What is Zipang's screening funnel?

Every Zipang candidate passes a 5-gate funnel: (1) CV intake with OCR parsing, (2) CEFR-aligned English proficiency check, (3) role-specific quiz, (4) structured 15–30 minute video interview, and (5) a paid trial task scored against the client's published KPI. Only candidates who pass every gate are placed into a client program.

What is the pass rate at each gate?

Aggregate pass-rates across 2015–2026 programs run by PT Lima Cakar Bumi are approximately: Gate 1 CV intake 100% (intake), Gate 2 English 70%, Gate 3 Role quiz 50%, Gate 4 Interview 40%, Gate 5 Trial task 35%. Cohort-specific pass-rates are available on request after the NDA is signed.

How is accuracy measured in production?

Accuracy is measured against client-graded gold sets and tracked at the per-task level. The Transperfect Dataforce interpolation program sustains 90%+ accuracy across 20 full-time production agents. The France retail AI annotation program tracks 3.4M production tasks per month with microsecond-level performance timing.

How does Zipang measure candidate English proficiency?

English proficiency is evaluated against the CEFR framework — reading, writing, and listening. The default pass band for BPO roles is B2 or higher. For voice-ready customer support roles, candidates additionally pass a live role-play before client interview. Regional languages (Mandarin, Japanese, Korean, Arabic) are tagged into sub-pools.

What KPI dashboards does Zipang provide?

Every program ships with a built-in KPI dashboard: accuracy, throughput, AHT, CSAT, and shift adherence — per agent, per cohort, per shift. Dashboards are built in-house (not white-labeled from a third party), and clients have direct access.

How is candidate retention measured?

Retention is tracked at 30 / 60 / 90 / 180-day windows. Any hire who does not meet agreed KPIs in the first 30 days is replaced at no charge. A 90-day non-performance window covers issues that surface later, with replacement candidates pulled from the active shortlist pool.

What is the trial task, and who pays for it?

The trial task is a paid pilot assignment that mirrors the client's actual production work. Zipang pays the candidate for completed trial hours at the agreed trial rate; the client only pays Zipang for the trial hours that produce usable output. Trial performance is the final gate before cohort placement.

How does Zipang handle underperformance after placement?

Daily KPI dashboards surface underperformance in real time, and weekly 1:1s with a dedicated account manager turn the data into coaching actions. A monthly business review covers trends and scaling decisions. If a hire still misses KPIs after coaching, the 30-day replacement window kicks in.

Methodology

How Zipang screens, measures, and reports

Name: Zipang Screening Funnel Pass-Rates
Creator: Zipang

The full 5-gate screening funnel, KPI definitions, and production telemetry that back our published numbers. Pass-rates, retention, and quality controls disclosed.

See pricing Hire BPO teams

Data foundation: Zipang research • client insights • about Zipang

The screening funnel

5 gates from CV intake to production

Every candidate passes through the same five gates. Pass-rates below are aggregate figures across 2015–2026 programs run by PT Lima Cakar Bumi. Specific role and cohort pass-rates are published on the request page after the NDA is signed.

Submit candidate CV

Candidates submit a CV through the public intake form. Documents are parsed via an OCR pipeline; PII is redacted before any human review.

~100% (intake)

English proficiency check

CEFR-aligned evaluation of reading, writing, and listening. The default pass band for BPO roles is B2 or higher; voice-ready candidates also pass a live role-play.

~70% pass

Role-specific quiz

Domain quiz for the target role (customer support, data entry, AI annotation, VA). Scenario-based, scored against the same rubric clients see in production.

~50% pass

Structured video interview

A 15–30 minute recorded interview reviewed against the role's competency matrix. Empathy, written clarity, and tone control are scored.

~40% pass

Paid trial task

Candidate completes a paid trial task scored against the client's published KPI. Trial performance is the final gate before cohort placement.

~35% pass

Aggregate pass-rates reflect 2015–2026 programs run by PT Lima Cakar Bumi. Specific role and cohort pass-rates are published on the request page after the NDA is signed.

Production quality

How accuracy and KPIs are measured

Every program ships with a published KPI dictionary. The definitions below are the ones clients and account managers reference during weekly 1:1s and monthly business reviews.

Accuracy bands

Sustained accuracy of 90%+ in production, measured against client-graded gold sets on a per-task basis. Outliers are re-reviewed by a senior cohort lead and fed back into quiz calibration.

Throughput

The France retail AI annotation program tracks 3.4M production tasks per month at the microsecond level. Throughput is reported per agent, per shift, and per cohort.

CSAT

Customer satisfaction is measured per interaction for CS programs, aggregated per agent, per shift, and per cohort. The 90%+ accuracy band on the Transperfect Dataforce program is the production accuracy benchmark.

AHT — average handle time

Average handle time is tracked per shift, per channel (chat, email, voice, WhatsApp), and per tier. AHT is reported alongside accuracy so high AHT does not mask accuracy regressions.

Adherence

Shift adherence and attendance compliance are tracked daily. The candidate screening pass already records prior shift compliance, so the shortlist is filtered to candidates who have demonstrated reliable attendance.

Reporting

Daily dashboards and weekly business reviews

Reporting is structured around three cadences — daily, weekly, and monthly — each with a clear owner and a defined output. Dashboards are built in-house, not white-labeled from a third party. Clients have direct access.

Real-time KPI dashboard

Per-agent, per-cohort, per-shift visibility into accuracy, throughput, AHT, CSAT, and adherence. Dashboards are built in-house, not white-labeled from a third party — clients have direct access.

Weekly 1:1 with account manager

Operational metrics, blockers, and escalations are reviewed weekly with a dedicated account manager. Coaching actions and outcome targets are documented in the dashboard.

Monthly business review

Quality, throughput, retention, and scaling options are reviewed in a structured monthly business review. Decisions on hiring, replacement, and expansion are recorded with clear ownership.

Dashboards are built in-house, not white-labeled from a third party. Clients have direct access to the underlying data behind every chart.

Retention

30-day replacement window, 90-day non-performance window

30-day replacement

Any hire who does not meet agreed KPIs in the first 30 days is replaced at no charge. The replacement candidate is pulled from the active shortlist pool — already screened through the same 5-gate funnel.

90-day non-performance window

A 90-day non-performance window covers issues that surface after the initial ramp — a client-side workflow change, a new product area, or a subtle accuracy regression. The window keeps both sides accountable for outcomes.

Replacement candidates are pulled from the active shortlist pool, which is fed by the same 5-gate funnel that has produced 432 deployed professionals across the France retail and Transperfect Dataforce programs.

Data and compliance

How candidate and client data is handled

The default posture is PII-redacted-by-default, NDA-first, and isolated-by-program. The sections below describe the controls behind every Zipang program.

PII redaction

The public talent pool shows aggregate stats — role supply, top skills, work-fit counts. Individual profiles stay PII-redacted until you sign in to the client portal. Contact details stay hidden until you formally shortlist a candidate.

NDA and IP assignment

A mutual NDA and a client-specific IP-assignment addendum are signed before any candidate name, profile, or trial work is shared. The default IP-assignment terms cover work product produced inside the program.

UU PDP 2022

Candidate and client data is handled in line with Indonesia's UU PDP 2022 (Undang-Undang Pelindungan Data Pribadi). Zipang, operated by PT Lima Cakar Bumi, is the data controller for candidate records and the data processor for client program data.

Client data isolation

Client-specific data, dashboards, and recordings are isolated per program. Cohort, role, and queue data for one client is not commingled with another client's data, and access controls follow the principle of least privilege.

Sample cohort

What the data looks like in practice

The figures below come from production programs Zipang has run since 2015. They are the same numbers cited on the employer portal and the about page.

432

Professionals deployed

France retail AI annotation program

3.4M

Production tasks / month

Tracked at the microsecond level

90%+

Sustained accuracy

Transperfect Dataforce program

14 days

Median time to first shift

From approved shortlist

These numbers reflect the France retail AI annotation program (432 deployed, 3.4M production tasks per month) and the Transperfect Dataforce interpolation annotation program (90%+ sustained accuracy).

Methodology FAQ

Frequently asked questions

The questions we hear most often when clients ask how the published numbers are produced. Answers map back to the funnel, KPI, and data sections above.

What is Zipang's screening funnel?: Every candidate passes a 5-gate funnel: (1) CV intake with OCR parsing, (2) CEFR-aligned English proficiency check, (3) role-specific quiz, (4) structured 15–30 minute video interview, and (5) a paid trial task scored against the client's published KPI. Only candidates who pass every gate are placed into a client program.
What is the pass rate at each gate?: Aggregate pass-rates across 2015–2026 programs run by PT Lima Cakar Bumi are approximately: Gate 1 CV intake 100% (intake), Gate 2 English 70%, Gate 3 Role quiz 50%, Gate 4 Interview 40%, Gate 5 Trial task 35%. Cohort-specific pass-rates are available on request after the NDA is signed.
How is accuracy measured in production?: Accuracy is measured against client-graded gold sets and tracked at the per-task level. The Transperfect Dataforce interpolation program sustains 90%+ accuracy across 20 full-time production agents. The France retail AI annotation program tracks 3.4M production tasks per month with microsecond-level performance timing.
How does Zipang measure candidate English proficiency?: English proficiency is evaluated against the CEFR framework — reading, writing, and listening. The default pass band for BPO roles is B2 or higher. For voice-ready customer support roles, candidates additionally pass a live role-play before client interview. Regional languages (Mandarin, Japanese, Korean, Arabic) are tagged into sub-pools.
What KPI dashboards does Zipang provide?: Every program ships with a built-in KPI dashboard: accuracy, throughput, AHT, CSAT, and shift adherence — per agent, per cohort, per shift. Dashboards are built in-house (not white-labeled from a third party), and clients have direct access.
How is candidate retention measured?: Retention is tracked at 30 / 60 / 90 / 180-day windows. Any hire who does not meet agreed KPIs in the first 30 days is replaced at no charge. A 90-day non-performance window covers issues that surface later, with replacement candidates pulled from the active shortlist pool.
What is the trial task, and who pays for it?: The trial task is a paid pilot assignment that mirrors the client's actual production work. Zipang pays the candidate for completed trial hours at the agreed trial rate; the client only pays Zipang for the trial hours that produce usable output. Trial performance is the final gate before cohort placement.
How does Zipang handle underperformance after placement?: Daily KPI dashboards surface underperformance in real time, and weekly 1:1s with a dedicated account manager turn the data into coaching actions. A monthly business review covers trends and scaling decisions. If a hire still misses KPIs after coaching, the 30-day replacement window kicks in.

Continue from methodology into the rest of the employer resources

PRICING

Pricing guide

Monthly ranges by role, what is included (BPJS, payroll, manager oversight, equipment), and the cost drivers that move the final number.

See pricing guide

EMPLOYER PORTAL

Hire BPO teams

Browse the redacted talent pool, request shortlists, and move candidates through client review. Side-by-side comparison with job boards and freelancer marketplaces.

Open employer portal

PLAYBOOK

Hire Indonesian customer support

Role-specific playbook for client support teams: language depth, channel mix, shift coverage, ramp expectations, and how to request the right profiles.

Read CS playbook

Hire BPO teams See pricing guide Research base