When I first set out to build Grokkoli, never in my wildest dreams did I think our system might prove effective at helping children with serious learning differences, such as cognitive impairment or dyscalculia. I envisioned creating something that would emulate the supportive guidance of a skilled tutor—an undertaking so challenging that I would have considered it a success even if it only worked for a small subset of learners.

But as our customer base grew, we noticed an increasing number of learners with ADHD or dyslexia using Grokkoli for the first time. To our surprise, these learners seemed to be progressing every bit as well as their peers. And over time, we continued to attract more of these learners to our system.

Eventually, we were approached by a family whose child was coping with 4 serious learning differences simultaneously. Up to that point, our system had demonstrated it could help learners with one or two learning differences—we had no idea if it would be effective for this learner, and we were upfront about this with this family. Instead, we promised them that if they took a chance with us, we would make reviewing and improving Grokkoli's behavior for their child our No. 1 priority.

In hindsight, this might have been the best decision we ever made. While things were a bit rough at first, we reacted quickly with changes both big and small, often within hours of reviewing this learner's sessions. Soon enough, we were able to get Grokkoli to work for them roughly 95% of the time. Her parents occasionally had to step in to provide additional support, but this already felt like a great achievement—we were proving our system could be helpful even in the most challenging of circumstances, and we loved how much more effective Grokkoli was becoming at teaching math more generally. Shortly after this, a non-special needs learner completed all of grade 2 with over 90% accuracy in just 6 ½ weeks—the fastest we'd seen yet. This wouldn't have been possible without the recent improvements we'd made.

Best of all, we now had a clear pathway for improving Grokkoli, which we've continued to follow ever since: reaching out to more and more families that have disclosed their kids' learning differences to us, and making improvements based on their feedback and where we see them struggle most.

At one point, we had a learner that was completely stuck, and we realized that the only way for Grokkoli to help them was for us to build a new feature that would be quite tricky to implement. We reached out to the family, convinced them to take a 1-month break from learning math while we made the necessary improvements, and paused their payment-collection. Time-consuming as it was, this enhancement proved immensely beneficial not only for this learner but for virtually all learners grappling with grade 3-5 math on Grokkoli.

In all, the past several months of development have been our most fruitful yet, and we are really pleased with how much better our system has gotten as a result. Grokkoli is far from perfect. But its weakest points have been systematically exposed and improved as a direct result of us working closely with special needs families using our system.

We're so grateful for the trust these families placed in us early on, at a time when we still couldn't quite promise them anything tangible in return.

What we've learned from this

Reflecting on these past several months, I think there's an important takeaway that goes against some of the prevailing wisdom often seen on Twitter and other social media platforms concerning early-stage ed-tech startups. There's a general recommendation out there that ed-tech startups should target affluent families initially, and gradually expand from there. At first glance, this seems to make sense. Schools have to be super-cautious and can't risk trying new unproven solutions, while most families are juggling multiple responsibilities and may struggle to prioritize educational investments amidst their busy lives. Affluent families on the other hand possess ample disposable income, and are less price-sensitive when it comes to their children's development. But despite these advantages, I don't think affluent families are the best audience for building educational solutions that really work.

If you're an ed-tech startup truly dedicated to cracking the ceiling of pedagogical effectiveness and bringing a fundamental shift in the economics of scaling high quality education, then special needs families are a far better early-customer to build for. Here's what it comes down to:

  • They will force you to create something that does real pedagogical heavy-lifting (as opposed to edutainment products or solutions that are only superficially effective).
  • They will force you to create something that works for the broadest and most diverse range of learners (when a learner with cognitive impairment, dyscalculia, math anxiety and ADHD is able to learn on your system, that pretty much guarantees everyone else is able to as well).
  • These families are the most in need of help and feel the pain of educating their children most acutely (so little out there works for them, including apps celebrated as best-in-class, and even specialized instructional support provided at school).

For us, and for the families we serve, this approach is paying off. Based on our internal statistics, here are some early learning-acceleration rates we've observed (compared to regular schooling) among special needs learners who use our system regularly (defined as a minimum of 3 sessions per week with 15 or more learning sessions logged to date):

Learners with Dyscalculia
Learners with Other Learning Differences

After years of falling further and further behind in school, these learners are catching up multiple grade-levels in a matter of months with us.

Rob and I will continue to improve Grokkoli and make it more and more effective at teaching math over time. And eventually, we will need to shift our focus more towards the average learner. But for now, we believe focusing on helping special needs families is not only the right thing to do, but the best possible path for us to create an exceptional education service that may one day help raise the floor in the quality of our education system. We strongly urge others in this space to do the same.