Here is the position most Indian education institutes are in. You pay Meta and Google for leads. What arrives is form fills. Then you pay a second time — in counsellor salaries, in calling teams, in weeks of follow-up — to work out which of those form fills is an actual student. By the time a genuinely qualified lead is sitting in front of an admissions counsellor, thousands of rupees have gone into it, and most of that was spent on the people who were never interested in the first place.
CareBridge Education reached interested students at ₹25 each. Not leads. Not form fills. People who stayed on a call, asked about the programmes, and wanted to hear more.
They got there by running three stages in a specific order — OBD first, WhatsApp for the numbers they could not legally call, then conversational AI calling for the people who responded. Most case studies show you the headline number and hide the method. This one shows both, including the parts that look unimpressive, because the unimpressive parts are where the useful lesson is.
CareBridge Education is a Hyderabad-based healthcare education provider running B.Sc Nursing, Physiotherapy, Occupational Therapy, Radiology and allied health programmes, with international licensing and placement pathways into the UK, Australia, the Gulf, Canada and the USA. High-ticket, long-consideration admissions — the kind where one enrolment is a multi-year commitment, and where a counsellor's time is the scarcest resource in the building.
They had roughly 50,000 old enquiries that nobody was calling.
The three-stage pattern we actually ran
The mistake most institutes make with a list this size is to point AI voice agents at all 50,000 numbers and hope. That is expensive, it is slow, and a large share of those numbers should not be receiving a promotional call at all. We ran it in three stages instead, and the order is the whole trick.
Why this order matters
Run it the other way round — conversational AI calls to all 50,000 — and you pay full conversational cost to discover that 49,400 people were not interested. Run only the OBD and you get keypresses, not qualification, so your counsellors still spend their days on semi-interested leads working out who is real.
That sequencing is why the cost per interested lead lands where it does, and it is the part most teams get backwards.
What this does to your counsellors' day
This is the operational benefit that institutes feel before they see it in a spreadsheet.
In the usual setup, a counsellor's day is mostly triage. Call a list, reach a third of it, discover most of those are students who filled a form months ago out of curiosity, and spend the remaining hours on the two or three who are serious. The skilled work — talking a nervous parent through clinical placements and licensing pathways — gets whatever time is left.
With the stages above, the semi-interested never reach them. A counsellor opens their day to a short list of people who have already heard what the programme is, confirmed they are eligible, and said they want to speak to someone. Every conversation is a conversation worth having.
That is the actual product. Not cheaper leads — a counselling team that spends its time counselling.
Why IVR alone doesn't work, and why outbound AI alone doesn't either
Most education institutes have an IVR. You know the shape: "Press 1 for admissions, press 2 for fees, press 3 to speak to a counsellor." It exists to protect the staff from call volume, and it works — by losing callers. A prospective student who navigates three menu levels to reach a busy counsellor usually doesn't navigate them twice.
The opposite failure is outbound AI calling with nothing behind it. You call 50,000 people. A few hundred are interested. Half of those were driving, in class, or at work when the call landed, so they call back two hours later — and hit the IVR. The agent that spoke to them has no idea the call is related. The context is gone, and so is the lead.
The combination is the point. Outbound AI calling opens the conversation at scale; the inbound path has to be ready to close the loop with the context intact. Treat them as one system and the funnel holds. Treat them as two products you happen to own and it leaks at exactly the moment the lead showed intent.
What those numbers actually mean
One percent interest sounds low until you weigh it against the alternative, which was zero — nobody was ever going to call these people. Each of those 600 heard what the programmes actually are, asked their own questions, and stayed on the call.
The 50 is the number that matters operationally, because it is the only one that consumes counsellor time. Fifty qualified conversations out of a list everyone had written off, without a rupee of new ad spend.
The number that actually changed the decision
Cost per interested lead: ₹25.
For context on why that matters: education and overseas-consulting lead costs on Meta and Google in India routinely run into the thousands of rupees per qualified lead, and much of that spend buys form fills from people who never answer the phone. Calling enquiries you already own skips the acquisition cost entirely — you are paying for conversation, not for clicks.
That is the structural argument for calling what you already have before buying more, and it is why this project ran at all. These enquiries were already paid for, years ago. The only question was whether anyone would ever talk to them.
Where the IVR earns its place
Three things have to be true on the inbound path for an outbound campaign of this size not to leak.
The return call has to be recognised. When one of those 500 interested leads calls back, the system should already know who they are, which programme they asked about, and what the AI agent told them. They should not re-explain themselves to a menu.
Routing has to be by intent, not by menu position. A caller asking about B.Sc Nursing eligibility and a caller asking about UK licensing after graduation need different people. An AI agent on the inbound leg can establish which in one question, rather than making the caller guess which numbered option describes them.
Nothing should dead-end. Out of hours, mid-queue, counsellor unavailable — every path needs a next step that holds the lead rather than dropping it. In practice that usually means handing the conversation to WhatsApp.
The 24/7 hotline is the piece most institutes are missing
This is the part of the setup we would argue hardest for. CareBridge runs a hotline that is answered around the clock by an AI agent — not an answering machine, not a callback form, an agent that picks up and has the conversation.
Admissions enquiries do not respect office hours. A student finishes coaching at 9 PM and starts researching. A parent in the Gulf calls at 2 AM India time because that is their evening. A lead who got an outbound call during a lecture rings back the moment they are free, which is rarely between 10 and 6.
Every one of those calls used to hit voicemail or an unattended IVR, and a prospective student who reaches voicemail at 9 PM calls the next institute on their list, not yours the next morning. With the hotline answered live, that caller is qualified, their questions are answered, and a counsellor slot is booked before they hang up — at 2 AM if that is when they called.
It also changes what the outbound campaign is allowed to do. When you know every return call will be answered, you can call 50,000 old enquiries without worrying that the interested ones will ring back into silence.
The WhatsApp layer
Education admissions are not decided on one call. A 17-year-old gets interested; a parent decides. That conversation happens over days, often in the evening, and usually in writing.
Our WhatsApp agent runs that stretch for CareBridge — programme details, fee structures, eligibility, campus and clinical-exposure information, and the international pathway questions that parents actually ask. It carries the call context, so a lead who spoke to the voice agent on Monday is not starting over on Thursday.
This matters more in education than in almost any other vertical we work in. The gap between interest and enrolment is measured in weeks, and the institute that is still present and answering at week three is the one that enrols the student.
What we would tell another institute
Start with the enquiries you already own. Before spending more on ads, find out what your dormant list is actually worth. The economics are not close.
Expect a 1 percent interest rate on aged records, and plan for it. Anyone promising dramatically more from a cold list is selling you something. The value is in the absolute number of conversations, not the percentage.
Fix the inbound path before you run the outbound campaign. Volume on the way out exposes every weakness on the way back in. An IVR that drops returning callers will quietly undo the whole exercise.
Measure cost per qualified lead, not cost per lead. Qualified is the only unit that maps to counsellor time, and counsellor time is the real constraint in admissions.
If this is your problem too
If you are an education institute in Hyderabad or anywhere in India sitting on old enquiries nobody has the capacity to call, the arithmetic above is reproducible. We run AI voice calling in Telugu, Hindi, English and 22+ Indian languages, with the inbound path and the WhatsApp follow-up built as one system rather than three.
Talk to us and we will put a live test call on your own number so you can hear the agent before you commit to anything. More on the vertical specifically on our education page.