How a multinational HR group assessed 200 candidates in 2 hours, speaking included

The talent assessment service of a European HR group with clients in more than 30 countries used to wait up to 24 hours for language test results, and its tests never covered speaking. After switching to Talketet, it assessed 200 candidates in 2 hours, with speaking included for every one of them. This is how it happened, and what other talent assessment providers can learn from it.

In Short

• Partner: European HR services group, 30+ countries, 20,000+ client companies
• Use case: Talent assessment service delivering outsourced language assessment for enterprise and institutional clients
• Solution: Talketet AI language assessment (speaking + writing), English, Italian and French

Results:

• 200 candidates assessed in 2 hours
• Time to results cut from 24 hours to 3 minutes
• Speaking assessed for every candidate, at scale

About the Partner

Our partner is one of Europe's largest HR services groups. It operates in more than 30 countries, with thousands of staff and over 20,000 client companies.

Its talent assessment service runs language testing for large enterprises and public institutions. When one of them needs to screen candidates for a hiring program, a training path or an international project, it hands the language assessment to this team.

The Challenge

The partner's clients screen large numbers of candidates across different programs. For each one, they need a fast, reliable answer to a simple question: can this person actually communicate in the language the role requires?

The tests used in the past could not answer it:

• Written only. They never tested speaking, the skill most roles depend on.
• Too long for candidates. A heavy experience at the top of the selection process.
• Too slow to deliver results. Scores took up to 24 hours, while hiring teams waited to move candidates forward.

Enterprise clients were not satisfied. A service meant to speed up hiring was slowing it down.

The Solution

The partner already worked with Talketet. When it looked for a better test, the shortlist was short: Talketet was the only provider that could assess speaking automatically.

The initial doubt was a fair one. Can AI really evaluate speaking and writing? To many organizations, the idea would have sounded too new to trust.

Our partner was willing to look at the evidence instead. Innovation is part of how it works, and the research Talketet carried out with universities and academic institutions showed that the scoring holds up against the CEFR framework. That was enough to move forward.

What the talent assessment service now offers on its projects:

• Speaking and writing assessment, CEFR-aligned and scored by AI
• Results in about 3 minutes after the candidate finishes
• About 30 minutes per candidate, a lighter experience than the old tests
• English, Italian and French, with more languages ready as new projects come in
• Proctoring for institutional clients that need a controlled testing environment

The Results

Speaking is now part of every assessment, and language testing has become a reason clients choose the service rather than a weak spot in it.

Speed at scale. The team assessed 200 candidates in 2 hours, and more than 400 in a single day for one client.

From 24 hours to 3 minutes. Hiring teams no longer wait for scores. Qualified candidates move forward the same day.

New enterprise business. The service helped win new large enterprise accounts, and the activity grew enough to need a dedicated team.

A cost structure that works. AI scoring made high-volume speaking assessment economically viable and improved the profitability of the service.

Lessons for Talent Assessment Providers

1. Clients want speaking, not just grammar. A written test tells you what a candidate knows. A speaking test tells you whether they can do the job. If your offer stops at written tests, you are answering a question your clients are not asking.

2. AI changes the economics of volume. Human-rated speaking tests only scale if costs scale with them. AI scoring lets you absorb peaks of hundreds of candidates a day and still keep healthy margins.

3. Compliance is the entry ticket for institutional clients. Large enterprises and public institutions ask about GDPR, the AI Act and proctoring before they ask about price. If you can answer with documentation, human oversight and a controlled testing environment, you get into conversations others never reach.

Want to offer speaking assessment to your clients at scale? Book a call with the Talketet team