Yammer Question & Answer Smart Detect

 
process.gif

Design Sprint

To kick off the project, the project PM and I cohosted a design sprint to go broad in blue sky ideas and then narrow down on use cases.

We categorized the user journey into three personas: Asker, Answerer, and Community. We sketched ideas based off of steps in the user journey.

Problem

Posing questions or requests is a common use case in communities—especially in “knowledge,” “practice” or “helpdesk”-based ones. However, this process is quite janky, and customers lament the lack of knowledge management tools in Office 365. Specifically:

1. Knowledge is unstructured and hard to scan or parse for people. As one scrolls through a group’s feed, it is difficult to know which conversations constitute questions, which questions are resolved, which ones need help, and so on. 2. It is cumbersome to find conclusions in large threads. Threads fold up replies as soon as there are more than

2. One must scroll through the entire thread to find the “best,”, “correct” or “most useful” response. This increases time and effort for people to find the information they need—slowing learning and decision-making. Active communities with longer conversations—especially during YamJam-like events—feel the brunt of this.

3. Current conversation model limits our ability to address these issues elegantly. For example, replies are ordered chronologically, and this limits our ability to reorder answers based on better heuristics e.g. marked as “most helpful” by poster, has highest number of votes, etc.

Team

Project Team included 1 Designer, 1 Researcher, 1 Front End Engineer, 2 Analysts, 1 PM, 1 Copy Writer. My role was the Lead Designer

Goal

For people seeking knowledge in communities, create an elegant experience to seek knowledge in a structured way, indicate closure, and learn from knowledge created by others. For people willing to share knowledge, help easy identification of conversations seeking help, and meaningful responses to them. This project will lay a foundation to create and scan knowledge in communities. Over time, we feel this will lead to improved knowledge creation, consumption and retrieval.

Research

Customers like EY, Unilever, Adidas, Velux, Lego and Thermo Fischer Scientific all have knowledge or helpdesk communities where people ask questions about specific topics. The designated community managers and, very often, active members of the community answer them. However, this process is far from optimal on Yammer Implication—We could build on user behaviors that exist already in organizations, we can make the Q&A loop faster, more efficient, and elegant from an experience perspective.

Some, like EY and Sanofi take the knowledge from the Yammer groups, sanitize it and put it in neat little documents on SharePoint sites. This sanitized knowledge is then publicized across the organization. Implication—providing some structure around questions and answers could make their life easier by allowing knowledge to be better structured and, consequently, found.

Customers like Centrica, Telefonica, Target and Vallourec have indicated the need to solve the problem of quickly finding the resolution in long threads.

 

Wireframes on Varying Solutions

3.1 Dismiss Pin Copy 4.png

What is the best route for the user to ask a question?

Taking ideas from the design sprint, I explored multiple variations and solutions. In the exploration phase, we considered questions as an attachment. This allowed for flexibility in adding an attachment to any Post Type.

3.1 Dismiss Pin Copy 8.png

The “Question attachment” would also allow for scanability. After qualitative user research, we concluded that this was disjointed from our current post type paradigm and users were confused by this.

B post result.png

Question as a Post Type followed our current post type paradigm and users understood how it worked.

However, for a long term solution we were looking for something truly seamless where it didn’t require users to take an extra step.

3.1 Dismiss Pin Copy 17.png

Explorations on the Visual Design for Best Answer 

It was  important the  Best Answer feel “celebratory” and almost badge like to provide positive feedback to whoever answered.  The solution must also provide accessibility  for those of poor eyesight, so low contrast solutions were eliminated.

Screen Shot 2018-11-01 at 2.28.25 PM.png

Mobile Wires

Screen Shot 2018-11-01 at 2.29.56 PM.png

Mobile Visual Explorations

Home Copy 6.png

As users type, related conversations in the group that might answer their question are surfaced. If users do not find their needed answer in Related Threads the question will be posted.

Home Copy 9.png

From there, admins and OPs can mark replies as Best Answer. This will allow the network to gather data on experts that actively answer questions, and route future questions to these expert type people.

Prototyping & User Testing

We tested the V1 prototype with 6 users for one hour each. They were current Yammer users who worked for corporations like Schneider, Habanero, and UK Post.

The uncertainties we were looking to test surrounded desirability, discoverability, and usability. We wanted to see if users found the feature useful enough to go through the trouble, if they could find the feature, and if they understood the concept of Best Answer.

Over all, we found that 5/6 users found the feature easy to find and all of them understood the process of selecting a Best Answer.

 
 Presenting at Microsoft Yammer all-hands

Presenting at Microsoft Yammer all-hands

Measurements of Success

Given that this is a major strategic direction that we, as an organization, want to invest in, we will seek to create an initial solution to the Q&A scenario and rapidly iterate from there. We will exercise qualitative methods such as usability testing and beta groups to reduce risk prior to shipping.

At the time of launch (Feb 2018), the feature will be available on our web, desktop, iOS and Android platforms.