A company hiring in one country has a process. A company hiring in eight usually has one of two things: a single process designed for headquarters that local teams quietly work around, or eight local processes that cannot be compared, reported on, or improved.
The workable answer is a deliberate split: a standard spine, with defined local variation.
What to standardise
These should be identical everywhere, and standardising them is what makes the rest possible.
- Stage names and their meaning. If "first interview" means a screening call in one market and a panel in another, no cross-market number means anything.
- Evaluation criteria per role family. The bar for a mid-level engineer should not depend on which office is hiring. Criteria travel; the questions used to assess them may not.
- The decision rule. Same threshold, same veto structure, everywhere.
- Data model. One system, one candidate record shape, one set of source categories. This is the constraint that makes everything else reportable.
- Disclosure and privacy baseline. Hold every market to the strictest standard you operate under rather than maintaining eight notices — see the notice structure.
What to localise
These genuinely differ, and forcing uniformity creates workarounds rather than consistency.
Channels
The dominant job board differs by market, sometimes sharply. A distribution strategy built around one global aggregator will underperform badly in markets where a national board holds the audience.
Language
Postings and candidate communication in the local language, even where the working language is English — unless the role genuinely requires English from day one, in which case say so and test it.
Notice periods and start dates
Where statutory notice runs to one, two or three months, offer-to-start planning changes completely. Time to fill is not comparable across markets with different notice regimes unless you measure to acceptance rather than to start.
Compensation bands
Local bands, and increasingly local pay transparency obligations. Publishing ranges is required in a growing number of jurisdictions and worth doing generally.
Background and right-to-work checks
Different documents, different providers, different legality. Some checks that are routine in one market are unlawful in another.
The three constraints that catch people
1. Data residency and transfers
Candidate data moving between countries needs a lawful basis, documented safeguards and disclosure. If your hiring system is hosted in one region and your recruiters sit in another, that is a transfer — including support access. This belongs in the tooling decision rather than in a later privacy review: where is your candidate data stored.
2. Access control
A recruiter in one market should not be able to browse candidate records from another without a reason. This is a data minimisation requirement and a practical one — the wider the access, the harder every subsequent question becomes.
3. Local approval chains
Offer approval that takes a day in one market may take a week in another, for entirely legitimate structural reasons. Building a single expected timeline across markets produces reporting that flags healthy processes as failures.
Making the numbers comparable
Three rules make cross-market reporting mean something:
- Measure to acceptance, not to start. Notice periods otherwise dominate the comparison and tell you nothing about your process.
- Segment by role family before market. Market differences are often role-mix differences wearing a flag.
- Report stage duration alongside conversion. Two markets with identical conversion can have completely different elapsed times, and only one of those is a problem you can fix.
Country breakdowns of these figures are the section missing from every major published benchmark — see EMEA mid-market hiring benchmarks.
The operating model that works
A small central function owns the spine — stages, criteria, the system, reporting, and the compliance baseline. Local teams own execution, channels, language and relationships.
The most common failure is the opposite: central owns execution and local owns nothing. Recruiters who cannot adapt their channel mix or their language to their market will route around the system, and then the data problem returns.
For running this in one system with local variation, see pipelines and multi-market hiring.