Over 200,000 people enrolled in Replit’s ‘100 days of code’ course. Amazing, right? But a graph shared by Y Combinator co-founder Paul Graham tells a different story. These numbers massively drop off - and fast.
This is not uncommon.
But why did it happen? The course looks well designed.
Was it the length? The delivery mode? Or something else entirely?
Here are 10 reasons why, each with some potential paths forward.
1) Talking to Humans
Who’s it for?
Problem: "100 Days" seems to cater to a lot of different groups. Is it for students? Novice programmers? Product managers? Solo entrepreneurs? It’s kinda hard to tell. Not being clear on who it’s for makes everything else - learning design, messaging, marketing - more difficult.
Path: Choose a single audience, or clearly articulate up to three sub-groups that people can identify with
Value proposition
Problem: Right now the course is anchored in "Learn coding”. This is indeed what it’s about, but it’s a mask. Skills are mentioned (in this case, Python), but we can go further.
Path: Show how these skills empower people. Frame as offering as a journey of transformation, not just a collection of knowledge. The ‘Who’s it for?’ question help us here. For example, perhaps our value prop for side project builders is ‘Launch the projects to kickstart your entrepreneurial journey’.
2) Structure & Design
Time Warp
Problem: The course description leads with ‘15 minutes per day’. But later it turns out that many days require 30 or more. Two problems here. First, people are busy. Finding 30 minutes a day can be a very different challenge to 15. Second, as a course designer it’s easy to underestimate the time people need to do something - so the estimates may well be off.
Path: Audit the challenges with a range of users to better understand the durations. Include more context on the expected duration based on skill levels, and within larger challenges build in 1 or 2 ‘checkpoints’.
Concept Bloat
Problem: A few people commented that later challenges became difficult to follow because the concepts became too large to understand. This is a common issue that’s two-fold: more advanced concepts usually have greater complexity; and often as designers we can get tired later in a build.
Path: Later challenges may be better as extensions or remixes of previous ones vs. net new (think ‘New Wine, Old Oak’). This is also where Designing in Reverse can help us avoid fatigue, assumptions and miscalculation.
Make it real with missions
Problem: The challenges feel appealing, but lack real-world applicability. However, just because something is ‘real world’ doesn’t always mean it’s interesting or relevant to people. Tricky.
Path: Thinking in themes and missions can solve this problem - and a few of the others, too. Here’s an example. A few of the existing challenges can be woven together into ‘the games arcade suite’. Imagine making a suite of 4-5 games that could be used on a simple phone or an old school arcade - Rock, Paper, Scissors; Tic Tac Toe; Pong, Defender. A concept like this blends real-world applicability, with a pop culture theme (nostalgia for arcades / retro phones), and a real output that demands a decent amount of investment - but not too much to put people off. We could even ramp this up with a ‘brief’, presented by a company or practitioner (ideally a Replit customer!).
Scaffolds & Bridges
Problem: The drop-off rate was serious. But day 3 blues, week 2 slump – these engagement dips are natural obstacles. How do we overcome them?
Path: Our Missions approach helps a lot here. But we can also add some scaffolding and other strategic interventions at these crucial junctures. First, know where the big dips are, then build bridges and scaffolds to help people cross them. Offer live Q&A sessions, badges, or provide personalized feedback to reignite the spark.
3) Modes & Connectivity
Multiple engagement modes
Problem: The entirety of the course is about 2000 minutes, or about 30 hours. And while you can dip in and out, there’s only one main mode of engaging.
Path: Offer multiple modes and journeys. How about a condensed group program - say two full days covering the top 40 concepts? There’s also likely a revenue model in here as it’s likely people would pay for it, especially if there’s some kind of credential at the end. Speaking of which - this opens up two other areas to explore…
Meaningful incentives
Problem: You get skills, projects, a sense of accomplishment. But in practice, these can feel strangely limited. There’s not much pushing people to engage.
Path: Badges, certificates, and leaderboards can feel corny, but when executed well they’re incredibly effectively. Well-placed incentives can provide nudges of encouragement, recognizing progress and motivating learners to push through dips in motivation. Perhaps more importantly, signals matter. There’s a reason why people spend so much on Louis Vuitton bags, go to Harvard, and want to work at Nike. The power of being ‘Replit 100 Certified’ shouldn’t be underestimated - and it’s also a handy piece of brand equity for Replit.
Social Dynamics
Problem: This drop-off rate is not unusual for self-paced courses (some stats put it as high as 96%). It’s easy to say ‘don’t quit’, but come on - we’ve all quit stuff. 100 days straight is a lot.
Path: We’re social animals. Connectivity and accountability matter. This can be as simple as discussion forums, through to collaborative projects, coaching, and the aforementioned group program. Don’t underestimate our desire for camaraderie and to feel seen and heard.
Just one more thing...
Do the numbers lie?
Problem: The graph appears damning. Or is it? How do we know? As Richard Feynman would say - don't fool yourself, and you are the easiest person to fool.
Path: It may actually be good news. Are people confident enough after day 3, 10, or 20 to go and do something with what they’ve learned? Has Replit in fact achieved success in what looks like failure? Could we do something else with all the remaining challenges - create a whole new product offering perhaps? This is where user connection matters - qual matters just as much as quant. Let’s learn from those humans, so we can better understand them, and make something they love…
So, there you have it. A few ways to level up not just 100 Days of Code, but any online learning experience.
By getting clear on the value proposition, thinking in thematic missions, and offer multiple engagement mode, you’ll be well-set to help people go somewhere they never thought possible…