A regional company with offices in Singapore, Kuala Lumpur, Jakarta and Hong Kong rarely has one hiring process. It has four, sometimes more, because each country has its own rules, each legal entity has its own approvers, and each office has built its own habits. Automation can either tidy this up or make it worse. Automating one global process that ignores country differences pushes recruiters back to spreadsheets. Automating four unrelated processes leaves headquarters with no consistent view.
The answer is to design the workflow in two layers and let rules switch the country-specific steps on and off. This guide shows how, with trigger rules for Singapore, Hong Kong, Malaysia and Indonesia, approval chains by legal entity, and a configuration table you can adapt. It expands the "local channel and workflow fit" factor in our guide to recruitment automation software. For choosing a platform, permissions and rollout order, see our separate guide to AI recruiting for multi-country hiring.
Why one global process breaks in APAC
Three differences break a single process:
- Rules before hiring. Some countries require steps before an offer, such as advertising a role locally before hiring a foreign professional.
- Rules about candidate data. What you may collect, when, and for how long differs by market.
- Approvers. A role in the Malaysian entity may need sign-off from a country manager and regional finance; the same role in Singapore may not.
When the workflow ignores these, recruiters fix them by hand, and the fixes are where mistakes and delays happen.
Split the workflow into a shared layer and a country layer
The shared layer is the same everywhere: stage names, scorecards, interview structure, reporting definitions and AI screening criteria for each role family. Keeping this layer identical lets headquarters compare markets fairly.
The country layer holds only what must differ: legal steps, data rules, approval chains, offer templates, job boards and candidate channels. Each item is a rule that fires when a job belongs to a given country or legal entity.
The aim is to keep the country layer as small as possible. If a step differs only because one office always did it that way, move it into the shared layer.
Country trigger rules
The examples below show how country rules turn into automated steps. Rules change, so confirm current requirements with your legal team. This is general information, not legal advice.
Singapore: advertise before an Employment Pass
Under the Fair Consideration Framework, employers applying for an Employment Pass "must first advertise the job on MyCareersFuture", and the advertisement "must be open for at least 14 consecutive days" (Ministry of Manpower). MOM also says you should not make an offer during that period. Exemptions include companies with fewer than 10 employees and roles with a fixed monthly salary of $22,500 and above.
Automated rule: when a Singapore role is flagged as possibly needing an Employment Pass, post it to MyCareersFuture, start a 14-day timer, and block the offer stage until the timer ends, unless an exemption is recorded.
Hong Kong: collect ID data late, delete rejected data on time
The PCPD's Code of Practice on Human Resource Management says employers should not collect a copy of a candidate's Hong Kong identity card until the candidate has accepted an offer, and that data on unsuccessful applicants may be kept for up to two years from rejection.
Automated rules: hide the ID upload field until the "offer accepted" stage for Hong Kong roles; schedule deletion of rejected candidates' data two years after rejection, unless the candidate agreed to longer retention.
Malaysia: bilingual notices and breach timelines
Malaysia's amended Personal Data Protection Act requires privacy notices in both Malay and English. Breaches must be reported to the Personal Data Protection Commissioner within 72 hours and to affected individuals within seven days (data breach notification guidelines).
Automated rules: show the bilingual privacy notice on Malaysian application forms and chat entry points; route any suspected data incident involving Malaysian candidates to the incident owner with the 72-hour clock visible.
Indonesia: impact assessments and fast breach notice
Indonesia's Personal Data Protection Law (Law No. 27 of 2022) lists automated decision-making among high-risk processing that needs an impact assessment (Article 34), and requires written breach notice within 3 x 24 hours (Article 46).
Automated rules: before AI screening is switched on for Indonesian roles, require a completed impact assessment in the job setup; keep a recruiter review step before any rejection; route incidents with the 72-hour clock.
For the full data rules in each market, see our guide to candidate data compliance in APAC.
Approval chains by legal entity
Approvals are where regional hiring slows most. Set them by legal entity and role level, not by recruiter habit:
- Requisition approval: hiring manager, then country head, then regional finance for new headcount.
- Offer approval: hiring manager and HR business partner, plus regional head for offers above a salary band.
- Exceptions: an offer outside the approved range automatically adds a compensation approver.
Run approvals in the tool managers already use. If your managers work in Lark or Microsoft Teams, approval requests should appear there, with the candidate summary attached, and the decision should write back to the ATS. A manager who has to log in to a separate system to approve will approve late.
For example, an offer for a regional role hired through the Malaysian entity might route to the hiring manager in Kuala Lumpur, the HR business partner, and, because the salary sits above the band, the regional head in Singapore. Each approver gets a card in Lark with the candidate summary and the proposed terms, and the offer letter is generated only when the last approval arrives.
A country-by-stage configuration table
| Stage | Shared layer | Singapore | Hong Kong | Malaysia | Indonesia |
|---|---|---|---|---|---|
| Job posting | Careers site | MyCareersFuture (14-day timer for EP roles), JobStreet | JobsDB | JobStreet | JobStreet |
| Application | Standard form | PDPA notice | Notice of purpose | Malay and English notice | Notice and consent |
| Screening | Same scorecard | Impact assessment before AI screening | |||
| Offer | Same structure | Blocked until FCF timer ends | ID copy only after acceptance | ||
| Rejected data | Retention rule | Keep while purpose remains | Delete after 2 years | Per notice | Per notice |
| Approvals | By level | By entity | By entity | By entity | By entity |
Rolling out country rules
Do not switch on every country at once. A practical order:
- Agree the shared layer first. Stage names, scorecards and reporting definitions, signed off by every country's hiring lead.
- Pilot one country. Usually the headquarters market. Configure its country rules and run real roles for a month.
- Add one country at a time. Copy the setup, add the local rules from the table, and test each rule with a sample candidate before going live.
- Review rules every quarter. Laws, thresholds and job boards change. Give each country rule an owner who checks it against the official source.
Common mistakes
- Building rules from habit, not requirements. Every country rule should trace back to a law, a contract or a clear business reason.
- Letting approvals multiply. Each extra approver adds days. Review approval chains when you review rules.
- Hard-coding thresholds. Values such as salary thresholds change. Store them as settings, not inside workflow logic.
- Forgetting retention. Rules that collect data need matching rules that delete it.
Customer example: Traveloka
Traveloka hires across 9 APAC markets. On Moka AI, it connects WhatsApp, Lark and HackerRank in one platform, and designs candidate outreach templates around each market's data protection rules. The team has created more than 4,500 interviews in the system. Read the Traveloka story.
Next step
List every step that differs by country in your current process, then sort each into the shared or country layer. Anything you cannot justify as a legal or business requirement goes into the shared layer. To see Lark approvals, WhatsApp messaging and AI screening on one candidate record, request a Moka AI demo, or read our guide to recruitment automation software.

Author
Moka AI Editorial Team
The Moka AI Editorial Team creates insightful, practical, and engaging content on HR, recruitment, and AI innovation. We explore emerging trends, share industry insights, and turn complex ideas into actionable knowledge to help HR professionals navigate the evolving world of work.



