All case stories
Case · No. 031 / 2025 Piprali Road, Sikar NEET / JEE Coaching 10 weeks

The coaching institute that stopped losing 11pm leads.

Vidya Pravah Coaching has prepared NEET and JEE candidates from Shekhawati for fourteen years. Their problem in 2025 wasn't teaching. It was that parents in Churu and Jhunjhunu were messaging at 10pm — and finding silence.

Live · Q4 2025

"A Hindi-first WhatsApp agent that knows the syllabus, the fees, and when to call a human."

2.4×
Enrolment conversion
00:00
First-reply time, 24×7
87%
Hindi queries handled by bot
40L
Estimated revenue recovered

The problem

A funnel with a hole that opened at 9pm.

Manish Sharma, the director, had analytics from his ad spend. The lift in WhatsApp messages between 8pm and midnight was clear. So was the response time the next morning: too late.

Most parents in Sikar's catchment cannot make decisions during the day — they run shops, drive auto rickshaws, work in the fields. Evenings are when they message coaching institutes, and Vidya Pravah was competing with two larger Kota chains who had call-centres. By the time Manish's office opened at 9am, half the leads had moved on. The dropped messages were costing him about ₹40 lakh a year in unrealised admissions.

He had tried a generic chatbot from a SaaS vendor. It replied in Hinglish that didn't sound like Sikar, couldn't quote the syllabus accurately, and routed serious enquiries to a generic "we'll get back to you." He shut it off in three weeks.

"

If the bot sounds wrong, the parent doesn't trust the institute. I would rather have no bot than a bot that embarrasses my staff.

MS
Manish SharmaDirector, Vidya Pravah Coaching

What we scoped

A WhatsApp agent that sounds like Sikar.

We spent three days at the institute and two evenings sitting with the front-desk staff as messages came in. The shape of the conversation was actually predictable — about 80% of incoming queries fell into eleven recognisable patterns:

  • Fee, batch, and timing questions."NEET 2026 ki fees kya hai?" "Sunday class hota hai kya?"
  • Faculty & track-record queries."Last year kitne selections aaye?" "Physics kaun padhata hai?"
  • Logistics for outstation parents."Hostel facility hai?" "Kya distance learning option hai?"
  • Demo class & visit booking."Visit karna hai, time mil sakta hai?"
  • Serious enrolment intent."Mera beta 11th mein hai, admission kaise hoga?"

Categories 1–4 are fully bot-answerable with the right grounding. Category 5 is where humans must take over. So we scoped a single agent — handle the first four with care, recognise the fifth, and hand to a named counsellor with full context.

What we built

A bot that cited its sources, in Hindi.

WhatsApp Business Cloud API

Verified, green-tick, BSP-routed

Deployed on the official Meta API via a verified BSP. Templates approved, message-template library built in Hindi and English. Two-week onboarding handled by us.

Grounded retrieval

Answers from real institute documents

The bot reads from the institute's actual fee structure PDF, batch schedule, faculty bios, and last-year results. Every answer cites which document it came from — staff can audit any reply.

Hindi-first, Hinglish-aware

Reviewed by native Sikar speakers

Tone, vocabulary, register all written with the institute's senior counsellor over two days. The bot uses "aap" not "tu", quotes fees in lakh not lac, and does not switch to English unless the parent does.

Hand-off discipline

One ring, full context

When a parent shows enrolment intent, the conversation summary is sent to the on-duty counsellor's WhatsApp with parent name, batch interest, and asked-questions. Counsellor replies in their own voice — same number, no transition.

How it ran

Ten weeks, including WhatsApp verification.

Week 01–02
Discovery & copy

Three days on-site. Two evenings on the inbox. Tone-of-voice document signed off by Manish.

Week 03–05
API onboarding

WhatsApp BSP onboarding, green-tick verification, template approvals.

Week 06–08
Build & shadow

Bot live in shadow mode — staff saw drafts, approved or edited, every reply logged.

Week 09–10
Cutover & counsellor training

Bot live to public. Counsellors trained on hand-off etiquette. First admissions cycle measured.

What changed

Forty-six new admissions that didn't slip away.

In the four months since the bot went live, Vidya Pravah has tracked 46 admissions sourced from after-hours conversations that, by Manish's own analysis, would otherwise have been lost. Enrolment conversion on inbound WhatsApp leads is now 2.4× what it was in 2024.

The softer outcome: front-desk staff stopped working evenings. The institute used to roster two people until 9pm on rotation. They don't anymore. Those staff hours were redirected to following up on warm leads from the previous month — which itself recovered another tranche of admissions.

"

One parent told me last month — "aapka WhatsApp office aadhi raat ko bhi answer karta hai, isiliye yahaan admission le rahe hain." I told her the truth. She laughed and said, "still, it answered when I was awake. The Kota one didn't."

MS
Manish SharmaDirector

What we'd change

One thing we got wrong.

Honest postmortem

We didn't account for entrance-exam panic season

In November, two weeks before NEET-PG counselling, message volume tripled overnight. The bot held up technically — but parents asking specific counselling-cutoff questions needed real expert input, not retrieval. Our hand-off threshold was tuned for normal volumes; it became a bottleneck. We added a "season mode" that lowers the hand-off threshold during specific four-week windows the institute tells us about in advance. We should have asked about the exam calendar before scoping. We do now.

Where are your evenings going?

A 30-minute call with a partner. We listen, then write you a plan.

Book a discovery call ↗