I was recruiting for the summer internship, and during one of the interviews, I was asked to design a product for LinkedIn. The exact problem statement goes like this:
“ Imagine that you are a product manager at LinkedIn, and LinkedIn has identified an opportunity to provide online learning services. You have been asked to develop a recommendation for the product and pitch to management.”
Often, as a product manager, we work on a small feature of a product, and that’s why working on a product idea right from research to design was exciting for me.
Working on this assignment, I felt it would be a good idea to share publicly once I am done and get some feedback.
Here is my approach to the problem and final deliverable.
Disclaimer: Please note that I don’t claim to be expert of any sort in product designing, and I am sharing this to learn and grow. Overall, I had limited time to work on this assignment, and hence I made multiple sweeping assumptions in designing this product.
Problem Diagnosis
The first step in designing the product was to understand the problem and establish the constraints and assumptions in designing the solution.
Identifying Target Customer Segment
First, I identified different target segments in both business and consumer segment. It’s important to note that currently, a big source of income for learning portal are businesses.
Next, in order to identify segment characteristic, I did telephonic conversation, face-to-face discussion, and survey.
After careful analysis, I decided to focus on 20–35 years age-group based on LinkedIn’s strength, willingness-to-invest in learning, and satisfaction level with the current solutions. This age-group, which is a big segment for LinkedIn, has willingness-to-invest in learning, and is unsatisfied with the current solution, would be an ideal target segment.
However, ‘mid and large business’ is also a lucrative segment from the revenue perspective. Therefore, we could start from the customer segment and then expand to business segment.
Starting from The Users
Using the segmentation analysis and user research, I developed two key personas — Diligent Dan and Achiever Amelia.
Afterward, I plotted an empathy map for our target user segment based on user interviews and qualitative survey to understand their attributes and behaviors.
I will illustrate with an example. During a 1–1 discussion with a user of learning portals, he told me that he finds forums, which are important for resolving queries around the course content, deserted. However, what he actually thinks is why can’t I figure this solution on my own. As a result, he searched on the web to find a solution for his query. Ultimately, unable to find a solution, he feels exhausted and unsure of what to do.
Next, I plotted the user journey for Diligent Dan, the first user persona, to understand what he wants to accomplish and his concerns around course recommendation and reviews.
Likewise, I plotted the user journey for Achiever Amelia, who is not happy with the lack of networking and soft-skill courses in online learning.
Prioritizing the User Needs
After creating an exhaustive list of user needs based on user journey, I measured needs against two dimensions — value to customer and dissatisfaction with the current solution, to shortlist the top underserved needs.
Product Design Overview
Next, I designed the product vision based on my understanding of user segments and goals of the project.
Value proposition: Social learning with a focus on micro-lessons, personalized curriculum, and engagement to prepare for tomorrow, today.
Marketplace approach: In online learning, content quality isn’t a big problem. We have the best teachers from worlds best schools. Instead, the problem is delivering the right content to the right user. Hence, I decided to follow the marketplace approach, which will provide us data and resource to figure out the right courses. On the LinkedIn portal, you could find courses from every provider including COurseera, Udemy, and LinkedIn
“The problem in the online learning industry isn’ content quality but rather imperfect course-student matching.”
Broad product timeline: It makes sense to start with the consumer segment to experiment the perfect the delivery before launching to the unforgiving enterprise customers. In addition, enterprises often have complex needs and a high barrier to entry (Changing LMS takes 6–12 months).
Feature Prioritisation
Subsequently, I came up with a list of features corresponding to the needs focused on delivering the value proposition and plotted these feature on a 2*2 matrix to shortlist top needs.
Next, I came up with a product roadmap in line with a long-term strategy.
Minimum Viable Product
We would launch a closed group MVP to test the key product hypothesis and accordingly decide whether it’s reasonable to move forward.
Leap of faith assumptions: First, a recommendation engine could recommend the right course despite stale, inaccurate nature of profile data. Second, collaboration among peers would lead to greater engagement and aid in clarifying questions.
Thus, related to these leap of faith assumptions, we need to test the following 3 features — Recommendation Engine, LinkedIn Groups, Office 365 based notes sharing.
Expectations from MVP: With respect to our industry peers, we need to have a higher impression to purchase conversion, purchase frequency per customer, % answered posts in LinkedIn groups, and net promoter score.
Product Architecture
Here is how I envision the high-level product architecture in the future. Our user Diligent Dan will enjoy personalized, micro-architecture based curriculum and engagement with peers.
Product Mockups
Next, I create low-fidelity product mockups and interacted with user groups to conduct feedback. For testing, I conducted 1 focussed group discussion and multiple 1–1 interview.
Initial Sketches
First Design Iteration
In this iteration, I conducted focussed to group discussion to get user feedback. I chose the name of the product “Grow”
In the focussed group discussion, I conducted two tests.
Open word choice: In this test, I asked users to describe the product experience using adjectives.
Barrier Assessment: In the second test, I asked users what’s stopping them to move to the next steps, and what could have made you more comfortable.
As the product mockup was low-fidelity, I was able to avoid the aesthetic feedback, which would have been early at this stage.
Second Design Iteration
While testing, I varied my first question. Instead of asking for adjectives, I asked users to describe the experience of going through the flow in just 1 sentence. I kept the second question unchanged.
Final Product Mockups
I created this mockup using Balsamiq. This is the time when I would cede control to a designer (not saying that designers didn’t control design before this step).
Metrics for Success
I followed “GAME” framework — goal, action, metric, and evaluate.
Goal: For the phase-1, our goal is to iterate and learn to provide customer delight.
Intended user actions: Actions we want our users to take that support this goal include achieve success by doing courses, select the right courses, delighted by the course experience, interact with peers, and makes progress in the course.
Metrics for Success: Our north star metric will be ‘Net Promoter Score’ — it’s the best measure of the goal of this phase — customer delight. Secondary metrics for this phase include the percentage of users selecting the courses suggested in their LinkedIn feed, the percentage of posts responded on Networking Groups on a weekly basis, average micro-lessons completed per users, and weekly searches by recruiters to find students who registered for a particular course.
Evaluate: I will take into consideration the trade-off I am making to optimize for these metrics. For example, while the customer delight is important, we would have limited cash to burn.
Viability and Usability Risks
Finally, I also briefly highlighted some risk involved in my product design strategy. For example, intense competition in our target consumer segment could lead to a price war and lower revenue. Likewise, mismatches resulting from inaccurate profile data or faulty recommendation engine model could erode customer confidence.
That’s it from my side. I would love to hear your views in comments.
