How College Vidya scaled Hindi-Hinglish online-degree lead qualification with Caller Digital's SARA
College Vidya (Blackboard E-Learning Pvt. Ltd.) is one of the largest online-degree platforms in India, with 100+ UGC-DEB approved partner universities and roughly one lakh annual admissions across Online MBA, BBA, BCA, MCA, B.Com and adjacent programs. Caller Digital deployed SARA — a custom-trained Hindi-Hinglish voice counsellor — for outbound lead qualification, counselling-session booking and LeadSquared-native disposition write-back across 20 structured fields per call.
High-velocity inbound lead funnel needing UGC-DEB-neutral qualification across 100+ universities, 20-field structured capture, and continuous 9 AM–9 PM coverage
College Vidya's counselling team needed to qualify every inbound lead within minutes — capturing 20+ fields per call (academic status, preferred course, preferred university, specialisation, budget, working status, callback time, loan interest) — and schedule qualified students with a senior human counsellor.
Manual qualification could not consistently sustain the 9 AM–9 PM coverage window. Regional Hindi-Hinglish handling varied across counsellors. And the platform's UGC-DEB neutrality — no preference for any of the 100+ partner universities — had to be enforced consistently on every call, with no placement guarantees, no salary commitments, and no comparative-ranking statements.
Inconsistent coverage across the 9 AM–9 PM window
Manual counsellor capacity dropped during peak inbound volume and evening tails. Students enquiring outside business hours often waited until next day.
Hindi-Hinglish handling varied by shift
Different counsellors applied different code-mix patterns. Some leaned shudh Hindi; others over-Englished. Brand voice inconsistency hurt trust on first-touch calls.
20-field structured capture was manual
Counsellors typed capture fields into LeadSquared after the call. Half-completed records were the norm.
UGC-DEB neutrality depended on counsellor discretion
Across 100+ partner universities, individual counsellors had implicit preferences. Off-brand utterances about university quality or placement carried real compliance and trust risk.
SARA — a nine-stage Hindi-Hinglish counsellor with LeadSquared-native 20-field write-back, 17 named objection handlers, and UGC-DEB neutrality enforced as a platform guardrail
Caller Digital deployed SARA — a custom-trained Hindi-Hinglish voice AI counsellor positioned as a brand-safe, neutral first-touch counsellor for College Vidya. SARA is female, calm, unhurried — a baseline emotional range with slight warmth on personal disclosures, no theatrical enthusiasm, no dramatic concern. Built to sound like a real counsellor on her 35th call of a Monday, not on her 5th.
The conversation runs a nine-stage state machine: greeting and identification, academic status (qualification, year, score), course discovery, specialisation, university preference, budget capture, working status and loan interest, counselling-slot pitch, and a clean close. SARA never goes backwards once a stage is passed. Seventeen named objection handlers cover the most common student and parent objections from the UGC-DEB online-degree funnel.
SARA writes back 20 structured fields to LeadSquared on every call: lead name, academic status, preferred university, preferred course, preferred specialisation, suggested universities, preferred budget, callback date/time, working status, lead qualification status, loan interest, AI Counsellor Summary (180–200 chars), lead temperature (HOT/WARM/COLD/NA), conversation URL, call recording URL, transcript URL, call time, voicemail-detected flag, 2–3-sentence call narrative, Add Activity flag.
Hindi-backbone Hinglish, telephony-trained
Hindi sentence structure with English technical terms (M-B-A, U-G-C, D-E-B, E-M-I spelled letter-by-letter). Silent language switching on two consecutive English-only student turns. Trained on real Indian counselling-call audio.
Nine-stage state machine, three absolute rules
Any word ≠ silence (never reconnect after a word). Twice + no progress = advance (never a third repeat). Interrupted mid-sentence resumes with 'Ha, mein bol rahi thi —' and repeats the full sentence.
20-field LeadSquared write-back per call
Structured fields written to mx_Custom_1 through mx_Custom_18 on every connected call, plus an Activity entry, call recording URL, transcript URL and conversation URL.
17 named objection handlers
Dedicated short-sentence Hinglish branches for the most common student and parent objections — 'abhi result nahi aaya', 'dusra university bhi dekh raha hu', 'fees zyada hai', 'placement guarantee chahiye', 'main sochke batata hu'.
UGC-DEB neutrality as a platform guardrail
Never pushes a single university across 100+ partners. Never guarantees placements or salary. Never makes comparative ranking statements. Defers exact fee numbers to human counsellor. Discloses AI honestly when asked.
Counselling session booking with round-robin assignment
Qualified students booked into a free counselling session with a round-robin-assigned human counsellor. Slot proposed inside the call; warm transfer to a logged-in counsellor on request.
Nine performance standards configured for the deployment, with the nine-stage state machine and 20-field LeadSquared write-back delivering structured first-touch coverage manual SDR teams could not
The Master Agreement is dated 30 April 2026; the deployment ran the standard 14-working-day timeline (3 days bot preparation, 7 days integration with LeadSquared and knowledge APIs, 2 days internal testing, 2 days UAT). SARA is live on the LeadSquared production instance.
| Dimension | Before | After |
|---|---|---|
| Coverage window | 9 AM–9 PM not consistently sustained | Continuous 9 AM–9 PM outbound coverage |
| Hindi-Hinglish handling | Varied across counsellors | Consistent Hindi-Hinglish on every call |
| Field capture in LeadSquared | Manual, often half-complete | 20 fields written on every connected call |
| UGC-DEB neutrality | Depended on counsellor discretion | Platform-enforced — no university preference, no placement guarantee |
| Objection handling | Improvised per counsellor | 17 named, scripted handlers |
| Session booking | Manual scheduling | Round-robin counsellor assignment from inside the call |
Frequently Asked Questions
An outbound Hindi-Hinglish voice AI counsellor (SARA) for first-level lead qualification of online-degree enquiries across 100+ UGC-DEB partner universities, with counselling-session booking, LeadSquared CRM write-back across 20 disposition fields, and Phase-2 inbound support.
SARA is configured as a brand-safe, UGC-DEB-neutral counsellor: she never pushes a single university across the 100+ partners, never guarantees placements or salary, never makes comparative-ranking claims between universities, and defers exact fee numbers to a human counsellor. She always discloses she is an AI when asked, and honours immediate DNC requests on any opt-out keyword.
Hindi-backbone Hinglish is the default register — Hindi sentence structure with English technical terms (M-B-A, U-G-C, D-E-B, E-M-I spelled letter-by-letter). The agent switches silently to English on two consecutive English-only student turns. Pure 'shudh Hindi' and corporate jargon are both avoided.
Lead ingestion is via real-time / scheduled LeadSquared API fetch with CSV fallback. The agent writes back 20 disposition fields (academic status, course interest, university preference, specialisation, budget, callback time, loan interest, working status, AI summary, lead temperature, recording URL, transcript URL, conversation URL, voicemail flag) to LeadSquared on every call, and creates an Activity per call. Round-robin counsellor assignment for warm transfers and session bookings is handled inside the call flow.
≥90% lead qualification accuracy, ≥95% information response accuracy on partner-university and program information, ≥95% lead information capture across 20 fields, ≥95% counsellor hand-off / live transfer success, ≥90% escalation routing accuracy, ≥98% outbound dial-out success, ≥99% retry logic execution, ≤3-second P50 bot initial response latency, and ≥95% counselling session booking success.
Per-minute consumption at ₹5.70/min on a 50,000-minute prepaid POC package (12-month validity), plus a one-time integration fee of ₹70,000 covering LeadSquared + up to five course/info APIs + round-robin counsellor hand-off. Deployment ran 14 working days: 3 days bot preparation, 7 days integration, 2 days internal testing, 2 days client UAT. Platform uptime committed at 99.5% monthly.
Run a similar voice AI rollout for your team
We work with Indian enterprises across BFSI, D2C, retail SaaS, and consumer services. Book a 30-minute consult to see what voice AI can do for your call queue.
© 2025 Caller Digital | All Rights Reserved