How to measure the ROI of corporate English training

Short answer: You cannot measure ROI on language training after the fact. It has to be designed in before the program starts, and it needs three things: a baseline taken before anyone is taught anything, a small number of metrics tied to work the company already cares about, and a repeat measurement on a fixed schedule using the same instrument. Attendance, hours delivered and satisfaction surveys are activity data, not outcome data, and a finance director will treat them as such. The rest of this article is how to set the three up.

Why most language training cannot be evaluated at all

Ask an HR lead how last year's English program went and you usually get one of three answers. People liked it. Attendance was good. The provider sent a report.

None of those survive a serious question from the person holding the budget. "People liked it" measures the teacher. "Attendance was good" measures the calendar. A provider's report, when it is only hours delivered and units completed, measures the provider.

These numbers are not wrong, they just describe the training rather than the change it was bought to produce. When the renewal conversation comes there is nothing to point at, so the program gets defended as a staff benefit rather than an investment, and benefits are the first thing cut when budgets tighten. If the only evidence is that the sessions happened, what the company bought was sessions.

The three things that have to exist before day one

1. A baseline, taken before any teaching

This is the step most often skipped, and it cannot be recovered later. Once training has started there is no way to go back and establish where people were, so any improvement claimed afterwards is an assertion rather than a measurement.

A baseline does not need to be elaborate. It needs to be consistent, quick enough that people actually complete it, and repeatable in exactly the same form later. At Achivr this is the Communication Audit, a 15 minute assessment scoring five dimensions: fluency, structure, vocabulary, confidence and professional register. What matters is less which instrument you use and more that you use the same one at every checkpoint. An assessment that changes between measurements produces numbers that cannot be compared, which is the same as having no numbers.

Run it on everyone in scope, including the people you expect to score well. A baseline with gaps invites the objection that you measured only the people who improved.

2. Metrics the business already recognises

The strongest measure of a language program is rarely a language score. It is something the company was already tracking and already unhappy about.

Examples that hold up in a review:

• Time to produce a client-facing document in English. If a proposal takes two drafts and a manager's rewrite, that is measurable in hours and those hours have a cost.
• Meetings that need a translator, a summary afterwards, or a second meeting. Count them before and after.
• Escalations caused by misunderstanding rather than by the underlying issue. Support and account teams usually know these by name.
• Internal mobility. How many people were passed over for a role that required working in English, and how many became eligible.
• Onboarding time for international hires when the receiving team can brief them directly.

Pick two. Not eight. Two metrics that someone outside HR already looks at will do more for a renewal than a dashboard of twelve that only L&D reads.

3. A fixed measurement rhythm

Improvement in communication is not linear and it is not fast. Measuring monthly produces noise that makes a real trend look like a flat line with spikes. Measuring once at the end produces a single data point that proves nothing about the direction of travel.

Eight weeks is a workable cycle. It is long enough for a change in how someone handles a meeting to become visible, and short enough that a program which is not working can be corrected before the budget year ends. Achivr engagements run in eight week blocks for this reason: each block opens with a diagnostic, has a mid point checkpoint, and closes with a report on the same dimensions as the last one.

What a defensible report actually contains

When the program is reviewed, the report should answer four questions in order.

Where did we start? The baseline, by team and by department, not just a company average. Averages hide the two departments that need attention.

What changed, on the same scale? Per dimension movement, so it is clear whether the gain is in confidence, in structure, or in professional register. "Overall improvement" is not actionable. "Structure improved across the sales team while written register did not" tells you what to do next.

What did it cost, per person who changed? Total program cost divided by participants who moved, not by participants enrolled. It is a less flattering number than cost per head, and a more credible one.

What did the business get? The two metrics you chose in step two, before and after.

A report that answers these four questions is usually about a page long.

Three mistakes that make ROI impossible

Training everyone at once. A pilot group of five to fifteen people produces a comparison you can point at. A company wide rollout in year one produces a large invoice and no control.

Changing the assessment mid program. This is normally done with good intentions, when a better instrument becomes available, but it destroys comparability with the baseline. Finish the cycle first, then change.

Letting the provider define success. If the only reporting you receive is generated by the company being paid, the incentives are obvious to everyone in the room. Agree the metrics in the contract, before the first session.

Where to start if the program is already running

The baseline is gone, but the program is not. Take the measurement now and treat it as the start of cycle one rather than the middle of an unmeasured engagement. You will not be able to claim whatever was gained in the first period, which is uncomfortable to explain, but everything after that point is evidenced. It is a reasonable trade to make once.

We run corporate English programs in eight week cycles for the reasons above. If that is useful: corporate program, case studies.