Why obsession with customer success beats defensible technology
Vaidy Bala
February 01, 2026
I hear this once every month or so when I show Spark to people - "Your moat is not strong enough"
A few days ago, I applied to a startup accelerator. As part of the application, I had to upload a pitch deck.
One of the implicit questions every pitch deck is expected to answer is:
"What is your moat?"
In other words, how tall is your wall - and how hard would it be for someone else to jump in and eat your lunch?
The answers people expect
In the deck, I wrote the kinds of things people expect to see:
We've built proprietary technology that isn't trivial to copy
We already have paying customers through PracticeNow, which gives us runway to build Spark patiently
Replicating Spark isn't easy. Replicating PracticeNow isn't a cakewalk either, because of the hundreds of business rules baked into it
I finished the deck and submitted the application.
Only after that did something become clear.
None of those things are our real moat.
They're reasonable answers. But they're not the truth.
Our real moat is how far we're willing to go to make experts and learners succeed.
When we focus relentlessly on their success, our own success becomes a by-product.
A lesson from Toyota
Kiichiro Toyoda was the second-generation leader of Toyota. Long before the company became famous for cars, he was helping his father grow the loom business.
One day, he was told that thieves had broken into the factory and stolen the detailed blueprints for their latest loom.
Everyone around him panicked. It felt like the company's future had just been stolen.
Kiichiro's response was calm:
They can follow those plans and eventually build a loom. But we improve our looms every single day. By the time they catch up to that design, we'll already be ahead.
They don't have the failures and the learning that went into these plans. We just need to keep improving."
Toyota's moat wasn't the blueprint.
It was the system - and the daily habit - that kept producing better blueprints.
What that means for Spark
It's the same with Spark.
Our features can be copied. Our technology can be reverse-engineered.
But only we have the failures and the lessons that got us here.
For example, in one early version of our course structure, the AI had split a long physics derivation right in the middle in an attempt to keep the section short. But it completely broke the flow of the teacher's explanation.
Our early adopter tutor called it out, and that feedback forced us to redesign the flow.
That's just one of so many other similar instances. Here's another one.
We built Spark with just two online high-school tutors.
They gave us two invaluable things:
Access to their Zoom accounts (real "classrooms") - and a fast-growing repository of real teaching content
Constant candid feedback from both themselves and from their 100+ students
The rest was on us.
Could we keep working on the product day after day?
Could we keep generating ideas, testing them, discarding what didn't work, and improving what did?
Could we keep showing up even when the work stopped being exciting?
Looking back at the last 10+ months, I think we did a good job on all of the above.
The real test is harder: will we care this much - every week, for the next decade?
If the honest answer to that stays "yes", the question of whether we have a moat becomes moot. 🙂
Onward!