Choose corporate language training by starting from the work situations staff must handle. Then measure participation, practice, job-task performance, use at work and one or two business results.
In short
- Start corporate language training with a needs analysis: which roles use the language, in which situations, at what level.
- Many programmes mix formats; compare them on practice per person and the evidence they produce.
- Measure in layers: participation and practice are inputs; job-task performance, use at work and business measures are outcomes.
- Calculate ROI only when you can isolate the training's effect and convert it to money credibly.
- Pilot for 90 days with a baseline task and success criteria agreed in advance.
Where should corporate language training start?
Corporate language training should start with a needs analysis, not a vendor demo. Corporate language training is a programme that helps employees use a language in their actual work: running a meeting, handling a customer call, writing an incident report. The goal is reliable performance in defined job situations, not general fluency.
A needs analysis is a short study of who needs the language, in which situations, how often, with what stakes and at what level. Needs vary by industry. In the 2016 English at Work survey by Cambridge English and QS of 5,373 employers in 38 countries, reading was rated the most important English skill in 11 industries and speaking in 9, and every industry reported a gap between the English skills required and those available.
Map roles, situations and levels
A one-page map turns "our people need better English" into goals you can train and measure.
| Role | Situations | Skill focus | Example goal |
|---|---|---|---|
| Customer service | Rebooking calls, complaint emails | Speaking, listening, short writing | Handles a rebooking call without passing it on |
| Engineers | Partner calls, incident reports | Speaking, technical writing | Explains a fault and next steps on a partner call |
| Managers | Meetings, presentations | Speaking, listening | Runs the weekly regional meeting in English |
| Finance and back office | Contracts, supplier emails | Reading, writing | Drafts supplier emails that need no rewrite |
Describe current levels with the CEFR, the Council of Europe's six-level scale from A1 to C2, defined through can-do descriptors. Record a short job task as a baseline you can repeat later. For needs-analysis questions and role-play cards, see our guide to ESP course design.
Set realistic time frames
Cambridge English estimates that it takes about 200 guided learning hours (lessons and supervised study) to move up one CEFR level. Cambridge calls this a guideline only; learning background, intensity, age and practice outside lessons all change it. At two hours a week, a year of training adds roughly 100 hours. So set goals as job tasks within the programme's time frame ("handles a rebooking call by month three"), not as "one CEFR level in three months".
Which corporate language training formats fit which needs?
Corporate language training formats differ mainly in how much speaking time each person gets, how flexible the schedule is, and what evidence they produce for measurement. Many programmes combine two or three of the six below.
| Format | Strengths | Limits | How to measure |
|---|---|---|---|
| In-house group classes | Shared goals; trainers can use company material | Little speaking time per person; hard across shifts and time zones | Attendance; job task at the start and end |
| External courses | Structured syllabus; often lead to a certificate | Generic content; fixed timetable or travel | Attendance, course results, manager check on use at work |
| 1:1 online tutoring | Fully tailored; suits senior staff and one-off events | Scheduling and cost per head; quality depends on the tutor | Sessions held, tutor reports, job task before and after |
| Self-paced apps | Flexible; useful for vocabulary, grammar and routine | Little speaking, writing or feedback; drop-off without goals | Active users and time on task (inputs only) |
| AI speaking practice | Rehearses job conversations between sessions, with instant feedback | Feedback quality varies; check the data policy; no human rapport | Practice volume, scores on repeated role-plays, manager observation |
| Blended | Live teaching for complex communication, practice in between | Needs coordination and one clear owner | All of the above, in one framework |
Language schools such as Berlitz run live online classes for company groups, and corporate platforms such as Speexx combine CEFR-based assessment, digital microlearning, and support from AI and live coaches. Compare providers on the same questions, using the RFP checklist below.
How do you measure the results of language training?
Measure the results of language training in layers. Inputs (participation and practice) show that the programme is running; outcomes (job-task performance, use at work and business measures) show whether it is working. Completion shows participation, not ability, so never report it as the result.
The Kirkpatrick Model, created by Donald Kirkpatrick in the 1950s, evaluates training at four levels: Reaction (is it relevant to the job?), Learning, Behaviour (are critical behaviours applied at work?) and Results (targeted organisational outcomes). The Phillips ROI Methodology builds on these levels, treats inputs as Level 0 and adds Level 5 for return on investment. The framework below maps language-specific measures to both.
| Measure | Level | How to collect | When | Watch out for |
|---|---|---|---|---|
| Participation | Input (Phillips Level 0) | Enrolment, weekly active learners, sessions attended | Weekly | Shows engagement, not ability |
| Practice volume | Input (Phillips Level 0) | Speaking minutes, role-plays completed, writing tasks submitted | Weekly or monthly | Volume is not quality; pair it with task performance |
| Task performance | Kirkpatrick Level 2 | The same job task (for example a customer call role-play and a work email), scored with a rubric | Baseline, mid-point, end | Same task and rubric each time; double-mark a sample |
| Manager observation | Kirkpatrick Level 3 | A checklist of three to five critical behaviours agreed with managers, plus work samples | One and three months after a module | Note barriers, such as no chance to use the language |
| Business outcome | Kirkpatrick Level 4 | One or two existing measures the language plausibly affects, such as satisfaction scores for English-language contacts, handovers or rework | Quarterly | Other factors matter; compare with a similar team or the prior trend |
For Level 1, run a short pulse survey early, not only at the end; Kirkpatrick Partners recommend pulse checks during launch and treat job relevance as the key reaction measure. To score speaking tasks consistently, use an analytic rubric such as our ESL speaking assessment rubric.
Can you calculate the ROI of language training?
You can calculate the ROI of language training only when you can isolate the training's effect on a business measure and convert that effect to money credibly. In the Phillips ROI Methodology, ROI (%) equals net programme benefits divided by programme costs, multiplied by 100; the benefit-cost ratio equals benefits divided by costs (Phillips and Phillips, ROI Basics, ATD Press, 2019).
Three of its guiding principles matter most for language programmes:
- Isolate the effect. Use at least one method, such as a control group, trend-line analysis or estimates adjusted for error (ROI Institute). A staggered roll-out, where a similar team starts a quarter later, gives you a comparison group without a separate study.
- Fully load the costs. Include provider fees and licences, employees' time in training, managers' and administrators' time, and content preparation.
- Stay conservative. Choose the most conservative option in each calculation, and count only the first year of benefits for short-term solutions.
Don't force an ROI figure when the group is small, when the business measure moves for many reasons at once (a new system, a reorganisation), or when converting it to money depends on guesswork. Report Level 3 and Level 4 results instead, and list intangible benefits: measures purposely not converted to money, in Phillips's terms, such as fewer calls passed to the same few bilingual colleagues. Kirkpatrick Partners take a different line and favour "contributive ROI", which reports the share of improvement reasonably linked to training rather than isolating training as the sole cause.
How do you run a 90-day language training pilot?
A 90-day pilot tests a corporate language training programme with one team before a wider roll-out, against success criteria agreed in advance. Ninety days will not move people up a CEFR level: at two to three hours a week, a pilot adds roughly 25 to 40 hours against Cambridge's guideline of about 200 per level. Judge the pilot on job tasks, practice and manager observation instead.
| Phase | What happens | Output |
|---|---|---|
| Days 1–14: prepare | Choose one team with a clear need; pick two or three target situations; record a baseline job task; agree success criteria and the manager checklist | Baseline scores; written success criteria |
| Days 15–45: start | Training begins; review participation and practice weekly; run a pulse survey in week three | Input data; Level 1 feedback |
| Days 46–75: check | Repeat the job task as a mid-point check; managers check the critical behaviours; remove barriers such as blocked practice time | Mid-point scores; list of barriers |
| Days 76–90: decide | Repeat the baseline task with the same rubric; collect manager observations; compare with the agreed criteria | Short report; decision to scale, change or stop |
Write the success criteria down before training starts, such as a minimum share of weekly active learners and a target rubric gain. Criteria agreed afterwards tend to fit whatever happened.
What should a language training RFP include?
A language training RFP (request for proposal) should make every provider answer the same questions on needs analysis, assessment, practice, content control, reporting, data protection and terms. Copy this 13-point checklist into your request.
- Needs analysis: Will you analyse roles and situations with us first? Can practice use our own materials, such as manuals and call scripts?
- Assessment: How do you measure progress from start to end? Is it CEFR-aligned and based on job tasks? Who scores it (a person, AI or both)? Can we see a sample report?
- Productive practice: How many minutes of speaking and writing does each learner produce per week, and what feedback do they get?
- Formats and access: Which mix of live teaching, self-study and AI practice do you offer? Does it work across shifts, time zones and mobile?
- Content control: Who reviews AI-generated content? Can our trainers edit and approve material before employees see it?
- Reporting: What do HR and managers see? Can we export the data?
- Data protection: Where is data stored, and under which law (for example GDPR)? How long is it kept? Are voice recordings used to train models, and can we opt out?
- Languages and accessibility: Which learning and interface languages and accessibility features do you offer?
- Integration and security: Do you support single sign-on and links to our LMS or HR systems? Which security certifications do you hold, if any?
- Pilot: Will you run a 90-day pilot with success criteria agreed in advance?
- Quality standards: Do you follow a recognised standard such as ISO 29991:2020 for language-learning services? Ask it as a question, not a requirement.
- References: Can we speak to a client of similar industry and size?
- Commercial terms: What are the pricing model (per seat or per active user), minimums, contract length and cancellation terms?
When AvoLingo fits, and when it doesn't
AvoLingo, an AI language-learning service by AvoLabs, adds speaking and writing practice between training sessions. It works alongside trainers, tutors and existing programmes rather than replacing them.
AvoLingo fits when
- You want practice built from your own documents and situations. Product sheets, work manuals or a described situation become role-plays, sentence practice and work vocabulary for each role. Trainers edit and preview drafts, and staff see nothing until a trainer assigns it.
- Staff need to rehearse real conversations. In a role-play, the AI plays the customer or colleague; the employee works through set tasks, can check a hint, and reads an explanation of their sentence after replying.
- You want instant feedback between sessions, such as pronunciation scores on sentences and sentence-by-sentence writing feedback. Trainers review the records afterwards.
- L&D needs participation and progress data. Admins see individual progress, team participation, practice results and the phrases a person found hard, as the AvoLingo for teams page shows.
AvoLingo doesn't fit when
- Live human conversation is the main need, for example executive coaching or negotiation preparation. AvoLingo has no live sessions with human tutors, so use a tutor or trainer for that.
- You need a placement test or a certified score. Use an external test as the benchmark.
- Procurement requires specific integrations or security certifications. Confirm these through our contact page before a pilot rather than assuming them.
- Nobody will protect time for practice. No practice tool works without it.
FAQ
How do you measure the ROI of language training?
Measure in stages (participation, practice, job-task performance, use at work, business measures). Apply the Phillips ROI formula only when you can isolate the training's effect and convert it to money; otherwise, report outcomes and intangible benefits.
How long does it take employees to improve their English?
Cambridge English estimates about 200 guided learning hours per CEFR level, as a guideline only. Set goals as work tasks within the programme's time frame, not as a promised level change.
Is completion rate a good measure of language training?
No. Completion shows participation, not ability. Pair it with a repeated job-task assessment and manager observation.
What should English training for employees include?
English training for employees should include a needs analysis by role, enough speaking and writing practice for each person, a job-task assessment at the start and end, manager involvement, and reporting that HR can act on.
Sources
- Cambridge English and QS (2016), English at Work: executive summary
- Cambridge English, Guided learning hours
- Council of Europe, The CEFR levels
- Kirkpatrick Partners, The Kirkpatrick Model
- Patricia Pulliam Phillips and Jack J. Phillips (2019), ROI Basics, chapter 1, ATD Press
- ROI Institute, The ROI Methodology
- ISO 29991:2020, Language-learning services — Requirements
Written by the AvoLingo team. We used AI to help research and draft this article; sources are linked in the text.
Back to all articles