Picture an engineer in your Vietnam office. She has gone to the company English class every Thursday for a year. Her emails are clear. Then on Monday a client in London asks her a direct question on a video call, and she hesitates before saying, "Let me confirm by email."
Most leaders read this as a teaching problem and start looking for a better instructor. The classes are probably fine. What is missing is speaking time.
Why does classroom-only corporate English training run out of speaking time?
Take a 60-minute class with 12 people. If the instructor said nothing at all, each person would still get only five minutes. Over a year of weekly classes, that adds up to about four hours of speaking.
Speaking well under pressure takes repetition. You say the words, hear what went wrong, and say them again. Four hours a year gives very little of that.
Here is what this means for cost. Speaking time per person is instructor hours divided by the number of people in the class. If you want each person to speak twice as much, you pay for twice the instructor hours.
Scheduling gets harder as well when the team works across three time zones. Organizations may struggle to give every employee enough individual speaking practice through scheduled instructor sessions alone.
The limit comes from the schedule, and the instructors are not the cause.
There is a second problem. Many programs report attendance. Attendance tells you people came to class. It does not tell you whether they can speak in the meeting. Researchers who study how training carries over into daily work have found that this is hard to evaluate, partly because good measurement tools are missing (Myers, University of Connecticut, 2009). If you cannot measure speaking, you cannot tell whether the class is helping.
What is the difference between AI English training software and traditional corporate English training?
AI English training software gives every employee individual speaking practice on demand, while traditional training limits speaking time to what scheduled classes allow.
Personalization is the first difference, and it works at three levels.
The first level is feedback. A class moves at the speed of the middle of the room. With AI practice, each person gets feedback on their own speech.
The second level is the material. Some traditional programs add industry-specific examples, but the lessons are still written before anyone at your company is in the room. ELSA Business can build practice from your own content: sales pitches, slide decks, manuals and your website. An employee practices explaining your product using the wording from your own pitch deck, and can roleplay the situations they actually meet at work.
The third level is the learner. The practice can be set to each person's country, industry, role and current proficiency. A sales manager in Jakarta and a support agent in Bangkok can end up with very different sessions, even inside the same program. Because the practice draws on your own material, someone at your company should check that material before learners use it.
Flexibility comes next. An employee can practice for ten minutes between meetings. Nobody has to find a time that suits three time zones.
Then there is capacity. AI-powered individual practice can give more people more chances to speak without instructor capacity growing at the same rate. The 500th learner does not take speaking time away from the first 499. Programs of this size exist: FPT Software has offered ELSA to more than 4,500 employees since 2020.
Where does AI practice fall short, and what does ELSA’s Meeting Companion add?
Practice with software is still practice. A learner who does well in an ELSA session has practiced in a simulated situation, and that does not prove the same result in a client meeting.
ELSA's Meeting Companion looks at that gap directly. In supported video meetings on Zoom and Google Meet, it scores the English an employee actually used on the call: pronunciation, fluency, grammar, vocabulary and intonation. The score is ready within minutes of the call ending, and the gaps it shows can shape that person's next practice.
Seeing how people speak on real calls gives you evidence of their spoken-English performance in actual meetings, which is why it is worth deciding what you will measure.
What did the NashTech program show?
NashTech is a technology services company whose engineers work with clients in English. It used ELSA Business to prepare new hires. In that program, 89% of participants improved proficiency.
Frequently asked questions
How much speaking time does a weekly corporate English class give each employee?
Take a 60-minute class with 12 people. If the instructor said nothing at all, each person would still get only five minutes. Over a year of weekly classes, that adds up to about four hours of speaking.
Why is attendance not enough to show that English training is working?
Attendance tells you people came to class. It does not tell you whether they can speak in the meeting. Researchers who study how training carries over into daily work have found this hard to evaluate, partly because good measurement tools are missing.
How can AI practice add speaking time without adding instructor hours?
AI-powered individual practice can give more people more chances to speak without instructor capacity growing at the same rate. An employee can practice for ten minutes between meetings, and nobody has to find a time that suits three time zones. FPT Software, for example, has offered ELSA to more than 4,500 employees since 2020.
What does ELSA’s Meeting Companion do?
In supported video meetings on Zoom and Google Meet, Meeting Companion scores the English an employee actually used on the call: pronunciation, fluency, grammar, vocabulary and intonation. The gaps it shows can shape that person’s next practice.
Before the next meeting that matters, your teams need more chances to speak. To see how ELSA Business could add individual speaking practice next to your current program, book a demo for ELSA Business.