The Teach-Back Method
If you can't teach it back, you don't actually know it.
The most effective study activity for the CPA exam, and the only one no other prep platform offers. Try it below, no signup.
Try Teach-Back Right Now
Answer the MCQ and teach it back. The AI grades your explanation in seconds.
Two 74s, 34 days apart. Then I changed how I studied.
I'm Nick Miller, Oregon CPA #14907. These are my real score letters from the Oregon Board. The habits that came out of those failures - drill, decode, demonstrate - are what ChatCPA is now.
Those two fails cost more than two attempts. My AUD credit expired the day after fail #2. My BEC credit expired during the delay. I passed six exams to get four credits.
Try the same teach-back method on real exam questions. Free trial, AI-graded with the rubric you just used in the demo.
Start Your Free Trial →The trap of feeling like you understand.
You read an explanation, nod along, feel like you get it. Then on exam day a different fact pattern tests the same concept, and your hand hovers between two options. The understanding that felt rock-solid is fuzzy. You guess.
Cognitive science calls this the fluency illusion: the brain learns to recognize answers, not produce them. The exam tests the second. Teach-back makes the gap visible - you produce the explanation, no notes.
What the research says.
The Feynman technique - explain it in your own words to test whether you know it - is the famous version. The research behind it is older, and three findings hold up:
- The generation effect. Producing information yourself creates stronger memory than reading the same information. Replicated across dozens of contexts since 1978. Slamecka & Graf, 1978.
- The self-explanation effect. Students who explain material to themselves perform far better on transfer tasks - applying concepts to new problems. Exactly what the CPA exam tests. Chi et al., 1989.
- Expecting to teach. Students who study expecting to teach the material learn it more deeply than students studying for a test. Same time, same material, different framing. Nestojko et al., 2014.
How ChatCPA's teach-back works.
Teach-back used to require a study partner - someone to listen and tell you where you got fuzzy. Most candidates don't have one. ChatCPA is that partner on demand.
Right or wrong, you see the canonical explanation, including why each wrong option is wrong.
The AI generates a prompt anchored to the concept tested.
30-60 seconds is plenty. Voice or text, your choice.
40% accuracy, 40% completeness, 20% clarity - what you missed, what to revisit, and a targeted follow-up for what's still fuzzy.
Reach "solid" or "mastered" twice on different days and the topic is locked in.
Why nobody else has this.
Becker, UWorld, and Gleim are content libraries. Their DNA is "watch lectures, do questions, read explanations." Teach-back doesn't fit that DNA - it needs real-time AI grading, voice transcription, and a rubric tied to each question's canonical explanation. Building it meant treating the research above as product requirements, not marketing copy.
Common questions.
No. Voice is recommended - speaking forces real-time articulation that typing doesn't. Typing works too, same rubric.
It grades against the verified explanation for each question. Strict on substance, fair on phrasing: a correct point in your own words is credited, a missed point is called out.
$59/month, $279 for 6 months, or $499 lifetime. The math favors lifetime: a typical 12-month run is $708 on monthly; lifetime is $499 once, no clock running.
No, and I want to be honest about that. The research above is general learning science - replicated for decades, but never run on CPA candidates specifically. What connects it to this exam is my own two 74s and what fixed them.
Try teach-back. Free.
The free trial includes voice and text teach-back on real exam questions, graded with the paid plan's rubric.