IndiVillage
HomeResourcesBlogGeneral
General

How do I know if my annotation vendor is using offshore labour, and should I care?

Offshore labour isn't inherently risky. What matters is labour practices, security, and accuracy. How to evaluate vendor models.
Author · Mark Pinnes
·
26 May 2026
·
10 min
IndiVillage specialists at workstation
IndiVillage Operating Centre · Bengaluru

How Do I Know If My Annotation Vendor Is Using Offshore Labour, and Should I Care?

You've received a proposal from an India-based annotation vendor. Your gut reaction: "Offshore = cheap and possibly low-quality." Your compliance team is already nervous. But before you reject it, consider this: Audere, a healthcare analytics platform, trusted an India-based team with 631,000 medical image interpretations. They achieved 98% accuracy. Healthcare systems adopted the results. The offshore bias is real—and it's costing you.

The Offshore Bias

The stereotype: offshore = cheap labour cuts corners on quality. Reality: offshore varies wildly. Some offshore vendors exploit workers and produce garbage. Others maintain world-class quality with ethical labour practices. The difference is vendor model, not geography.

The real distinction: dedicated vs. gig.

A vendor with 1,000+ dedicated full-time employees in rural India (paid fairly, 96% retention over 16 years) is operationally different from a platform that sources from 50,000 gig workers earning $2/hour. The latter cuts corners and burns through staff. The former compounds expertise.

Geography (India, Philippines, Eastern Europe) is a red herring. Vendor model is what matters.

Vendor Model Comparison

Dimension Dedicated Model Gig Model
Employment Full-time FTE; 1,000+ staff Contract/freelance; 10,000+ workers
Retention 96% annually over 16 years 30-50% annually; constant churn
Wages Fair market + benefits $2-5/hour (OpenAI Kenya investigation, 2023); no benefits
Learning 18+ months domain expertise 8 weeks ramp-up before departure
Quality 98%+ accuracy sustained 85-90% accuracy (high error rate)
Accuracy trend Improves year-over-year (expertise compounds) Flat or declining (new workers every cycle)
Cost 30% of Silicon Valley rates 10% of Silicon Valley rates
Scale Bounded by geography + hiring (repeatable) Unbounded but unpredictable (exploitative)

Dedicated model costs more than gig, but delivers higher quality and sustainability. Gig model is cheaper upfront but quality suffers and long-term reliability is poor.

How to Know if They're Offshore

Ask directly. "Where are your annotation teams based? Are they full-time employees or contract workers? What's your turnover rate? How long have they been with your company?"

Evasive answers are a red flag. "We have global centres of excellence" (vague) vs. "We have offices in Bengaluru, Hyderabad, and Delhi, with 800+ annotators, 96% retention" (specific).

Look for proof. Published case studies should mention team location (or not mention it, which is fine—doesn't mean they're hiding anything). Audere's case study explicitly names IndiVillage's India-based operations. That transparency builds trust.

Ask about economics. If the vendor can't explain why their cost is 30% of Silicon Valley-based competitors, they're either: (a) cutting corners somewhere, (b) highly efficient, or (c) underpaying staff. Vendors willing to share economics openly are usually option (b).

Labour Practices as Quality Proxy

Here's the mechanism most people miss: labour practices are correlated with accuracy.

Fair wages. Staff earning living wages stay longer. Turnover is lower. New hires are rare. Your annotation team has 18 months of domain learning, not 8 weeks.

Benefits. Healthcare, education stipends, retirement contributions. These signal a company that expects long tenure. Gig-economy vendors don't offer benefits (staff are "contractors," not employees).

Career progression. If your only path is "annotator → more annotation," you leave. If there's a path to QA lead, training specialist, or team lead, you stay. IndiVillage promotes annotators into these roles.

Stable employment. Full-time vs. contract matters psychologically. Full-time staff treat mistakes seriously (their job is at stake). Contract staff shrug (there's always another gig).

96% retention over 16 years is proof that staff choose to stay. That's not possible with gig-model exploitation.

Security Implications of Offshore

"Offshore" and "secure" are orthogonal. An offshore vendor with SOC 2 Type II, DPA, audit trails, and role-based access controls is more secure than a US-based vendor with none of these.

What actually matters:

Data residency. Where does your data live? If it's in a secure India-based data centre (encrypted, backed up, access-controlled), that's fine. If it's on someone's laptop, that's not fine. Location is irrelevant; controls are everything.

Access control. Who can see your data? Role-based permissions (annotators see their batches only, not others' work) prevent unauthorised access. This has nothing to do with geography.

Audit trails. Every annotation logged with timestamp and user ID. You can prove who did what on which date. Again, geography doesn't matter; logging does.

Legal framework. DPA in place, GDPR compliance documented, HIPAA BAA signed (if applicable). These are contractual controls that work regardless of vendor location.

Breach response. 24-72 hour notification, incident investigation, remediation. This is a process requirement, not a geography requirement.

A Real Scenario: Offshore Security Under Regulatory Pressure

A fintech platform processing transaction data under PCI-DSS requirements evaluated two vendors: one US-based, one India-based. The US vendor claimed "we're local, so your data is safe here." The India vendor provided: SOC 2 Type II report, DPA, PCI-DSS compliance documentation, data residency in Amazon AWS India region (encrypted at rest and in transit), access logs showing role-based permissions, and incident response procedures. The fintech's compliance team initially preferred the US vendor for familiarity. After reviewing the India vendor's SOC 2 Type II report and security controls, they realised the India vendor had more rigorous controls than the US vendor (which had no SOC 2). They selected the India vendor. Two years later, when a minor security incident occurred (a contractor's laptop containing cached data was stolen), the India vendor's incident response—24-hour notification, forensic investigation, remediation—met PCI-DSS requirements perfectly. The offshore location never became a compliance issue because controls were in place.

IndiVillage is India-based. It also has SOC 2 Type II, DPA, HIPAA compliance (Audere case), and published security controls. The offshore location doesn't override the security controls—it's orthogonal to them.

Accuracy is Geography-Agnostic

99%+ accuracy from Bengaluru = same rigour as Silicon Valley. There's no physics that makes annotation more accurate if done in California.

Proof: Audere case. Healthcare data (sensitive), India-based team (offshore), 98% accuracy (trustworthy). Healthcare systems wouldn't deploy inaccurate annotations—their liability is real. They verified the accuracy independently before adopting.

Worked Example: Offshore Accuracy Verification

A computer-vision platform building an autonomous-robotics model needed 500,000 egocentric video frames annotated for obstacle detection. They received proposals from three vendors: one Silicon Valley-based (native English, local), one India-based (IndiVillage), and one Philippines-based (gig-model). The Silicon Valley vendor claimed "local presence ensures quality." The India vendor published case studies: Taranis (4.5M+ images, 96%+ accuracy), Machani Group robotics (18 months zero-drift QA, 99.4% on robotics workloads). The Philippines vendor offered lowest price, no case studies. The platform ran a 500-image pilot with all three. Results: Silicon Valley 94% accuracy, India 99% accuracy, Philippines 87% accuracy. The India vendor cost 40% more than Philippines, 20% less than Silicon Valley, but delivered highest accuracy. Geography had zero bearing on quality. Vendor model did. They signed with the India vendor.

If accuracy is your concern, ask for proof—case studies, customer references, third-party validation. Don't assume offshore means lower accuracy. Test it.

ESG and Ethical Considerations

Here's where offshore actually matters, positively:

Ethical offshore > unethical onshore. An India-based B Corp with fair wages, women leadership, and community investment is more ethical than a San Francisco-based vendor exploiting H-1B loopholes.

Impact sourcing. Rural India impact sourcing (IndiVillage model) creates jobs in low-income areas, funds education, builds gender equity. If values alignment matters to your buyer, ethical offshore wins.

Cost-quality-ethics alignment. Ethical offshore delivers all three: cost-efficient (India), high-quality (retention compounds expertise), and values-aligned (labour practices, impact measurement).

Gig-economy onshore fails on all three: expensive, low-quality (turnover), and exploitative (workers paid $2/hour).

The ESG Advantage: A Procurement Case Study

A major European agricultural conglomerate evaluating AI-powered crop diagnostics faced an ESG mandate: vendors must demonstrate labour equity, environmental responsibility, and governance transparency. They received proposals from two annotation vendors: one US-based (no published labour metrics), one India-based (IndiVillage—B Corp certified, 50% women workforce, 96% retention). The conglomerate's procurement team requested evidence of vendor labour practices. The US vendor claimed "we're committed to diversity" but provided no documentation, retention data, or wage transparency. IndiVillage provided: B Corp audit report, women leadership breakdown (50.3%), salary ranges relative to local median (1.4x), retention rates by year, education sponsorship programmes funded through profit-sharing. The ESG checklist heavily weighted the India vendor. Beyond ethics, the procurement team recognised that the offshore vendor's labour stability (96% retention) directly predicted annotation accuracy and long-term partnership reliability. They signed with IndiVillage, citing both ethical alignment and operational excellence. The offshore location was not a liability—it was a competitive advantage because the vendor's labour practices were verifiable.

The FAQ

Q: If a vendor is offshore, should I require extra security audits?
No. Require the same security controls (SOC 2, DPA) regardless of location. Let controls prove security, not geography.

Q: What if my compliance team objects to offshore?
Show them proof. Audere case study (healthcare + offshore + 98% accuracy). SOC 2 report (security controls). DPA (legal framework). HIPAA BAA (regulatory compliance). Data overcomes bias.

Q: Can offshore vendors handle sensitive data?
Yes, if they have proper controls. Healthcare systems use IndiVillage (offshore) for medical interpretations. GDPR applies whether the vendor is in Germany or India—controls are what matter.

Q: Are offshore vendors cheaper because they cut corners?
Some do. Others are cheaper because cost of living is lower. Cheaper ≠ worse quality if the vendor invests in retention and training. Audere proof: cheaper AND high quality.

Q: What's the risk of offshore labour dependency?
Concentration risk: if a vendor has one office and it's disrupted (power outage, political instability), work stops. Mitigation: ask about redundancy. IndiVillage has 11 offices—work reroutes if one goes offline.

Q: Should I require my vendor to be local?
Only if you need frequent face-to-face meetings or have specific regulatory requirements (some governments require local vendor presence). Otherwise, it's limiting unnecessarily.

Q: What happens if an offshore vendor experiences a geographic disruption (power outage, political instability)?
Ask about redundancy. Does the vendor have multiple offices or data centres in different regions? IndiVillage has 11 offices across rural India—if one office goes offline, work reroutes to another without interruption. A vendor with a single office carries concentration risk. Mitigation: include business continuity clauses in your contract specifying recovery time objectives (RTO) and recovery point objectives (RPO). Example: "In case of office disruption, vendor shall reroute work to alternate facilities within 4 hours with zero data loss."

Q: If I'm using an offshore vendor, should I require data to stay offshore, or can it be transferred?
That's a business decision, not a security one. Some buyers require data residency in the vendor's home country (India data stays in India, EU data stays in EU). Others permit temporary transfers to other secure data centres for backup. Your DPA should specify data residency requirements. Example: "Data shall remain in India-based AWS region. Transfer outside India requires 30-day written notice and customer written approval." The key is specificity. Let the vendor know your requirements upfront.

Q: If an offshore vendor experiences a security breach, how do I know I'll get timely notification?
Contractual language and third-party audit. Your contract should require 24-72 hour breach notification, investigation report within 7 days, and remediation timeline. The vendor's SOC 2 Type II audit includes their incident response procedures—review it. A vendor with current SOC 2 has been audited on their breach response process. If they lack SOC 2, ask for their incident response plan in writing. Offshore location doesn't slow notification—clear contractual obligations do.


Your Next Best Action

Don't reject an offshore vendor based on location. Request: labour practices documentation (wages, retention, benefits), security certifications (SOC 2, DPA), customer references (proof of accurate work), and incident response procedures. Judge by facts, not geography bias.

Work with us
Run a specialist audit.
100 frames. Your modality. Your accuracy target. Returns in 48 hours.
Run a specialist audit
Talk to a delivery lead →