Efekta R&D Blog // 8th December 2025
Introduction
At Efekta, we are uniquely positioned to lead the future of evidence-based language learning. Our global scale, rich learning data, and advances in AI-driven speech analysis enable us to measure student progress with unprecedented precision — especially when it comes to spoken English performance.
While industries like fitness have long offered deeply measurable progress tracking, language learning has lagged behind, relying heavily on anecdotal guidance. With Classroom AI, Efekta is closing this gap — providing students, teachers, and organizations with objective, data-driven insights that meaningfully accelerate language development.
How Classroom AI Works
Classroom AI is a proprietary system that analyzes student speech during online 1:1 lessons, group classes and AI practice conversations. It ingests and processes multiple audio and video streams, then uses a pipeline of modules to extract quantitative speaking metrics.
Vocabulary diversity (e.g., unique word count)
Sentence complexity
Fluency indicators
Predictive proficiency level based on Efekta & CEFR scales
These metrics are backed by Efekta’s large proprietary dataset (1M+ observations), enabling highly accurate level predictions based on a single lesson.
Accuracy of the Model
Internal validation shows strong alignment between AI-based predictions and human evaluations:
~75% match with teacher-estimated CEFR level
~90% match with a more robust teacher-led verbal assessment taken within ±30 days
This level of predictive accuracy enables Efekta to track changes in student speech over time and identify learning behaviours that correlate with measurable progress.
Research Study: The Impact of Providing AI-Driven Speaking Metrics
A controlled eight-week study was run with 346 adult ESL learners in China, split into test and control groups. The test group received access to Classroom AI speaking metrics.
Key Outcomes
33% increase in lesson bookings
18% higher rate of sustained learner engagement
Faster improvement in:
Total unique words spoken
Average sentence length
Post-study interviews reinforced that students who had visibility into their progress felt more motivated and maintained learning momentum longer.
Takeaway: Objective feedback drives engagement — and engagement drives improvement.
Understanding Learning Progress: Time Is the Most Important Input
Analysis across 727 Efekta students shows that time-on-task is the strongest predictor of progress.
How Much Time Is Needed?
2–3 hours of spoken English (≈10–15 1:1 lessons) → detectable speech progress
5+ hours of spoken English (≈25+ 1:1 Lessons) → detectable level progression
Conclusion: Students who stay active long enough do progress — therefore learners need help establishing consistent habits.
Things We Learned
Consistency of Study
Students active in 80% of available weeks are
➡ 73% more likely to progress a level
compared to students active ≤60% of the time.
Zone of Proximal Development (ZPD)
Optimal distribution:
60–80%: appropriately challenging lessons
15%: 1–2 levels below
Up to 10%: stretch lessons 1–2 levels above
These are just examples of a wide range of insights we continuously derive from analysing our course. Our clients benefit from our curiosity around learning behaviours and our bias towards experimentation to improve our product.
Why This Matters
Classroom AI is helping Efekta provide personalized, evidence-based learning journeys. Today’s insights form the foundation of a future where every learner receives:
A tailored curriculum
Smart prompts to adjust lesson difficulty
Suggestions for ideal study frequency
Adaptive feedback grounded in real speaking behaviour
Conclusion
AI-driven insights are enabling Efekta to finally deliver what language learners have always needed: clarity. Clarity on where they stand, how fast they’re progressing, and which actions most effectively move them toward their goals.
By combining Classroom AI’s powerful speech analytics with our massive library of learner behaviour data, Efekta is building a new era of measurable, personalized language education — one that is more motivating, more predictive, and ultimately more effective.
