Alpha School's Secret Sauce

Spoiler: it's not AI

Something big is happening in education.

Or at least that’s what you might think if you read the hype on social media.

“Two hours of AI-powered academics is producing what traditional schools can’t achieve in six.” I’m a teacher at one of those traditional schools. Let’s say I want to learn from what Alpha School is doing. What’s the lesson for me? What should I do differently?

Last Year

I learned about Alpha School from Edward Nevraumont’s essay last summer. Here are a few things I learned from that essay:

• Based on the NWEA MAP assessment, Alpha School students were learning at about 2.6x the rate of a typical public school student in their “two-hour learning” model.¹

• Most of the education technology that Alpha School used was “off the shelf” edtech common in regular schools. The major app was iXL, a platform common in many schools that claims 1 in 4 students uses its program.

• There was basically no generative AI being used in the program. The “AI,” if we want to call it that, was limited to a spaced repetition system of the kind that’s been common in edtech for a decade.

• The heart of the program was motivation. Alpha School used a complex extrinsic reward system to incentivize effort, creating a token economy and giving students prizes based on hitting their “minimum” lesson goals and extra rewards for going beyond those minimums. The entire culture was oriented around motivation as well, with a small teacher guide-to-student ratio, adults in the room focused solely on motivation rather than instruction, and a message that students are earning their time back by completing academics quickly and learning cool stuff in the afternoons. A lot of the afternoon stuff looks cool! You can read more in Nevraumont’s essay.

This Year

According to a few different sources, Alpha School overhauled its edtech tools this school year:

Alpha is radically transforming its academic apps this year…

The most important thing to start with is that Alpha is rolling out a new academic platform, TimeBack, for the 2025-26 school year.

TimeBack is a $100m+ project to build an AI platform that allows students to learn 10x faster than sitting in a classroom in standard school.

(source)

There has never been an EdTech industry. 99% of EdTech products produced 0 efficiency gains and belong on the garbage pile of technological history.

The Timeback closed feedback loop enabled Alpha School to find the 1% of EdTech products that work and achieve 2x learning in 2 hours. This was enough to become the number one school in the country for academic growth.

(source)

Science of learning heavyweights like Carl Hendrick are tweeting nice things about Alpha School:

If you click through the links you’ll find a bunch of examples of generative AI, a whole new suite of tech tools, and some big claims about how incredible Alpha School’s new system is.

This shift isn’t surprising to me. iXL is a pretty run-of-the-mill product. It does some things well, other things not well. Alpha School was getting results by designing their entire program around motivation, and motivating students to work hard using mediocre edtech tools. They are billionaire-funded and put a ton of money into new technology. They overhauled their program. What happened next?

The Results

A few weeks ago, Alpha School students took their mid-year MAP assessment. MAP is a popular tool for measuring academic growth. Joe Liemandt, Alpha School’s principal, tweeted about the results. From his post:

The Alpha District Winter 2025-2026 MAP results are not merely high; they are statistically exceptional, placing the district’s average performance well beyond the 99th percentile of national norms.

Let’s look! Here are the math results:

Let’s take these numbers at face value. I’ll drop a bunch of statistical notes in a footnote.² I’m going to average the multipliers in the table because I don’t have access to the raw data.³ The average multiplier? About 2.5x. For reading it’s about 2.8x.

That looks familiar! Last year, Alpha School’s NWEA growth was 2.6x.

Again, let’s take these numbers at face value for a minute. Last year, the program was mostly iXL + a culture laser-focused on motivating students. There was little to no generative AI involved. This year, Alpha School has overhauled their academics, released their own platform, and incorporated generative AI throughout. They are finally doing what they say they are doing: AI-driven schooling. And the results are…more or less the same?

This is completely fascinating to me. 100 million dollars, tons of hype on the internet, grand claims about the future of education. And the results haven’t budged from bribing kids to try hard on iXL?

What If It’s Not the Technology?

Here’s my hypothesis. I believe students at Alpha School are learning quickly. I think Alpha School is being creative with the statistics to paint themselves in the best light possible, but I absolutely believe kids are learning more quickly than typical public school students. I’m a public school teacher. I see the inefficiencies every day. I don’t doubt that it’s possible to do better.

But the reason Alpha School students are learning quickly is motivation, not technology. AI has nothing to do with it. The apps and all the money spent to build them aren’t the key variables here. It seems like Alpha School has figured out how to motivate students to put in high amounts of effort to a personalized learning system. That’s it. That’s what’s driving their results.

A great article that came out two years ago on this topic called it the 5 percent problem. Personalized learning works great when students are motivated to use the personalized learning platform well. In a typical school that’s about 5% of students. Alpha School has probably increased that number — I’m sure it’s more than 5% who can succeed in their program. For that group who can succeed at Alpha School, they use extrinsic motivation to supercharge learning and get them learning much faster than in a typical school. But the apps themselves aren’t the secret sauce. Motivation is doing most of the work.

There’s also a dark side to using motivation as your secret sauce. A recent article described a high level of surveillance, using webcams to monitor students at all times, including when they are outside of the physical Alpha School. Another article from October describes the stress a student experienced as she struggled to complete a math lesson on three-digit-by-three-digit multiplication.⁴ She spent hours outside of school working on the lesson, and while she eventually completed it she fell behind Alpha School’s learning goals in the process and started a spiral of stress and anxiety. She ended up leaving the school.

Look, clearly Alpha School’s program works for some students. There is a chorus of parents singing the school’s praises on social media. But those parents are paying $40,000 or more each year. They chose Alpha School. That changes what the school can ask of students and the level of motivation the school can expect. Is this really the future of education?

Where Does This Leave Me?

I work in one of those traditional schools that Alpha School is leaving in the dust. What lessons should I take away from all of this?

I’m sure defenders of Alpha School have lots of responses here. New campuses are opening across the country. The program is growing. The technology is still new. Maybe results will improve! I’m open to new evidence. Let’s see how they do. My prediction is that their results will stay broadly the same. If that’s the case, I’m not going to take the AI stuff too seriously. If Alpha School can make significant improvements on their old numbers, I’ll take another look at their technology and think about whether it offers something of value for me as a teacher.

If motivation is really Alpha School’s secret sauce? I don’t know that Alpha’s success means anything for me. I don’t have families paying $40,000 for their kids to take my class. I’ve tried iXL. The platform didn’t motivate many of my students. I don’t have a system around me designed to maximize the potential of personalized learning, and if I did I’m skeptical it would work for all of my students.

Meanwhile I’m working hard to improve at this outdated, boring thing called “whole-class instruction.” My students take the same MAP assessment as Alpha School students. The last round, I got the best results I’ve had as a teacher here: 1.5x! My students are right on your heels, Alpha School. And you, dear reader, can have access to this innovative pedagogy! If you are willing to move to Leadville, Colorado and pay me $40,000, your kids can join my math class. Just kidding, my class is free. I’ll even throw in all the other subjects, breakfast, lunch, and sports in a bundle we like to call “public education.”

But seriously, I don’t know what to make of all this. It’s cool Alpha School has created a program where their students learn really fast. I wish we could be honest about what makes that program work. Maybe public schools have something to learn! I know we are struggling with motivation right now. Motivation is a hard problem to solve. But all of Alpha School’s marketing claims that AI is their secret sauce. That’s just not true. If Alpha School wants to fork over a little slice of that $100 million they’re spending on their technology to study motivation in public schools, sign me up. Let’s learn some stuff together.

1. I’ll note that the two-hour learning model is maybe a bit exaggerated. According to Nevraumont’s review, the academic portion of the day took from 8:30am to 12:00pm. Some of that time was in a “morning meeting” and students got frequent short breaks to run around outside so they aren’t learning for every minute of that time, but it also isn’t quite what you might imagine if you picture two hours of learning each day. There are also multiple anecdotes of students working in the evenings and weekends, either because they didn’t finish their minimum requirements during that two-hour learning, or because they want to get extra “Alpha Bucks” to purchase prizes. Let’s contrast that with a typical public school. Two-hour learning only focuses on core academic subjects. A typical public school might spend four hours each day on core academics (the rest is breakfast, lunch, recess, homeroom, and specials). The 2.6x claim is still impressive, but feels a bit less eye-popping when you think through these numbers.

2. A few statistical points: Alpha School should be comparing their MAP results to their own students’ learning before they join, not to the national median. On average, students whose families can pay $40,000 for Alpha School are probably already learning faster than the median. The 2.6x measure is also a bit weird. Here is an Alpha School parent describing some of the statistical issues, in particular that measuring with standard deviations would be more helpful than multipliers. The basic thing happening here is Alpha School playing with the numbers to paint themselves in the best light possible. I’m also trusting these self-published numbers. I don’t have any particular reason to think Alpha School is lying, but I’d love to see an independent analysis.

3. Averaging the multipliers the way I’m doing here is almost definitely wrong. The number of students in each grade level isn’t the same, and I should be averaging the underlying student numbers and not the grade level numbers. If anything I’m probably overestimating the true number, because the highest multipliers were in higher grades and my impression is that most Alpha School students are elementary-aged. Let’s call the 2.5x and 2.8x numbers rough estimates, and if Alpha School wants to release the raw numbers I’m happy to redo the analysis.

4. The first article requires a subscription, and the second article is paywalled. Feel free to email me if you’d like to read the whole thing.