In this initiative, we tried to solve a million-dollar industry problem on skill discovery. Our goal was to help our users effortlessly discover and learn our new skills.
Even though the overall usage of Ask Sam is high, the usage across skills is inconsistent. High-performing skills contribute only around 30%, leaving behind 70% of low-performing skills.
How to improve adoption of low-performing skills
We, along with the product, investigated various reasons behind low-skill adoption. “Lack of awareness” emerged as a critical pain point that could be a potential opportunity if solved.
To evaluate the same and build upon it, we carefully planned and executed a discovery phase. The goal was to collaborate and ideate together as an experience team.
The team brainstorming exercise helped us to generate more user pain points and solution directions. Team involved: Produt manager | Design | Engineering | Bussiness
We conducted stakeholder interviews with the operations team. We needed to understand how we handled this problem at the operations level.
“Apart from the generic training, there is no specific training given to the associate on using applications” - Club operations
Knowing that discoverability is a universal problem among conversational assistants, it was essential to evaluate what measures have been taken across the industry.
Most of the assistants have dedicated skill documentation for users to explore and learn.
Featured skill cards, skills suggestions, by the way, etc. were some new initiatives
We also had a similar explore skills page and “what’s new” feature to communicate new and available skills. But the usage was minimal.
As a next step, we did multiple club visits, in-person interviews, and observation studies to know our users better.
Our research goal was:
“Our users did’nt know
that certain skills could exit ”
They had a very strong perception about the limited capabilities of Ask Sam
While we designed Ask Sam to perform various capabilities and features, some of our users continued to perceive Ask Sam as a tool to perform only certain key skills like product look-up, schedule, floorplan lookup etc. Neither did they knew about other skills, nor did they expected Ask Sam to perform other things.
The product team's conceptual model of Ask Sam and users mental model of Ask Sam turned out to be different. The expectation is for them to match to have the best experience.
Product team conceptual model. VS
User's mental model
Based on the information collected on the discovery phase we decided to go with the following approach.
We need to push the right information upfront to the user
rather than waiting for the user to pull them
Our goal was to make sure the information is easily discoverable for the user. Towards that goal, we must place them at the right touch points in the user journey.
Cross application promotion
Bring users immediate attention
User might disregard if busy
Dependency on other platforms and teams
Both the options were effective but required close collaboration and integration with the other apps and service teams. We decided to narrow down the scope to our application and identify touch points within that.
Chat screen within Ask Sam was the page where our users land first and stay actively to ask questions. So, we decided to meet where our users are, Chat screen. Within the options considered among various components, we decided on using a chat component itself, as its more organic, and scalable.
Our goal was to identify what part of conversation should we push the skill information.
Before they start a conversation
Greeting were always ignored by the user, as it was a static content.
If written well, along with the right micro interactions, greeting can engage user before they start asking questions.
Follow up after a conversation
Following up with a skill suggestion, after a conversation can also be a great opportunity for engagement.
Generic follow up
Follow up suggesting a skill which they have not used yet.
Contextual follow up
Follow up suggesting a skill which are related to the previous conversation.
To keep them actively engaged, we must always provide value in the information we show. So, our goal was to make the skill suggestion relevant to the user group. To maintain relevancy, we created a skill mapping with respect to user groups. To reduce redundancy, we integrated with real time skill usage so that user will only see relevant skills that are not used.
Information should be:
Based on teh above crieteria, we finalized on 4 main information components.
To finalise on the design components, we conducted some prototype testing with the club users. This helped us to understand the following.
Which design was:
Tested with variations in card behavious
One we identified all the design components, we worked on creating individual assets for each skills card. That involved designing the content and the iconography.