Failure mode 1: Duplicate candidate submissions
A senior Java engineer in Bangalore is actively interviewing. He has sent his CV to 3 staffing agencies in the last 30 days. Two of those agencies submit him to your open role within 4 days of each other. Both claim "first submission" status. Neither will concede. Your TA team spends 3 hours mediating, the candidate gets confused and annoyed, and whoever loses the attribution battle pulls the candidate.
This scenario happens 15-30 times a month at a typical mid-market enterprise running 30+ open roles through 8 vendors. The cumulative cost is a week of TA bandwidth per month plus the reputational hit of candidates who decide not to pursue your company because the process felt disorganized.
The fix is a shared candidate deduplication layer, not a policy. Policies (timestamp-based attribution, 90-day exclusion windows) exist at every enterprise but are unenforceable without tooling because nobody agrees on what the timestamp was or which email thread counts. A shared ATS or vendor portal that does automatic deduplication on email, phone, and LinkedIn URL at submission time eliminates the problem at the source. The candidate is flagged as duplicate before two vendors can both claim submission; the first submission (with timestamp) wins; the second vendor gets an automated notification and moves on.
The implementation bar is lower than most TA leads assume. A shared vendor portal with submission deduplication on email + phone + LinkedIn URL is a 3-month build or a ready-to-buy tool in the ₹50k-2 lakh/month range depending on vendor count and seat licensing. Payback is typically under 4 months from TA time savings alone, before counting placement value preserved.
Failure mode 2: Vendor black holes (no feedback loop)
A staffing agency submits 15 candidates to your open role in week 1. They hear nothing for 2 weeks. They escalate to the TA lead, who responds "we are still reviewing." By week 4, the agency has de-prioritized your requirement because their recruiters cannot justify spending time on a client who never gives feedback. You complain you are not getting enough submissions on your role; the agency says they stopped trying because your process is broken.
This is the most common multi-vendor failure mode because it is silent: the TA team does not realize they are starving the pipeline until the role has been open for 45 days and three good vendors have disengaged. The cost is absorbed as "the market is tight" when actually the agencies are just routing their best candidates to clients who respond.
The fix is a structured feedback SLA, enforced by the shared portal. Every submitted candidate must get a status update within 72 hours: rejected (with reason), shortlisted for interview, on hold (with date of next review). No candidate sits in limbo past 72 hours. The portal sends automated reminders to the TA reviewer at 48 hours and an escalation at 72.
Agencies respond to feedback discipline by prioritizing your requirements. A vendor seeing a 100% 72-hour feedback rate will send you their best candidates first; a vendor seeing 40% of submissions ghosted will send you their B-team and save the A-team for responsive clients. The feedback loop is not courtesy; it is how you get access to the good candidates.
Failure mode 3: Unclear vendor tiering
You have 8 vendors but no clear tiering. Every new requirement goes to all 8. Each vendor sends 10-15 submissions. You get 100+ CVs to review per role. 60% are low-relevance because the vendors are hedging (they know 7 others are also submitting, so they send volume and hope). Your TA team drowns in triage.
The fix is explicit three-tier vendor tiering with different access rules. Tier 1 (preferred, 2-3 vendors): first access to every requirement for 3 days before Tier 2 gets invited. Submissions capped at 5 per role to enforce quality over volume. Tier 2 (secondary, 3-5 vendors): activated after day 3 if Tier 1 has not closed. Submissions capped at 10. Tier 3 (emergency, 2-4 vendors): activated only for niche or stale roles. No cap but quality-monitored.
The tier a vendor lands in depends on their performance over the trailing 6 months: submission-to-interview conversion rate, placement rate, average time to first-submission, candidate quality ratings from the hiring manager. A vendor with 60%+ shortlist-to-interview conversion and under 48-hour time-to-first-submission is Tier 1. A vendor at 30% conversion and 5-day time-to-submission is Tier 3 and has to earn their way up.
The system has to be transparent. Every vendor sees their own tier and the metrics that determined it. Vendors tolerate being in Tier 3 if they can see exactly what they need to improve to move up. Vendors rage-quit opaque tier assignments.
Failure mode 4: Attribution disputes on placements
Six months after a placement, Vendor A disputes the placement fee. They claim they submitted the candidate 4 weeks before Vendor B, were ignored, and now Vendor B is collecting the ₹4 lakh placement fee. Vendor A has email evidence of the earlier submission. Vendor B has ATS evidence of the later submission. Your TA lead has 6 hours of meetings, a legal escalation, and a strained relationship with whichever vendor loses.
This is the single most expensive multi-vendor failure because the disputes are real money (₹3-8 lakh per placement) and the evidence usually exists on both sides. Without a single source of truth, the dispute becomes a he-said-she-said that costs legal hours regardless of who wins.
The fix is an audit trail every party trusts because they all write to it. A shared portal with immutable submission timestamps, visible to all participants, ends 90% of disputes at the source: Vendor A can see their own submission timestamp, Vendor B can see theirs, the system enforces the rule, nobody has to litigate. The remaining 10% of disputes (where both vendors submitted within the 30-day exclusion window via different channels) get decided by the portal audit log, which the vendors pre-agreed would be the source of truth when they signed the vendor agreement.
The audit trail must include: submission timestamp (system time, not vendor time), candidate identifiers (email + phone + LinkedIn URL), the JD version the candidate was submitted against, the submission source (portal, email, referral), and the disposition history (reviewed, shortlisted, rejected, hired). If any of these are missing, you do not have an audit trail, you have a log file.
The SLA structure that actually changes behavior
Most vendor SLAs read like legal documents and drive no behavior change. Effective SLAs are short, measured, and tied to tier assignment, not just contract clauses. Here is a structure that works at mid-market Indian enterprises.
- Time-to-first-submission: target under 48 hours for Tier 1, under 72 for Tier 2. Missed twice = tier downgrade review.
- Submission-to-interview conversion: target 50%+ for Tier 1, 35%+ for Tier 2. Trailing 6-month rolling average.
- Offer-acceptance rate: target 70%+ across all tiers. Below 50% means the vendor is over-selling candidates and wasting your time.
- Feedback adherence (from your side): 100% of submissions get disposition within 72 hours. Violations tracked by TA lead and reviewed monthly.
- Escalation response: vendor must respond to TA escalation within 4 business hours during business hours.
Tooling patterns that work
A functional multi-vendor pipeline runs on three tooling layers. First, a shared vendor submission portal with built-in deduplication and audit trail. This is table stakes; without it, everything else is duct tape. Second, a tier-and-SLA enforcement layer that automatically applies submission caps, tier-based access timing, and feedback-deadline escalations. Third, a vendor performance dashboard that shows each vendor their metrics relative to the tier thresholds.
You do not need to build any of this. Commercial vendor management platforms (VMS) exist in the ₹1-5 lakh/month range for mid-market Indian enterprises. The ones worth evaluating have a strong submission deduplication engine, native DPDPA-aligned data handling (consent capture, 12-month retention defaults, candidate rights workflow), and a vendor-facing portal the vendors actually like using.
The common failure is adopting a VMS and not enforcing it. Vendors keep emailing CVs as attachments because "the portal is slow" or "it is easier for me." The TA team accepts the emails "just this once" to not break the placement. Six months later the portal has 40% coverage and you are back to where you started. The rollout bar is: all submissions go through the portal, or they do not count. No exceptions, including for Tier 1 vendors during their complaint phase in months 1-2.
What to measure monthly
A healthy multi-vendor pipeline produces five numbers every month that trend in the right direction. Submission volume per tier (Tier 1 should dominate, Tier 2/3 should be shrinking over time as sourcing improves). Duplicate submission rate (should fall below 3% within 6 months of adopting a portal). Average days-to-first-shortlist (target under 5 days, great if under 3). Placement distribution across vendors (no single vendor over 60% of placements, otherwise you have concentration risk). Vendor satisfaction score (quarterly pulse survey, target 4+/5 on "would recommend partnering with this company to another vendor"). If any of these drift, investigate immediately; vendor relationships recover slowly and fail fast.