Operations hiring across several markets drifts question by question. Each market adapts the first round to what it believes the role needs. The adaptations are all reasonable on their own, but the shortlists stop being comparable at group level. Nobody notices the drift until two markets compare notes and find they were never asking the same thing. The first round is also the highest-volume stage, so a small drift there compounds fastest.

Five markets, five first rounds

Rundahl hires operations staff in five countries. The first-round questions had been written locally, translated locally and revised locally. By the time anyone compared them, they had been five different documents for years.

Group hiring could see how many people each market shortlisted. It could not see what each market had actually asked that week, or how closely one market’s bar matched another’s. A shortlist rate on its own answers the wrong question.

A candidate who would have cleared the first round in one country might have been declined on a question another market had quietly dropped two years earlier. Nobody at group level had any way to notice, because the questions themselves were never reported upward, only the outcomes.

One set of questions, five languages

The first round now runs on Yardstick Agents in the applicant’s own language, against the same questions everywhere. The answers come back into one structured record that group hiring can read across markets.

Translation is reviewed by a fluent speaker in each market before the round opens. A question that drifts in translation reintroduces the exact problem the shared round was meant to remove. A literal translation that reads oddly in one language is rewritten rather than left to stand. Recruiters in each market now spend less time defending their own version of the round, and more time discussing borderline candidates.

Local requirements stay local

Where a market has a genuine legal or regulatory requirement, it is added as a named criterion for that market alone. The difference is declared on the record, instead of buried in the wording of a supposedly shared question. A market lead still signs off on every shortlist; the shared questions inform that read, they do not replace it.

Results across the group

All 5 markets now score candidates against one comparable rubric in the first round. Group hiring reads the outcome as a single set of results, rather than five separate reports that never quite lined up.

The spread in shortlist rates across markets has narrowed by 52%. That is evidence the gap was mostly in the question, not in the candidates each market happened to attract. Two markets that once looked very different now shortlist at close to the same rate.

100% of translated questions are reviewed by a local speaker before a round opens. A market can still flag a question that reads badly once real candidates start answering it, and two questions have been rewritten this way since launch.

What they are doing next

Rundahl is considering whether the same shared first round should extend to its two smaller markets, which currently still run their own separate process end to end.