How MetaPret verifies interpreters before the match.

The structural problem

Most interpreter platforms describe their verification process in a single word: “vetted.” That word usually means a self-uploaded resume, a self-paced tutorial, and a hardware compliance check. None of it tests interpretation skill against a live interpretation task.

The result is a market where buyers cannot tell whether their interpreter is good. Agencies charge premium prices for interpreters whose ability the buyer cannot audit. Interpreters with the strongest networks get the bookings, not necessarily the interpreters with the strongest skill. The whole market runs on proxies for quality, not quality itself.

MetaPret inverts this. We test the part that matters, the actual ability to interpret, before any interpreter appears as a match candidate. We publish the methodology because we can. Most platforms cannot.

Layer 1: General verification

When an interpreter applies to join MetaPret, they cannot accept a single engagement until they pass Layer 1, our general skill verification.

Layer 1 has three sections, each measuring a different dimension of interpretation ability:

1. Consecutive interpretation. A 90-second audio segment in one language. The interpreter delivers it in the other language. We score three things: how faithfully the source's meaning and structure are carried (content fidelity), how precise and idiomatic the target-language rendering is (target-language fluency), and how clearly it's delivered (delivery).
2. Listening comprehension. A longer audio segment with embedded ambiguity: a hedging phrase, an idiom, a numeric range with modality. We score whether the interpreter catches what the speaker actually said versus what an inattentive interpreter would assume.
3. Domain expression. Specialized phrases pulled from business, legal, medical, and financial contexts. This isn't memorization. It tests whether the interpreter can render specialized concepts naturally and precisely.

Every section is scored by two evaluator-interpreters who work independently. They don't see each other's scores. A third reviewer settles significant divergence. An AI Monitor tracks scoring patterns over time to flag drift in either direction.

Only interpreters who pass Layer 1 enter the MetaPret network. The bar is set high enough that not everyone who applies qualifies.

Layer 2: Per-engagement verification

This is the layer most platforms do not have. And it is the one that matters most for the client.

When a client submits a request, they describe what the engagement is. Medical conference. M&A negotiation. Factory tour. Executive dinner. Cross-border earnings call. Each of these is a different domain with different vocabulary, different stakes, different stylistic norms.

MetaPret generates a Layer 2 test specifically for that engagement. The medical conference test is not the M&A test, is not the factory tour test. Different content. Different scoring weight. Different difficulty.

Only interpreters who pass Layer 2 for that specific engagement become candidates for it. An interpreter who is excellent in IR calls may not pass the medical Layer 2. An interpreter who is excellent in factory tours may not pass the legal Layer 2. We do not match by self-asserted expertise. We match by tested expertise, tested for this engagement.

This is the part of verification that compounds over time. Every Layer 2 test is also a calibration data point. An interpreter's accumulated Layer 2 history becomes evidence of where their real strengths are.

The 15-point rubric

Both layers use the same rubric. Three dimensions, up to 5 points each, 15 total.

A score of 13 or higher on a Layer 2 test means the interpreter can be matched to that engagement. Below 13, we do not match, even if Layer 1 was strong.

The rubric is the same rubric we use internally. We do not publish a sanitized version for marketing purposes. It is the operational document.

What you see as a client

You do not see the rubric, the scoring breakdown, or the evaluator notes. Those are internal. You see one thing: a list of verified interpreters who passed Layer 2 for your engagement, sorted by fit.

You pick one. We confirm. The match happens.

No interview process. No resume guessing. No paying for a name and hoping the work matches. Verification is done before the match exists. That is the entire model.

If the match does not meet your expectations, we want to hear about it and will work with you to resolve it. Contact cs@metapret.net.

Frequently asked questions

Submit a request to see the score sheet.

Verified candidates take the Layer 2 test for your engagement. You compare the scores and pick.