Built at a kitchen table by a parent whose kid came home with a 63% and a knot in his stomach. Mathsmith runs on the learning science the big platforms skip — and refuses the mechanics that make practice feel like punishment.
The district's practice-app logins broke mid-semester — again — right as a module test landed hard. So one weekend, the practice site got rebuilt from scratch: every skill matched to the exact modules the teacher emails home, every wrong answer met with the name of the actual mistake ("you flipped the first fraction — only the second one flips"), and review scheduled on the forgetting curve instead of the grind.
When the teacher's plans change on a Sunday, the site matches them by dinner. Our in-house testing department is eight, ten, and twelve years old, and they are merciless.
“Do not worry about your difficulties in mathematics; I can assure you that mine are still greater.” — Albert Einstein, in a 1943 letter to a junior-high student
Most practice apps borrow their mechanics from casinos and standardized tests. Mathsmith borrows from good coaching — and from the research on how confidence is actually built.
A first-try correct gets a cheer. Correcting yourself on the second try gets the parade — because error-recovery is the exact skill a struggling kid needs to believe in.
No score that collapses when you're tired. A miss means a hint and a second swing for half credit — then the same idea returns a few problems later for another shot.
Eight frustrated attempts on one skill? It goes "back in the coals" until tomorrow, with a short video to watch instead. Grinding a tired kid teaches only dread.
"Precise work," not "you're so smart." The research here is unambiguous: person-praise builds fear of failure; process-praise builds persistence.
| ⚒️ Mathsmith | IXL | |
|---|---|---|
| How practice is structured | Mixed-skill sessions + spaced review on a 1/3/7/14/30-day memory schedule — the two highest-utility techniques in the learning literature | One skill at a time until a score threshold — feels productive today, reliably loses to interleaving on next month's test |
| When your kid is wrong | The exact misconception gets named, then a retry. The app knows the difference between "flipped the wrong fraction" and "forgot to flip at all" | A generic worked solution — and the famous SmartScore drop that can erase three answers' worth of points in one late miss |
| Feelings at the table | No timers, no cliffs, recovery celebrated loudest. Designed by the people who eat dinner with the student | Anxiety is load-bearing: search "IXL SmartScore" and read what parents say |
| Fit to the actual class | The teacher's Sunday email becomes deployed practice by dinner — this week's test modules lead the queue | 4,000 pre-built skills you match to the syllabus yourself |
| Your kid's world | Each learner gets their own themed site — name, mascot, colors, celebration style — generated from four onboarding questions and refined until they love it | One interface for 16 million students |
| Where IXL genuinely wins | Coverage is exactly what your kid is learning right now — and only that | Pre-K–12, every strand, calibrated on millions of students, with polished teacher dashboards. Real advantages, honestly stated |
| Price | Free. It costs us about $0/month to run | ~$120+/year for families; $8–12 per student per subject at district scale |
— we kept the rows IXL wins. fair fights only. —
Every problem is a memory retrieval event — the most robust finding in learning research.
Dunlosky et al., 2013Skills return on an expanding schedule, mixed within sessions; interleaving roughly doubled delayed-test scores in classroom RCTs.
Rohrer et al.New skills open with a watched example before the first attempt, then fade to independence.
SwellerWrong answers matched against known error patterns and named specifically.
Hattie & Timperley, 2007Praise targets the method, never the child — the difference between building persistence and building fear.
Mueller & Dweck, 1998Periodic "why does that work?" prompts — one of the strongest, least-shipped effects in the field.
Chi et al.An accuracy governor holds sessions near ~85% success — the empirically optimal error rate.
Wilson et al., 2019Zero countdowns, anywhere. Timed drill is directly linked to math anxiety.
BoalerEach learner's site is their own little world. This one belongs to a ten-year-old who loves a certain heroic bird — his choice, his colors, his cheering section:

