How to Build a Corporate Language Training Program From Scratch

Nothing exists yet. There is no budget line, no vendor, no list of who needs what. Somebody in an operations review said the words "we have a language problem," a VP nodded, and now the file is yours. Building a corporate language training program from scratch is a design job before it is a purchasing job, and it comes down to six decisions: which business problem you are solving, exactly who needs language, what level counts as finished, what that costs, who delivers it, and what you will report. Put those six answers on one page and everything after is scheduling.

Most first attempts skip straight to demos. A provider shows a platform, the platform looks good, licenses get bought for anyone who raises a hand, and eleven months later the renewal conversation stalls because nobody can say what changed. The design work below is what prevents that. It usually takes six to ten weeks and costs almost nothing except meetings. Once it is done, the actual launch is covered in our guide to corporate language training program implementation, which picks up the day your contract is signed.

What does building a language training program from scratch actually involve?

Two separate phases, and confusing them is the most common early mistake. Design decides what the program is for and who it serves. Rollout decides how it runs week to week. You cannot compress design into a vendor call, because the vendor has no way of knowing which of your conversations are breaking.

Here is the sequence that holds up across company sizes:

Run them in that order. Skipping step one and starting at step five is how companies end up paying for licenses that 40 percent of learners never open. For the wider strategic picture, including how programs are positioned to leadership, start with our pillar guide to corporate language training.

How do you run a language needs assessment?

Use the three-level structure that federal HR practice and most L&D coursework are built on. The Office of Personnel Management frames needs assessment as organizational, occupational, and individual analysis. Applied to language, that translates cleanly.

Organizational analysis asks whether language is genuinely the constraint. It often is not. If a bilingual customer walks out because the wait was 40 minutes, Spanish training solves nothing. Look at where the business already spends money working around language: interpreter invoices, translation vendor spend, escalation and transfer rates, rework on documents, safety incidents involving instruction comprehension, turnover in mixed-language teams. Those line items are your evidence, and they are also your comparison baseline later.

Occupational analysis is where language programs are won or lost. Go role by role and write down the actual conversations the job contains. Not topics, conversations. A shift supervisor delivering a corrective instruction. A dispatcher confirming an address under time pressure. A nurse explaining discharge medication. Pull real artifacts while you do it: call recordings, ticket transcripts, incident reports, the documents that keep getting sent for translation. Twenty specific conversations beats a curriculum outline every time, and it is the raw material a good provider will ask you for anyway.

Individual analysis measures where people are now. Do this with a real assessment that includes speaking and listening, not a self-rating on a form. Self-reports skew in both directions: employees who studied a language in school overestimate badly, and heritage speakers who use the language daily at home underestimate because they never learned to write it. Our explainer on CEFR levels covers what each band means in practice.

Budget three to five weeks for the assessment. Interview supervisors rather than surveying employees at this stage. Supervisors know which conversations get escalated, and employees are reluctant to write down that they did not understand something.

Which employees and which languages make the first cut?

Narrow harder than feels comfortable. The pressure to be inclusive at launch produces cohorts of 200 people at eleven proficiency levels across six time zones, which cannot be scheduled and cannot be measured. A first cohort of 15 to 40 employees in two or three levels is small enough to run properly and large enough to prove something.

Four criteria do the sorting:

• Contact frequency. How often does this role hit the language barrier? Daily beats monthly.
• Consequence. What happens when the conversation fails? A lost sale, a safety event, and a mild inconvenience are not the same priority.
• Distance to target. Someone at A2 reaching B1 is a one-year project. A true beginner reaching B2 is a multi-year commitment, so do not put both in the pilot.
• Schedulability. Can this person reliably attend a fixed weekly session? Shift and field roles need a different delivery model, which our comparison of language training formats works through.

Language choice is usually decided by your workforce and your customers rather than by preference. Roughly one in five workers in the US civilian labor force is foreign born, and closer to one in four in the West, according to the Bureau of Labor Statistics. Census tabulations from the 2017 to 2021 American Community Survey put 22 percent of residents aged five and over speaking a language other than English at home, with 38 percent of that group reporting they speak English less than very well. In most US operations that points at Spanish first, English second for incoming staff, and everything else third.

One caution about direction. Programs designed by headquarters tend to teach headquarters. Budget follows the people who have development plans, and those are rarely the people standing in front of customers. If your needs assessment pointed at the shop floor or the branch counter, the cohort belongs there.

What should the program be accountable for?

Write objectives as tasks a graduate can complete alone, then attach a proficiency level to each one. "Improve Spanish" is unmeasurable. "Complete an account opening, including the fee schedule, without a transfer" is a target a supervisor can verify on a Tuesday.

Then decide in advance which of the four classic evaluation levels you will actually report. The Kirkpatrick Model runs from reaction to learning to behavior to results, and each level is harder to measure than the one before it. Almost every new program reports level one, satisfaction surveys, because it is easy. Almost none reports level three, on-the-job behavior, because it requires deciding what to watch before training starts. Pick your level three metric during design, while you still can. Interpreter minutes per month, escalation rate, transfer rate, incident counts, and time-to-resolution are all defensible. Our guide on how to measure language training ROI covers what to instrument and when.

Set the review date now too. Twelve weeks is too early for a level jump and about right for attendance and behavior signals. Nine to twelve months is when a CEFR reassessment means something.

How do you size a budget before you have one?

Work from hours, not from list prices. LanguageCert publishes guided learning hours per CEFR level: about 95 hours to reach A1, another 95 to A2, then roughly 180 more to B1. Two 60-minute sessions per week yields around 90 hours across a working year once holidays and coverage gaps come out. That arithmetic is what turns a vague request into a proposal a finance partner can evaluate.

Benchmarks help you sanity-check the ask. Training magazine's 2025 Training Industry Report put total US training expenditure at $102.8 billion, an average of $874 per learner, with employees receiving about 40 hours of training per year. Only about 7 percent of training budgets went to outsourcing, averaging $217,178 per organization. ATD's 2025 State of the Industry report found direct learning expenditure of $1,054 per employee, 13.7 formal learning hours per employee, and a cost of $165 per learning hour.

Those numbers carry a useful warning. A language program needing 180 hours to move one CEFR level does not fit inside an average of 13.7 hours per employee per year. Language training is not a normal training line item, and presenting it as one invites a rejection on cost per hour alone. Present it as a capability build with a multi-year path and a first-year milestone. Concrete ranges by delivery model are in our breakdown of corporate language training cost, and the argument for the spend is laid out in the bilingual workforce ROI analysis.

Should you build it in-house or buy it?

Very few companies should build. Hiring instructors, writing curriculum, running placement testing, and managing attendance is a small business inside your business, and it needs a specialist to survive its first vacation season. Building in-house makes sense in three situations: you have hundreds of learners in one language at one site, your terminology is so proprietary that no external curriculum applies, or you already employ credentialed language instructors.

Everyone else buys, and the real decision becomes which vendor and on what terms. A hybrid is common and works well: an external provider handles instruction, assessment, and reporting, while you keep the parts nobody outside can do, which are scheduling around operations, supervisor accountability, and supplying real workplace scenarios.

Do not evaluate providers on a demo. Send the same document to three of them and compare answers, using our language training RFP template as the base. Then apply the questions in our guide on how to evaluate a language training vendor, which covers instructor credentials, class size caps, substitution policy, assessment methodology, and what happens when a cohort falls behind.

Which policies have to exist in writing before day one?

This is the shortest section and the one most often skipped. Six things need a written answer before the first class, because each one becomes an argument later.

1. Is study time paid? Unpaid study is the single most reliable predictor of a cohort collapsing around week five, and it has nothing to do with motivation. If sessions are on the clock, say so in writing, and tell supervisors that coverage is their problem to solve.
2. Who is eligible? Open enrollment sounds generous and produces unmanageable cohorts. Define eligibility by role and by assessed level.
3. Does language skill get paid? If bilingual employees are already acting as unpaid interpreters, training more of them without a differential simply spreads the resentment. Put the skill in the job description and pay for it. Our guide to hiring bilingual talent covers how to structure that alongside recruiting.
4. What happens with assessment results? Placement scores are performance data about employees. Decide who sees them, whether they enter the HR file, and what a low score does not trigger.
5. What is the attendance rule? Name the threshold and the consequence. Quiet attrition with no rule is worse than a stated one.
6. Are you accidentally writing a language restriction? Teaching employees a language and restricting which language they may speak are opposite policies, and companies drift into the second while pursuing the first. Under EEOC guidance on national origin discrimination, a blanket English-only rule covering all workplace times is presumed to violate Title VII. Narrower rules require business necessity and advance notice.

What does the design phase look like week by week?

Ten weeks, part-time, one owner. The schedule below assumes an HR or L&D lead spending roughly four hours a week on it.

At the end of week ten you have a fundable proposal rather than a request. Then move to the rollout mechanics, cohort scheduling, protected learning time, and reporting cadence, which are covered in our implementation guide.

What derails a program built from scratch?

• Fluency as the goal. Nobody reaches fluency in a corporate program, so promising it guarantees a failed review at month twelve. Task completion at a named level is the goal.
• No baseline. If you did not record interpreter spend and escalation rates before launch, you will have no way to show improvement afterward. Capture the baseline in week one, not week thirty.
• Buying a platform instead of designing a program. Self-serve licenses have their place for motivated individuals. They do not move a cohort of shift workers to B1.
• Mixed-level cohorts. Putting A1 and B1 learners in one class loses both. Group by assessed level even when it splits a team.
• No supervisor on the hook. Attendance is a scheduling decision made by line managers. If no manager is accountable, sessions lose to coverage every time.
• Sponsor with no operational stake. A sponsor whose numbers are not affected by the language problem will not defend the budget in a tight quarter.
• Design in isolation from operations. A schedule built without the staffing planner is a schedule that will be overridden in week three.

The design phase also has a quiet advantage worth naming. Career progression is what employees say motivates them to learn, and organizations that treat development as a real system rather than a perk retain people better, per LinkedIn's Workplace Learning Report. A language program with named objectives, a level target, and a documented path reads to employees as investment. The same budget spent on licenses nobody assigned reads as a benefits page nobody visits.

Build your language program with Edlingo

Edlingo works with employers at exactly this stage, before there is a program to run. We assess every candidate against CEFR with speaking and listening included, run supervisor interviews to build the conversation inventory, group cohorts by level instead of by department, write objectives your line managers can verify, and report attendance and progress per session from the first week. Start with one site, one cohort, and one measurable target. See how delivery works on our for employers page, or go straight to corporate Spanish training and corporate English training for the two programs US employers request most.

Scope Your First Cohort →

Working through the whole decision? Read the pillar guide to corporate language training, price it with our cost breakdown, then run the launch with our implementation plan. Ready to talk? Contact us and we will build the cohort list with you.