Eldar designs
Case study / SAAS / 2023 / Research / Prototyping

Reimagining Helpdesk.media’s Knowledge Base for support agents



Helpdesk Media Foundation, a European non-profit, serves as a lifeline for victims of the war in Ukraine through an emergency hotline. The foundation supports 40 hotline managers and over 3,000 volunteers who assisted more than 35,000 people in 2022–23. Central to their operations is a secure conversational support system, similar to Zendesk or Intercom. However, the outdated knowledge base had become a bottleneck, hindering the efficiency of the support agents.

Foundation's website (RU+EN)
CEO’s article in the New York Times (EN)
About the Foundation’s helpline service in the Nieman Lab (EN)

P.S. All the designs here were translated from Russian to English for a better comprehension

Objective

My responsibility as a product designer was to overhaul the existing Knowledge base to ensure the features were up-to-date, easily accessible, and efficient, thereby improving response times and decision-making during crises. The first version of the Knowledge base was hastily designed for a rapid launch of the project in the beginning and for a very long time it has not been significantly updated. After some time it didn’t meet new requirements and couldn’t face new challenges. Some users created their own loopholes to improve their experience.
The old version of the Knowledge base integrated into the Conversational Support System
Version of the Knowledge Base for contributors where they create, edit and delete articles was used by some support agents due to the better search


Research

User interviews from regular users and stakeholders revealed significant dissatisfaction with the search and navigation functionalities of the knowledge base. Support managers found the system cumbersome, often resorting to external tools to manage frequently used responses.

‘The article search is so broken. I think, it is beyond repair. You won’t be able to fix it.’ – a support manager once said

Other insights from the interviews:

‘It’s very difficult to search for something by topic. We have a large knowledge base but there is no convenient “catalogue” inside the chats’ – Support manager ‘Volunteers can’t see the table of contents, hiding it is really bad’ – Support manager
‘There are messages that I send to every chat every day, it's inconvenient to search for them all the time so I had to save them to a text file on my computer’ – Volunteer‘I search for the same templates over and over again and even store several of them in my Notes app’ – Support agent


Defining problems

Problems discovered during research and analysis fell into three basic groups:
  • Inefficient Search Mechanism: Users were unable to locate relevant information quickly, impacting response times.
  • Poor Navigation: The lack of structured information made the system difficult to learn and use.
  • Reusing articles: Users don’t have quick access to reusable materials, impacting response times.
In the meantime, giving the limited resouces we had, we decided to measure the usage of the current knowledge base so that when we implement a new version, we would be able measure the outcome not just verbally but also statistically. The two basic metrics we decided to implement were the amount of clicks on an article to embed it into a chat and the amount of clicks on an article in the knowledge base editor.



Ideation and design

Together with my Design Lead we brainstormed solutions prioritized by the RICE model. We decided to revamp search engine and bring in some structure to the articles to enhance user experience. Favourites is a rather large and complicated feature yet we wanted to launch something super small to test the idea and bring joy to our users as well.

List of problems and potential solutions scored by RICE model


Some aspects were crucial:
  • Do we preserve the current version of the knowledge base as part of the chats or remove it and work on the open version of the one that lives in another directory and at that time functioned as an editor for knowledge base contributers?
  • There are several ‘types’ of articles in the knowledge base, the most important two are Instructions that agents use as an answer to a question in chats and Templates that agents use as intro, defining and closing statements, modifying for their usage — should we keep them in one place or make them separate and put them into different places?
Our stakeholder wanted us to make an open version on another page whilst I was convinced that having an integrated knowledge base in the chats is a better option because the users would find answers/templates in it and embed them into chats faster than if they would look for articles on a different page and copy and paste information into a chat from there. Leading conversational support systems on the market have an integrated knowledge base in chats as well. Moreover, the stakeholder had reservations about the original version due to its age and poor performance. However, she trusted my opinion and agreed with an integrated knowledge base.

In the research users also mentioned that instructions and templates do look similar because of the previous design decisions but they don’t mind having both types of articles in one place. In fact, they were happy it was that way because they search for an instruction the same way they search for a template and having them placed in two diferrent places wouldn’t benefit them. Thus, we I decided to keep them together.

Favourite articles could be accessed from the home screen of the Knowledge base
Quick shortcuts from the home screen could lead to articles grouped by type for an easier navigation

Var 1


Var 2


Development

We redesigned the knowledge base to include:
  • Article hierarchy: Articles were grouped by categories, helping nagivating and exploring the knowledge base.
  • Enhanced search and filters: Users could now search by article type or keywords, improving the discoverability of information.
  • Favourites could be easily accessed from the home directory to speed up the response time.


Validation and Testing


Due to limited resources, full-scale usability testing was bypassed. However, feedback from prototypes tested with select users and stakeholders was overwhelmingly positive, confirming we were on the right track. Once coded, I tested the knew knowledge base for bugs on our development stage and it was ready to be launched.
Article types and favourites reachable in one single tap
Flexibility thanks to new filters and faster article search
Articles are readable with easy to reach actions


Launch and Results

The new knowledge base was rolled out with a tutorial to aid in transition. Early metrics indicated a significant increase in usage and a decrease in reliance on the external editor system.

The redesigned knowledge base improved the operational efficiency of Helpdesk Media Foundation, directly enhancing the support provided to war victims. This project not only met its objectives but also demonstrated the power of user-centered design in critical service environments.
Adoptionof the new knoledge base was quick Response timeof support agents was improved Fixes and featuresSearch mechanism was fixed, new filters and favourites added Discovery
and learning of articles in the knoledge base became easier

‘This is all we ever wanted’
– a happy user said in the end


Special thanks

♥︎ To our stakeholder for believing in our ideas
♥︎ To our software developers who were eager to shovel a lot of backend code
♥︎ To our users for helping make all the changes possible