA & Content Strategy Saved €2.1M
From Disarray to Structure:
How IA & Content Strategy Saved €2.1M
From Disarray to Structure:
How IA & Content Strategy Saved €2.1M
A telecom provider invested heavily in self-service, yet support volumes and costs remained high. The root causes turned out to be completely fragmented information, a messy and illogical information structure, and a poorly tuned search function.
A telecom provider invested heavily in self-service, yet support volumes and costs remained high. The root causes turned out to be completely fragmented information, a messy and illogical information structure, and a poorly tuned search function.
A telecom provider invested heavily in self-service, yet support volumes and costs remained high. The root causes turned out to be completely fragmented information, a messy and illogical information structure, and a poorly tuned search function.
My contribution
My contribution
I transformed a fragmented knowledge ecosystem into a scalable self-service platform.
As Information Architect, Content Strategist and UI Designer, I:
I transformed a fragmented knowledge ecosystem into a scalable self-service platform.
As Information Architect, Content Strategist and UI Designer, I:
I transformed a fragmented knowledge ecosystem into a scalable self-service platform.
As Information Architect, Content Strategist and UI Designer, I:
restructured the information architecture,
established new guidelines for better content usability,
redesigned the user interface of the knowledge base and self-service portal,
improved overall usability,
set up a governance model.
restructured the information architecture,
established new guidelines for better content usability,
redesigned the user interface of the knowledge base and self-service portal,
improved overall usability,
set up a governance model.
restructured the information architecture,
established new guidelines for better content usability,
redesigned the user interface of the knowledge base and self-service portal,
improved overall usability,
set up a governance model.
This resulted in a better user experience, higher findability, greater customer satisfaction with less dependence on the Support departments, and a more efficient support organisation.
This resulted in a better user experience, higher findability, greater customer satisfaction with less dependence on the Support departments, and a more efficient support organisation.
This resulted in a better user experience, higher findability, greater customer satisfaction with less dependence on the Support departments, and a more efficient support organisation.
Results achieved
Results achieved
business impact
business impact
business impact
Measurable results for the organisation.
Measurable results for the organisation.
Measurable results for the organisation.
Training new support staff
50
50
%
%
less time
needed
Support emails per week
300
300
emails
instead of 3000
FTE Support teams
6
FTE
instead of 16
FTE Support teams
6
6
%
FTE
instead of 16
Avergae call duration
Average call duration
60
60
seconds
shorter
Issues resolved first time
increase
up from 57%
72
72
%
%
Annual Support Costs
€
€
2,1
2,1
M
M
reduction
Average call duration
60
Seconds
shorter
Issues resolved first time
72
%
increase
up from 57%
Annual Support Costs
€
2,1
M
less
operational impact
operational impact
Improvements in processes, content and information architecture.
Improvements in processes, content and information architecture.
Productivity Support Teams
customers helped better
in less time
Information Structure
increased consistency
through the introduction of templates
Findability
finetuned search engine
makes information easy to find
Findability
finetuned search engine
makes information easy to find
Content Quality
UP-TO-DATE CONTENT
through automated periodic reminders
Content Management
INCREASED EFFICIENCY
through new processes
Escalation routing
FEWER ERRORS
by linking the right forms to the right content
Content Quality
UP-TO-DATE CONTENT
through automated periodic reminders
Content Management
INCREASED EFFICIENCY
through new processes
Escalation routing
FEWER ERRORS
by linking the right forms to the right content
CUSTOMER impact
CUSTOMER impact
The benefits end users experienced directly.
The benefits end users experienced directly.
Customer satisfaction
70
%
Net promoter score
(up from 50%)
Self-service
90
%
FOUND ANSWER
independently in the knowledge base
Phone support query
26
%
FASTER & Better
handled
Customer satisfaction
Net promoter score
up from 50%
70
%
Self-service
found answer
independently in the knowledge base
90
%
Phone support query
Net promoter score
up from 50%
26
%
Project overview
Project overview
My role
My role
Information Architect • Content Strategist • UI Designer
Information Architect • Content Strategist • UI Designer
Duration
Duration
6 months
6 months
team
team
6 Content specialists
6 Content specialists
2 stakeholders from the support centres
2 stakeholders from the support centres
1 Trainer
1 Trainer
SCOPE
SCOPE
Knowledge base
Script for Support staff
Content governance and operational processes
Knowledge base
Script for Support staff
Content governance and operational processes
Operational pain points
Operational pain points
Knowledge base / Information
Knowledge base / Information
Information in the knowledge base was hard to find; the search function did not help users;
Customers often contacted (by phone or email) Support for information even though that information was already available in the knowledge base;
Customers often had to contact Support multiple times before their issue was resolved;
Support teams also struggled to find the required information.
Information in the knowledge base was hard to find; the search function did not help users;
Customers often contacted (by phone or email) Support for information even though that information was already available in the knowledge base;
Customers often had to contact Support multiple times before their issue was resolved;
Support teams also struggled to find the required information.
Costs Support
Costs Support
Cost (average) per support call: €25 (more than thousand calls each day).
Cost (average) per email: €7 (on average 3000 emails per week).
FTE Costs Support staff:
Cost (average) per support call: €25 (more than thousand calls each day).
Cost (average) per email: €7 (on average 3000 emails per week).
Costs FTE Support staff:
Cost (average) per support call: €25 (more than thousand calls each day).
Cost (average) per email: €7 (on average 3000 emails per week).
Costs FTE Support staff:
-
-
-
-
-
-
-
-
Call centre staff:
Support staff email:
Complaints department:
Trainers:
Call centre staff:
Support staff email:
Complaints department:
Trainers:
206
9
7
6
206
9
7
6
Total:
Total:
228
228
Training / Onboarding
Training / Onboarding
Onboarding period for new support staff: 2 weeks.
High Support staff turnover (some left right after onboarding)
Onboarding period for new support staff;
High Support staff turnover
ORGANISATIONAL CONSEQUENCES
ORGANISATIONAL CONSEQUENCES
Rising operational costs;
Inconsistent customer experience resulting in a low NPS;
Declining trust in the knowledge base;
Limited scalability of the support organisation.
Rising operational costs;
Inconsistent customer experience resulting in a low NPS;
Declining trust in the knowledge base;
Limited scalability of the support organisation.
Research
Research
Before determining a strategy, I conducted thorough diagnostics:
Before determining a strategy, I conducted thorough diagnostics:
Research ACTIVITIES
Research ACTIVITIES
18 in-depth interviews with support staff
7 customer interviews about their search behaviour
Analysis of 3 months of search logs (which search terms worked, which didn’t?)
Audit of all knowledge articles - more than 1,750 (content analysis)
Full day in usability lab with 7 participants
Card sorting sessions with 14 participants
18 in-depth interviews with support staff
7 customer interviews about their search behaviour
Analysis of 3 months of search logs (which search terms worked, which didn’t?)
Audit of all knowledge articles - more than 1,750 (content analysis)
Full day in usability lab with 7 participants
Card sorting sessions with 14 participants
Root cause analysis
Root cause analysis
The problem was not a lack of content, but a fundamentally failing information architecture that broke down on 3 levels:
The problem was not a lack of content, but a fundamentally failing information architecture that broke down on 3 levels:
01.
Fragmented and Inconsistent Content
Fragmented and Inconsistent Content
02.
Chaotic Information Architecture
Chaotic Information Architecture
03.
Search didn’t work — partly due to a language gap
Search didn’t work — partly due to a language gap
root cause analysis
root cause analysis
01.
01.
Fragmented and Inconsistent Content
Fragmented and Inconsistent Content
Content had grown uncontrolled over the years, with little to no governance.
Content had grown uncontrolled over the years, with little to no governance.
Over 1,750 articles, but 14% were duplicates or outdated.
No fixed templates — every article looked different.
No ownership — nobody was responsible for checking if content was still up-to-date.
Different writing styles per author (some short, others lengthy)
Authors were not trained in writing for the Web
Over 1,750 articles, but 14% were duplicates or outdated.
No fixed templates — every article looked different.
No ownership — nobody was responsible for checking if content was still up-to-date.
Different writing styles per author (some short, others lengthy)
Authors were not trained in writing for the Web
Result: Support staff gave contradictory answers. Customers were confused. Trust in the knowledge base declined.
Result: Support staff gave contradictory answers. Customers were confused. Trust in the knowledge base declined.
root cause analysis
root cause analysis
02.
02.
Chaotic Information Architecture
Chaotic Information Architecture
The categories, sub-categories and sub-sub-categories (817 in total!) reflected the internal company structure, not how users thought.
Apart from that, the sheer number of categories and sub (and sub-sub) categories made navigating through them next to impossible for people who every now and then needed a quick answer.
It was like finding your way in a labyrinth designed to make you get lost.
And to make matters even worse, even authors no longer understood their own structure. They got lost themselves and then just 'invented' yet another new category.
The categories, sub-categories and sub-sub-categories (817 in total!) reflected the internal company structure, not how users thought.
Apart from that, the sheer number of categories and sub (and sub-sub) categories made navigating through them next to impossible for people who every now and then needed a quick answer.
It was like finding your way in a labyrinth designed to make you get lost.
And to make matters even worse, even authors no longer understood their own structure. They got lost themselves and then just 'invented' yet another new category.
Insight
Insight
I ran card sorting sessions with 5 customers and 9 support staff.
I ran card sorting sessions with 5 customers and 9 support staff.
Customers and staff thought in task-oriented categories:
Customers and staff thought in task-oriented categories:
• “Manage my account”
• “Manage my account”
• “Calling abroad”
• “Calling abroad”
• “Setting up voicemail”
• “Setting up voicemail”
But the knowledge base was organised according to internal departments:
But the knowledge base was organised according to internal departments:
• “Finance”
• “Finance”
• “International”
• “International
• “Services”
• “Services”
This mismatch was fatal: either no results or totally irrelevant results.
When browsing for relevant information users quickly got lost in the navigation and gave up and picked up their phones or sent an email.
This mismatch was fatal: either no results or totally irrelevant results.
When browsing for relevant information users quickly got lost in the navigation and gave up and picked up their phones or sent an email.
root cause analysis
root cause analysis
03.
03.
Search failed to deliver relevant results work
Search failed to deliver relevant results work
For 3 months I analysed search logs and compared search terms with the terms used in articles and metadata. I evaluated how the search engine responded to user queries and compared the results with the expected outcomes to improve search relevance.
For 3 months I analysed search logs and compared search terms with the terms used in articles and metadata. I evaluated how the search engine responded to user queries and compared the results with the expected outcomes to improve search relevance.
Search success rate: 38% of searches led to usable results. Mostly there way too many results, and those results were often not relevant at all for the search query.
Search success rate: 38% of searches led to usable results. Mostly there way too many results, and those results were often not relevant at all for the search query.
Critical findings:
Customers and the knowledge base spoke different languages!
Example:
Customers and the content management team spoke different languages!
Example:
Customers and the knowledge base spoke different languages!
Example:
• Customer searched: “account” → Knowledge base: “statement” → 0 results!
• Customer searched: “account” → Knowledge base: “statement” → 0 results!
• Customer searched: “top up” → Knowledge base: “add credit” → 0 results!
• Customer searched: “top up” → Knowledge base: “add credit” → 0 results!
The content team didn’t speak the customer’s language. They wrote in corporate jargon.
The content team didn’t speak the customer’s language. They wrote in corporate jargon.
The search engine was poorly configured resulting in irrelevant search results!
Examples:
Somebody scoping the search to only content dealing with a specific abonnement still got results related to prepay etc.A word mentioned only once in passing in the article carried the same weight as a match in the article title. This was yet another reason completely irrelevant search results kept surfacing.
The search engine was poorly configured resulting in irrelevant search results!
Examples:
Somebody scoping the search to only content dealing with a specific abonnement still got results related to prepay etc.
A word mentioned only once in passing in the article carried the same weight as a match in the article title. This was yet another reason completely irrelevant search results kept surfacing.
The search engine was poorly configured resulting in irrelevant search results!
Examples:
Somebody scoping the search to only content dealing with a specific abonnement still got results related to prepay etc.
A word mentioned only once in passing in the article carried the same weight as a match in the article title. This was yet another reason completely irrelevant search results kept surfacing.
Transformation Strategy
Transformation Strategy
Instead of isolated fixes, I designed a complete transformation of the service ecosystem built on four pillars:
Instead of isolated fixes, I designed a complete transformation of the service ecosystem built on four pillars:
01.
01.
Content Rationalisation
Content Rationalisation
Not: “add more content” Instead: “better curated, consistent content”
Not: “add more content” Instead: “better curated, consistent content”
02.
02.
Information Architecture Redesign
Information Architecture Redesign
Restructuring the taxonomy around how users think
Restructuring the taxonomy around how users think
03.
03.
Search algorithm improvement
Search algorithm improvement
Finetuning the search engine
Finetuning the search engine
04.
04.
Governance for Sustainability
Governance for Sustainability
Making the improvements stick long-term
Making the improvements stick long-term
1.
Content Rationalisation
Not: “add more content” Instead: “better curated, consistent content”
2.
Information Architecture Redesign
Restructuring the taxonomy around how users think
3.
Search algorithm improvement
Finetuning the search engine
4.
Governance for Sustainability
Making the improvements stick long-term
transformation Strategy
transformation Strategy
01.
01.
Content
Content
Rationalisation
Rationalisation
Step 1: Full content audit
Step 1: Full content audit
I analysed all articles (more than 1750) for:
I analysed all articles (more than 1750) for:
Duplicates (remove )
Outdated content (archive, don’t delete),
Related articles (merge to reduce amount of strongly related search results).
Conflicting information (reconcile)
Duplicates (remove )
Outdated content (archive, don’t delete),
Related articles (merge to reduce amount of strongly related search results).
Conflicting information (reconcile)
Result: 1,750 → 500 articles. Sharper selection, better quality.
Result: 1,750 → 500 articles. Sharper selection, better quality.
Step 2: Template & writing guidelines
Step 2: Template & writing guidelines
I designed a rigorous template for every article:
I designed a rigorous template for every article:
Tab “What is it?” — short definition
Tab “How does it work?” — steps (bullets, not paragraphs)
Tab “What if it doesn’t work?” — troubleshooting
Work instructions visible only to support teams including the right escalation form when needed
Tab “What is it?” — short definition
Tab “How does it work?” — steps (bullets, not paragraphs)
Tab “What if it doesn’t work?” — troubleshooting
Work instructions visible only to support teams including the right escalation form when needed
All articles were fully rewritten according to this structure.
All articles were fully rewritten according to this structure.
Why this structure? Usability research showed that scannable, stacked information works better than continuous prose. Stacked information is ideal for scanning content quickly. And users want a direct answer, not a story.
Why this structure? Usability research showed that scannable, stacked information works better than continuous prose. Stacked information is ideal for scanning content quickly. And users want a direct answer, not a story.
I had to convince the content team that rules were better than freedom. There was some resistance from some of the content managers.
I had to convince the content team that rules were better than freedom. There was some resistance from some of the content managers.
Step 3: Introducing basic ownership
Step 3: Introducing basic ownership
This was the hardest step. Without ownership, quality declines immediately. As a first foundation for the full governance model developed under Pillar 4, I implemented:
This was the hardest step. Without ownership, quality declines immediately. As a first foundation for the full governance model developed under Pillar 4, I implemented:
Peer-review process — new articles must be approved by a colleague,
Periodic audits — Content owners received periodic notifications to check for stale content,
Work instructions visible only to support teams including the right escalation form when needed,
Appointing content owners — per article or topic
Peer-review process — new articles must be approved by a colleague,
Periodic audits — Content owners received periodic notifications to check for stale content,
Work instructions visible only to support teams including the right escalation form when needed.
Appointing content owners — per article or topic
This required organisational change.
This required organisational change.
All articles were fully rewritten according to this structure.
All articles were fully rewritten according to this structure.
transformation Strategy
transformation Strategy
02.
02.
Information
Redesign
Architecture Redesign
Information Architecture
This was the core of my work.
This was the core of my work.
Step 1: From Diagnosis to Design
Step 1: From Diagnosis to Design
This phase built directly on the card-sorting exercise already described in the Root Cause Analysis (5 customers and 9 support staff, grouping 60 topics into clusters).
Rather than running a second round of research, I performed a treejack test to evaluate the category labels that had come about during the card sorting sessions. When treejack test confirmed the results from the card sorting sessions I used those clusters as the foundation for the new architecture.
Key findings guiding the redesign:
This phase built directly on the card-sorting exercise already described in the Root Cause Analysis (5 customers and 9 support staff, grouping 60 topics into clusters).
Rather than running a second round of research, I performed a treejack test to evaluate the category labels that had come about during the card sorting sessions. When treejack test confirmed the results from the card sorting sessions I used those clusters as the foundation for the new architecture.
Key findings guiding the redesign:
task orientation instead of a structure mirroring the internal departmentstext here,
3 navigation levels maximum
task orientation instead of a structure mirroring the internal departmentstext here,
3 navigation levels maximum
Result: 1,750 → 500 articles. Sharper selection, better quality.
Result: 1,750 → 500 articles. Sharper selection, better quality.
Step 2: New taxonomy
Step 2: New taxonomy
Based on card sorting findings I designed a new architecture
Based on card sorting findings I designed a new architecture
Result: 817 categories and (sub)sub-categories → 130 categories. — an 84% reduction in structural complexity.
Result: 817 categories and (sub)sub-categories → 130 categories. — an 84% reduction in structural complexity.
Step 3: Maximum 3 navigation levels
Step 3: Maximum 3 navigation levels
Critical design decision: no one should need more than 3 clicks.
I tested this in 7 usability sessions. Participants who had to navigate deeper than 3 levels gave up and said they would rather call customer service.
Critical design decision: no one should need more than 3 clicks.
I tested this in 7 usability sessions. Participants who had to navigate deeper than 3 levels gave up and said they would rather call customer service.
Now I would pick up my phone and call Support"
Now I would pick up my phone and call Support"
transformation Strategy
transformation Strategy
03.
03.
Search
Optimisation
Functionality Optimisation
Search Functionality
This is where the language gap was solved, the weighing was improved upon, a list of synonyms was established
This is where the language gap was solved, the weighing was improved upon, a list of synonyms was established
Step 1: Building a synonym list
Step 1: Building a synonym list
Based on search log analysis, I created an extensive synonym list:
“Account” → linked to articles about “statement” and “invoice” “Top up” → linked to “add credit” “Unreachable” → linked to “call forwarding” … (100+ synonyms)
The search engine now searches not only for exact matches, but also for synonyms.
Based on search log analysis, I created an extensive synonym list:
“Account” → linked to articles about “statement” and “invoice” “Top up” → linked to “add credit” “Unreachable” → linked to “call forwarding” … (100+ synonyms)
The search engine now searches not only for exact matches, but also for synonyms.
Based on search log analysis, I created an extensive synonym list:
“Account” → linked to articles about “statement” and “invoice” “Top up” → linked to “add credit” “Unreachable” → linked to “call forwarding” … (100+ synonyms)
The search engine now searches not only for exact matches, but also for synonyms.
Step 2: Optimising Search Relevance
Step 2: Optimising Search Relevance
Beyond fixing the vocabulary gap, I fine-tuned how the search engine ranked results on the CMS search. Matches in the title and metadata were weighted more heavily than matches in body text, since they're a stronger relevance signal, with query term frequency in the article as a secondary factor.
I also closed the customer-editor vocabulary gap by building a custom synonym thesaurus, sourced from real search-term analytics — so searches like "rekening" now correctly surfaced articles written around "factuur," instead of returning zero results.
Beyond fixing the vocabulary gap, I fine-tuned how the search engine ranked results on the CMS search. Matches in the title and metadata were weighted more heavily than matches in body text, since they're a stronger relevance signal, with query term frequency in the article as a secondary factor.
I also closed the customer-editor vocabulary gap by building a custom synonym thesaurus, sourced from real search-term analytics — so searches like "rekening" now correctly surfaced articles written around "factuur," instead of returning zero results.
Step 3: Continuous optimisation
Step 3: Continuous optimisation
Every week:
Every week:
Analyse the previous week’s search logs,
Identify searches with 0 results,
Add missing synonyms,
Measure improvement the following week.
Analyse the previous week’s search logs,
Identify searches with 0 results,
Add missing synonyms,
Measure improvement the following week.
transformation Strategy
transformation Strategy
04.
04.
Governance
Sustainability
for Sustainability
Governance for
THE DILEMMA
THE DILEMMA
A one-time clean-up is easy. Keeping things clean sustainably is much harder. I observed that the organisation cleaned up content. But after 6 months, everything fell apart again. I wanted to prevent this.
A one-time clean-up is easy. Keeping things clean sustainably is much harder. I observed that the organisation cleaned up content. But after 6 months, everything fell apart again. I wanted to prevent this.
My Governance Model:
My Governance Model:
1. Content ownership,
2. Peer review,
3. Periodic audits,
4. Feedbaclk loop from Support teams
1. Content ownership,
2. Peer review,
3. Periodic audits,
4. Feedbaclk loop from Support teams
Content OWNERSHIP
Content OWNERSHIP
Every article / topic gets a clear owner,
The owner is responsible for accuracy, quality and being up-to-date,
15% of their time allocated to content management.
Every article / topic gets a clear owner,
The owner is responsible for accuracy, quality and being up-to-date,
15% of their time allocated to content management.
PEER REVIEW
PEER REVIEW
New or updated articles must be approved by a peer,
Review checklist: “matches the template?”, “written in the customer’s language?”, “no duplicates?”, "up to date?", "Used the right metadata?"
Right category/sub-category?
New or updated articles must be approved by a peer,
Review checklist: “matches the template?”, “written in the customer’s language?”, “no duplicates?”, "up to date?", "Used the right metadata?"
Right category/sub-category?
PERiodic Audits
PERiodic Audits
Each article got an expiry date,
Expiry date flagged outdated articles,
When audit is due the content owner receives an automated notification,
If audit is not performed within a certain period the article would be invisible until audit.
Each article got an expiry date,
Expiry date flagged outdated articles,
When audit is due the content owner receives an automated notification,
If audit is not performed within a certain period the article would be invisible until audit.
Feedback Loop from Support
Feedback Loop from Support
A weekly meeting with stakeholders was set up to discuss which topics generated the most questions,
This feeds directly back into content priorities,
A weekly meeting with stakeholders was set up to discuss which topics generated the most questions,
This feeds directly back into content priorities,
Interim Solution: Internal Support Platform
Interim Solution: Internal Support Platform
the challenge
the challenge
While I worked on the long-term transformation, support staff still had to help customers every day. The old knowledge base was still no better.
Support staff wasted an average of 1-2 minutes per call searching for relevant information. With 200 staff, 8 hours a day, 5 days a week, that’s more than 3,000 lost hours per week.
I designed an internal support platform streamlined around the actual workflowas a quick win:
This resulted in a better user experience, higher findability, greater customer satisfaction with less dependence on the service desk, and a more efficient support organisation.
While I worked on the long-term transformation, support staff still had to help customers every day. The old knowledge base was still no better.
Support staff wasted an average of 1-2 minutes per call searching for relevant information. With 200 staff, 8 hours a day, 5 days a week, that’s more than 3,000 lost hours per week.
I designed an internal support platform streamlined around the actual workflowas a quick win:
This resulted in a better user experience, higher findability, greater customer satisfaction with less dependence on the service desk, and a more efficient support organisation.
The main products: abonnements, prepay, white label each visualized as a tab. text here,
Per tab a list of all topics ordered alphabetically linking to articles in new structure exclusive for that product.
A bulletin board for information on local outages, etc. (pulled from a separate platform).
A section dedicated to browsing the main topics linking to the most requested information.
A quick links section dedicated to the top 5 of most frequent asked questions for that specific product.
This resulted in a better user experience, higher findability, greater customer satisfaction with less dependence on the service desk, and a more efficient support organisation.
This resulted in a better user experience, higher findability, greater customer satisfaction with less dependence on the service desk, and a more efficient support organisation.
DESIGN CHOICES
DESIGN CHOICES
Why these features? Because I studied the support team.
I spent more than 20 sessions in the call centre, listened to real conversations, and recorded where staff wasted time:
Why these features? Because I studied the support team.
I spent more than 20 sessions in the call centre, listened to real conversations, and recorded where staff wasted time:
Searching the knowledge base (1-2 min) or walking to a colleague of who they thought would know the answer (all the while having the client waiting on the phone).
Searching another application for the right escalation form (30 sec).
Copy-pasting info into info notes (30 seconds to a minute) into yet another application.
Frequently checking another application for up-to-date information concerning outages.
This platform addressed all of these pain points directly.
This platform addressed all of these pain points directly.
Platform results:
Platform results:
Search time: 1-2 minutes → 10 seconds.
Average call duration: at least 60 seconds less.
Estimated cost savings: €5 per call.
With 200 staff: € 60 k per day.
strategic advantage
strategic advantage
This platform did more than save costs. It:
This platform did more than save costs. It:
Let support staff directly experience that better design makes their job easier.text here,
Reduced resistance to the larger transformation (“look, it actually works!”).
Generated valuable feedback for the eventual self-service platform.
