Inspired by social media and often compared to LinkedIn by university stakeholders and Reddit or Discord by students.
Communities encourage thoughtful and relevant content, shared goals, and true social connections.
Unique features that differentiate the approach and drive outcomes include:
A patented point system (rewards interactions and those whose content is relevant to everyone)
A feed favoring content relevant to the group (not individual interests)
Stakeholders in solving learner engagement and persistence problems in education include:
Instructors often decide technology use in their classes. Each have goals (e.g., specific learning objectives, assessing students, etc.) and needs (e.g., easy setup, fits into other workflows, etc.) that vary by subject and modality (e.g., online vs. on-campus).
Use cases outside of courses require different setup workflows and functionality compared to courses and are setup by staff with significantly different goals and approaches.
Learning, Curriculum, and Program Designers often influence technology adoption and can ensure better practices. They are often concerned about accessibility, student and instructor support, and repetitive task difficulty (e.g., configuring many courses).
They budget and pay for technology and care about ROI, using adoption from other stakeholders and proof it produces better results than cheaper alternatives to make decisions. However, academic freedom, decentralization, and their associated politics mean that even with ROI, institutional leaders can rarely force adoption or dictate how something is used.
The most frequent and numerous users, yet their experiences, workflows, and feedback are fundamentally impacted by other stakeholder choices. Their feedback influences others, but it is often interpreted differently depending on the stakeholder (e.g., Educators expect students to complain about difficult things).
Education is a price-sensitive industry with many enterprise needs. To achieve impact and grow revenue, Yellowdig had to use limited engineering and success resources to create repeatably positive outcomes for all other stakeholders.
All stakeholders are influencers: Everyone thinks about Learners in EdTech, but designing just for them will not optimize outcomes if it means others' needs are poorly met. Every stakeholders thoughts will impact user growth and revenue.
Rowing in different directions: More stakeholders introduce more possibility for misalignment on the most important goals, requiring more upfront institutional alignment and more ongoing supports to make sure the desired outcomes are achieved.
Stakeholder needs can conflict: The more stakeholder personas and the less they are pursuing the same goals and outcomes, the more likely helping one stakeholder could hurt another.
It is sometimes necessary to withhold functionality from one stakeholder or design very different experiences for them.
The best roadmap items will often be ones with smaller positive impacts across many stakeholders rather than a bigger positive impact for only one.
In complex environments like universities, a product must be balanced in meeting all stakeholders' needs and include supports in and around the product (e.g., in-app onboarding, external training webinars, etc.) and from their internal teams and processes.
I began at Yellowdig when there were approximately 10 employees, as a User Experience Researcher and Learning & Academic Support Manager, to understand users and help build and deploy a platform that had been previously built by the CEO as Product Manager and a small set of engineers. My mandate was to use my research skills and learning science background to improve feedback and drive impactful adoption that ultimately created growth and revenue.
In the ensuing years I became Head of Customer Success, Vice President, Product & Success (formally "Academic Product Engagement"), and Senior Vice President, Product & Success (formally "Academic Engagement"). I was the first to occupy an SVP role, which was the highest role below our CEO at the time.
As I was promoted to VP I directed about half of the company headcount. The company grew while I was VP from about 20 to about 50 total employees. I was in charge of our Product Design team and responsible for Product Management, managed our Quality Assurance team, and ran our client enablement operations (Success and Technical Support). I also regularly worked with the Marketing and Partnerships teams to align efforts with the product and product communications, develop strategy based on user and client analyses I was doing, and created sales enablement materials, participated in promotional events, and helped with strategy and participating in our highest-stakes sales and success meetings.
In my roles, I founded, performed, and grew teams to handle an array of functions. These included doing formalized user experience research, creating and leading a product design team that better responded to user and client needs, strategically updating the product roadmap, holistically managing product releases and post-launch support strategies, and creating and training the Success team to offer end-to-end client experiences that reduced Technical Support needs while growing adoption and revenue.
My initial mandate at Yellowdig was to answer two questions:
What makes so many Educators wildly successful and happy with Yellowdig while others are decidedly not excited?
What can we do to make more of those happy Educators and Learners?
Using mixed methods like interviews, student behavior data, studying differences in how instructors managed communities, asking for course designs from syllabi, and studying our marketing and support materials, I came to find Yellowdig was being introduced as an auto-grading discussion board.
Most Instructors were not aware of what the platform was actually intended to help with. Their Institutional Leaders were often paying for Yellowdig to provide a better student experience and increased term-to-term retention, not necessarily to make teaching easier or more automated. Every stakeholder would appreciate less work as a side perk, but it was effectively being communicated as the only motivation behind Yellowdig.
They applied discussion board (DB) methods they already knew ("post once, comment twice").
They used weekly assignments with Sunday deadlines to focus students on one important course topic per week.
They performed their normal DB role minus the effort to individually award points. They corrected Learners who needed it, occassionally taking their points, without really taking part in the conversations.
At this time, they did not have the data to understand the chain reaction that was happening or helpful guidance for avoiding it:
Data: Engagement is normally low in DBs and Yellowdig normally beats them, so by comparison Yellowdig looked better than DBs. They had no reason to suspect their outcomes were actually bad relative to what was possible in Yellowdig and we were not showing them it was.
Helpful Guidance: Yellowdig had not yet provided best practice recommendations at scale and was finding it hard to convince Educators to try them. If you look for advice on creating better online discussions, the normal advice is "create better prompts" and that is the advice most tried to follow in Yellowdig.
Here is the discussion paradigm chain reaction I observed:
Students heard "discussion board" and those familiar with them immediately fell into a low engagement behavior pattern.*
Show up once a week.
Create a post to answer the instructor's question.
Post it at the end of the week to more easily comment on two peer responses at the same time.
Read just enough posts to find two that are easy to comment on.
Leave and never think about it again.
The gamified point system was configured for "auto-grading," rather than allowing Learners to earn more points from other students as they created helpful content for them.
Learners made mediocre content just good enough to not have points taken by the Educator.
If the Educator did not visibly participate, learners assumed they were lazy (auto-grading) and did not really care ("Why should I care if they don't?").
When Educators did participate, it was often just to give corrections, which was interpreted negatively (i.e., Darth Grader).
Learners all answered the same question, so responses were usually text heavy and repetitive. At a glance, any stakeholder would see a WoT ("wall of text") that was not engaging or using Yellowdig's exciting features.
Learners were often told to respond to Educator posts, making them stick to the top and breaking the feed algorithm.
Learners sifted through a lot of repetitive, mediocre content while learning little and getting a sense few others cared.
Seeing little of value, Learners stopped wasting any unneccessary time reading or coming back.
Instructors and Institutional Leaders noticed all of these things that had made them seek out Yellowdig as a replacement for their LMS discussion boards in the first place.
This deteriorating participation was the opposite of the exciting spaces where outcomes often got better as social connections formed. Even very small impacts of any one of the links in this chain reaction gets compounded over time by network effects.
I later came to see student expectations about discussion boards (#1) as the hardest to overcome. Doing so requires convincing a large group of students who already learned something is going to be a waste of time to ignore what prior experience taught them.
That is why a first change I recommended (list below) was for the company to stop using the term discussion board.
*The discussion paradigm is a structurally poor starting point for conversation; I became an outspoken critic (e.g., this whitepaper) as I amassed more understanding of the behavioral patterns the paradigm establishes by design.Unlike writing a paper or studying for a test, in social learning activities one Learner's demotivation negatively impacts the experience for others. In talking with Educators and Designers, I liked to point out that every disengaging student gives the gift of demotivation to every other student. Demotivation over the course of a term is contagious and clearly shows up in discussion boards.
It was clear many more people could like Yellowdig, even if it still needed many things at the time, but the discussion board (DB) framework impacted every stakeholder:
Educators remained frustrated spending time on disengaged students, even with "auto-grading."
Learners read and learned less and built fewer social connections, yielding fewer positive impacts they could feel.
Institutional Leaders saw fewer of the engagement and retention improvements they were paying for.
At the time, Yellowdig experienced many immediate and long-term impacts that were evidence of other friction:
When I arrived, web traffic was frequently 100x normal baseline on Sunday nights, creating use surges and instability.
More procrastinating Learners lied more often about having technology problems to try to get missed points. It created support investigations, unclear signals about tech failures, and friction with instructors. This and other integration issues showed courses using the DB paradigm created at least 10x the support tickets.
Adoption was slower and churn risk higher wherever the DB paradigm was used. I later found client organizations systematically using DB designs had NPS scores ~50 points lower.
Instructors and most other stakeholders seeing Yellowdig with DB designs came to a reasonable immediate conclusion; Yellowdig did not really "work" to engage students. Not surprisingly, students who are not having a better and more engaging learning experience in the moment also do not magically show amazing longer-term outcomes like higher retention.
Previously maintained as factual documentation by engineers, I rewrote the Knowledge Base for Educator and Learner guidance.
Added articles for different user groups, like an Orientation Guide for Learners.
Made it more opinionated and a tool for helping partners manage change:
Highlighted Yellowdig's purpose: Driving engagement and retention.
Created a Getting Started section institutions could easily share with new adopters, introducing important concepts and including templates for course designs to make YD easier to adopt and improve first-time success.
"Scary" warnings with data and justifications were added to some articles to dissuade common poor choices.
Added recommendations for how instructors should behave and interact, including some things that were common ways instructors spent time that did not help (e.g., putting accolades/badges on every post).
Changes reflecting the above were prioritized in the product to help quickly provide just-in-time guidance.
Simple vocabulary edits, help icons, and copy additions were written for pages being worked on for other releases.
Some additions initially got pushback from engineers because it was not "clean". I explained most setup pages were not core to the experience and showed Educators were only using them a few times per year. It needed to support good, confident choices more than be clean to look at and quick to use.
Help article permalinks and links to our commercial site were used to keep in-app copy simple while providing far more extensive guidance in areas easily managed by other teams and without engineering resources.
Some examples illustrate easy choices to add needed functionality. Others tell stories demonstrating tension in removing choices or withholding functionality from certain stakeholders to improve or safeguard the experience for others.
Vocabulary Change: Discussion to Community (Everywhere - Success; Growth)
Yellowdig's Classic to Engage Transition (Product - Quality Assurance and Functionality Choices)
Self-Directed Certification & Knowledge Base (Support - Scalability; 24/7 Access)
*ML Point Default Project (Product - Setting Optimization and UX Improvement)
*Community Health Dashboard (Product - New Feature)
*Action Center & AI Summaries (Product - New Feature)
Excavator: Go-to-Market App - AI Data Compiler & Report Generator (Operations - Success; Growth)
Yellowdig's Product Update Feed
*More coming soon!
In my research and success capacity, partners I worked with to research outcomes and who templated course designs all generated average NPS ratings in excess of 50 (highest = 57) across various educators; the product itself is very capable of creating exceptional experiences for most Educators when implemented as suggested.
The NPS ceiling for Yellowdig's product was always quite high given some early good decisions, but changes I made in it and to our enablement operations significantly raised the floor that could be expected among new clients.
When COVID-19 hit, I suggested offering existing clients free use for new Educators who got certified prior to September 2020.
I knew Institutional Leaders would appreciate the help moving operations online.
One of my earlier analyses had shown the course was impactful and certified Educators needed almost no human support and had exceptional LTV.
Knowing Designers and Educators were wary of free offers adding tools for a single term, I further suggested we allow the free use until client renewals, giving everyone time to find an expanded budget.
The offer was communicated to clients and was very highly utilized. Our existing user base quadrupled from the prior Spring.
As my analysis suggested, new Educators needed little support and added it to more courses after the first term.
We did not hire a single Success employee and needed to hire only one additional Technical Support employee to manage a 300% increase in Learners.
That was not achieved by offering a poor experience. Most Educators that adopted Yellowdig during the emergency move to online instruction continued to use Yellowdig even after their courses returned to in-person instruction.
Institutional Leaders ultimately licensed over 90% of the added free use at their next renewals.
Steady growth in G2 rankings, maintaining 6th out of over 150 companies in their Digital Learning Platform category as of November 2026.
These come along with many positive G2 reviews and category leadership awards for the usability and support functions I directed (Summer 2024, Winter 2024, and Spring 2025).
TIME’s recognition as one of 15 World EdTech Rising Stars
Listing as a HolonIQ top education company
A particular honor for me as a Product Manager, the Product Manager for Boston University's Online MBA Program, Charlotte Wang, chose Yellowdig for her appearance on a "My Favorite Product" series on Build It Ship It.
With the product and user journey improvements presented here, I was able to reliably create large persistence and retention improvements.
The outcomes could be replicated and measured in various ways with partners among their own students (e.g., University of Maryland; Arizona State University), helping financially justify larger contracts and marshalling university help in driving expansion.
In one case, the persistence outcome was so big relative to any other technology they had seen that they initially dismissed it as a fluke. Real "belonging" is powerful for addressing retention issues.
The consistent outcomes and scalable enablement allowed Yellowdig to market it's 1% improvement guarantee and offer a freemium model without significant risk. It reduced hesitancy for new Educators and clients to adopt and helped drive growth.
With only in-platform resources, self-serve enablement, and a team of six Success and Support staff, my teams sustained over a 96% renewal rate and 20-30% revenue and user growth rates among clients.
Most universities have course redesign policies limiting redesigns to every 4 or 5 years. To sustain >20% growth, many existing clients were adding Yellowdig to courses nearly as fast as their standard redesign policies would allow.