Khan Academy vs. a $150/hr Tutor: What Research Says About Math

At $150 an hour, a private math tutor feels like it should work better than a free website. That’s the intuition most parents operate on. And for a subset of kids in specific circumstances, they’re right.

But the research is less flattering to tutoring — and more flattering to Khan Academy — than most parents expect. There are also some very specific conditions under which Khan is nearly useless. Understanding the actual evidence helps you spend less money and get better results, which is the only outcome that matters here.

What Khan Academy Actually Is (and Isn’t) as a Learning Tool

Khan Academy is not a tutor. That distinction matters because people often evaluate it as a replacement for human interaction, which sets up a comparison it was never designed to win.

Khan Academy is a mastery-based adaptive learning platform. The core mechanism: a student watches a short explanatory video, then works through a sequence of practice problems. The platform tracks performance and adjusts problem difficulty. It requires repeated correct answers before a topic is marked “mastered” and the student can advance. This mastery gate is the feature that distinguishes Khan from passive video watching — which is important given what we know about the fluency illusion in educational video learning.

What it cannot do: diagnose why a specific child is stuck. Khan doesn’t know that your 9-year-old struggles with fraction division because she’s been carrying a misconception about numerator-denominator relationships since second grade. A good human tutor can diagnose that in one session.

What it can do: provide unlimited, patient, adaptive practice at zero cost, available at 11pm when a child is panicking before a test.

Khan Academy’s Own Efficacy Data vs. Independent Studies

Khan Academy commissioned a major external evaluation in 2014–2015 with SRI Education, and a subsequent evaluation in 2016 by the RAND Corporation. The RAND study is the more cited and more rigorous.

RAND’s 2016 analysis, led by researchers Pane, Steiner, Baird, Hamilton, and Pane, examined Khan Academy use across 20 schools and found that students who used Khan Academy for math for at least 30 minutes per week showed statistically significant improvement in math scores compared to matched students who did not use it. The effect size ranged from approximately 0.15 to 0.20 standard deviations, depending on the grade and the comparison condition.

That’s a real effect. An effect size of 0.15–0.20 is consistent with what Hattie’s Visible Learning database calls a “developmental effect” — the kind of gain you’d expect from a year of normal schooling. It’s not transformative. But it’s measurable and consistent, and it happened with a free tool used ~30 minutes per week.

The important caveat from the RAND study: the effect was almost entirely concentrated among students who used it consistently and for practice problems specifically, not among students who primarily watched videos. This aligns with the broader cognitive science: the platform works because of its retrieval practice mechanism, not its explanatory videos.

Murphy et al. (2014), an independent RAND review of Khan Academy effectiveness in five California schools, found more mixed results — positive for some grade levels and subjects, neutral for others. They noted that outcomes depended heavily on teacher integration: when teachers assigned Khan Academy as structured practice with accountability, it worked; when students used it on their own without structure, effects were minimal.

The National Education Policy Center at the University of Colorado has also reviewed Khan Academy and noted that while the evidence base is more solid than for most EdTech products, it remains limited to specific implementation conditions and the effect sizes are modest.

What Private Tutoring Research Shows About Outcomes

The research on private tutoring is more complicated than its reputation suggests. The most rigorous modern review is Kraft and Falken’s (2021) meta-analysis in Annals of the American Academy of Political and Social Science, which found that high-dosage tutoring (3+ sessions per week, in-person, curriculum-aligned) produced effect sizes of 0.37–0.53 standard deviations — meaningfully above Khan Academy’s measured effects.

But “high-dosage tutoring” at 3+ sessions per week is not what most families hire when they call a private tutor. Most private tutoring is one session per week, 60 minutes, often the week before a test. And the research on low-dosage tutoring is considerably less impressive. Kraft and Falken specifically note that one-session-per-week tutoring consistently shows weaker effects than high-dosage models.

The What Works Clearinghouse has reviewed a number of tutoring programs and finds strong evidence for some structured tutoring interventions (particularly curriculum-specific programs like Reading Recovery), but much weaker evidence for the generic “find a tutor on Care.com” model that most parents actually use.

Private tutoring also has enormous variance in quality. The market for private tutors is unregulated. A tutor charging $200/hr may be an excellent diagnostician and teacher or may be a college student who was good at calculus. There’s no certification or standardized effectiveness data. The research on tutoring effectiveness is largely drawn from structured intervention programs, not unstructured private arrangements.

Khan vs. Private Tutor: Side-by-Side

Factor	Khan Academy	Private tutor (typical)
Cost	Free	$60–$200/hr depending on market
Accessibility	24/7, any device	Scheduled sessions, varies by availability
Accountability	Self-paced; requires child motivation	External accountability from human relationship
Diagnosis of root cause	No — tracks performance, not underlying misconceptions	Yes, if tutor is skilled
Adaptation	Algorithm-based difficulty adjustment	Human judgment; quality highly variable
Evidence base (math outcomes)	RAND 2016: d = 0.15–0.20 at 30 min/week	Kraft & Falken 2021: d = 0.37–0.53 for high-dosage
Best condition for effectiveness	Consistent practice, 30+ min/week, problem sets not videos	3+ sessions/week, curriculum-aligned, skilled tutor
Failure mode	Fluency illusion if student only watches videos	Low-dosage, poor diagnosis, test-prep only

The honest read: a skilled tutor doing high-dosage, curriculum-aligned work with your child will probably outperform Khan Academy. Most private tutoring arrangements are not that. Under real-world conditions — once a week, general practice, variable tutor quality — Khan Academy’s measurable effect sizes are competitive with and sometimes exceed what families actually get from private tutoring.

When to Use Each — A Decision Framework for Parents

Use Khan Academy when:

• Your child is keeping up overall but needs more practice on specific topics
• You can build a 30-minute daily habit (the effect disappears at lower frequency)
• Your child is self-directed enough to work through problem sets (not just watch videos)
• The issue is fluency — they understand the concept but get shaky on the arithmetic
• You want a baseline measurement: Khan’s teacher dashboard shows mastery levels by topic, which helps identify gaps

Consider a private tutor when:

• Your child has a specific, persistent misconception that isn’t clearing up with more practice
• The difficulty is new content, not just fluency — something genuinely hasn’t been understood yet
• Your child is significantly behind grade level in a way that will compound over time
• Test prep with a deadline — a structured human plan for SAT/ACT math is more efficient than self-directed platform use
• Your child is completely disengaged from self-directed digital tools

Don’t use either alone when:

• There’s a consistent pattern of confusion across multiple topics — that may indicate a learning difference worth assessing with a specialist
• The child refuses to engage with Khan but is also passive with a tutor — that’s a motivation and engagement issue, not a tool issue

For a broader picture of how these tools fit into a child’s overall STEM development, our article on the engagement cliff in educational apps explains why digital tools — including Khan Academy — stop working for many kids after a few months.

The Hybrid Approach That Outperforms Both

The strongest outcomes in both the tutoring and EdTech research come from combining structured digital practice with human diagnostic feedback. You don’t need to choose one or the other.

The model: Khan Academy for daily practice (habits and fluency), one tutor session every two to four weeks focused specifically on diagnosing misconceptions that show up in the Khan data. The tutor doesn’t re-teach everything — they review the mastery report, find the weak spots, and spend the session on root-cause diagnosis. Then Khan handles the daily practice on those specific topics.

This is more cost-effective than weekly tutoring — you’re buying two to four hours of human time per month instead of four-plus, and the human time is targeted rather than exploratory. The NWEA’s research on math intervention supports this kind of targeted diagnostic approach: knowing specifically which concepts are weak, rather than providing general math support, is the variable that most strongly predicts improvement.

A practical note: Khan Academy’s teacher dashboard is free and public. You can create a “teacher” account, add your child as a student, and see their mastery levels by specific topic — fractions, equivalent expressions, linear equations, probability, whatever. That report becomes the agenda for the tutor session. Instead of “my kid is struggling with math,” you walk in with “she’s at 40% mastery on adding and subtracting negative numbers and 25% on interpreting functions.” That’s a completely different and more efficient conversation.

For a deeper look at how to evaluate whether any learning tool is producing real understanding vs. surface familiarity, see our piece on why hands-on STEM learning builds deeper retention.

Key Takeaways

• The RAND 2016 study found Khan Academy produced a real, measurable math improvement (d = 0.15–0.20) among students who used it for 30+ minutes per week — almost entirely through problem practice, not video watching
• High-dosage tutoring (3+ sessions/week, curriculum-aligned) produces effect sizes of 0.37–0.53 — stronger than Khan, but most private tutoring is not high-dosage
• Real-world private tutoring quality is highly variable; the research on structured tutoring programs doesn’t automatically transfer to the general private market
• Khan Academy fails when students primarily watch videos without completing problem sets — the platform’s mechanism is retrieval practice, not passive viewing
• The hybrid approach — daily Khan practice plus monthly diagnostic tutor sessions targeting specific Khan weak spots — outperforms either tool used alone
• Khan Academy’s mastery dashboard is a free diagnostic tool worth using even if you also have a tutor

FAQ

Is Khan Academy good for kids who hate math?

Not on its own. Khan Academy requires some self-motivation to work through problems when things get difficult. Kids who have significant math anxiety or strong negative associations with the subject often need a human relationship — encouragement, calibrated challenge, and real-time adjustment — before digital practice becomes useful. Khan can work well as a follow-on tool once a tutor has rebuilt some confidence.

At what grade level does Khan Academy become less effective?

Anecdotally, parents report decreasing use in high school, particularly for Algebra II and beyond. The evidence base also thins at higher grade levels. Khan is strongest for foundational math through pre-algebra. For advanced topics like calculus or statistics where conceptual depth matters more than procedural fluency, the tutor model shows stronger advantages.

How do I find a tutor who’s actually effective?

Ask for their specific approach to diagnosis: “How do you figure out why a student is stuck?” A good tutor should be able to describe a process — working backwards through a problem, checking procedural vs. conceptual understanding, asking the student to explain their reasoning. A tutor who says “I explain the topic clearly” is describing re-teaching, not diagnosis.

My child already has a school math teacher. Why isn’t that enough?

Classroom teachers have 25–35 students and limited time for individual diagnosis. They can identify who is behind but rarely have the one-on-one time to find out exactly why. Khan Academy and private tutors both address this individual-diagnosis gap in different ways — the platform through adaptive algorithms, the tutor through direct conversation.

Does Khan Academy help with standardized tests like the SAT?

Khan Academy has a free, official SAT prep partnership with College Board. The evidence on its SAT effectiveness is genuinely strong: a 2017 College Board study found that 20 hours of Khan SAT prep correlated with a 115-point average score increase. This is one of Khan Academy’s highest-evidence use cases. For ACT, the evidence is weaker and the platform is less tailored.

My 7-year-old finds Khan boring. Is it the wrong tool?

Probably yes for now. Khan Academy’s design skews toward upper elementary through high school. For early elementary math, programs specifically designed for younger children — like IXL, ST Math, or structured practice with manipulatives — often show better engagement and comparable learning outcomes. Khan becomes most effective when children have enough reading fluency to process the problems independently.

---

About the author

Ricky Flores is the founder of HiWave Makers and an electrical engineer with 15+ years of experience building consumer technology at Apple, Samsung, and Texas Instruments. He writes about how kids learn to build, think, and create in a tech-saturated world. Read more at hiwavemakers.com.

Sources

1. Pane, J. F., Steiner, E. D., Baird, M. D., Hamilton, L. S., & Pane, J. D. (2015). Continued Progress: Promising Evidence on Personalized Learning. RAND Corporation. https://www.rand.org/pubs/research_reports/RR1365.html
2. Murphy, R., Gallagher, L., Krumm, A., Mislevy, J., & Hafter, A. (2014). Research on the Use of Khan Academy in Schools. SRI Education / RAND. https://www.sri.com/work/publications/research-use-khan-academy-schools
3. Kraft, M. A., & Falken, G. T. (2021). “A blueprint for scaling tutoring and mentoring programs in public schools.” Annals of the American Academy of Political and Social Science, 695(1), 143–164. https://doi.org/10.1177/0002716221992490
4. NWEA. (2022). MAP Growth Norms Study. NWEA Research. https://www.nwea.org/research/
5. College Board. (2017). “Official SAT Practice: Early Empirical Findings.” https://collegereadiness.collegeboard.org/about/scores/sat-practice
6. What Works Clearinghouse. (2022). “Tutoring Programs.” Institute of Education Sciences. https://ies.ed.gov/ncee/wwc/Topic/33
7. Hattie, J. (2009). Visible Learning: A Synthesis of Over 800 Meta-Analyses Relating to Achievement. Routledge.