Playbooks · 5 min read
EdTech on WhatsApp: Fees and Eligibility Need a Human
An edtech company's messaging problem is not the same as a tutor's, even though the words in the enquiries look identical. A tutor gets thirty messages a week from people who mostly live nearby. An admissions team gets thousands, from every state, in every language, most of them from people who clicked an ad four seconds ago and will not remember doing it.
That volume changes what automation is for. It is not there to answer. It is there to make sure a counsellor's day is spent on the enquiries worth spending it on.
The enquiry that arrives with no context
A large share of edtech enquiries begin with a click on an advertisement that opens a chat. The person arrives having read one line of creative. They have no relationship with you, they have not chosen a course, and often the first message is just "hi" or the pre-filled text from the ad.
At that moment nobody knows anything useful. Which course, which exam, which language, which state board, whether they are the student or the parent, and whether they can be reached again.
So the automation's first job is the one a counsellor would otherwise spend six messages on:
- Establish who is asking. Student or parent. It changes the entire conversation and it is one question.
- Establish what for. Which course, exam or class, from a short list rather than an open question — open questions at this stage return "any".
- Establish language. Not by guessing from the name. Ask, and then actually route to a counsellor who speaks it.
- Set an expectation with a real number in it. "A counsellor will call you today between 4 and 7" is honest and checkable. "Our team will connect with you shortly" is what everyone says and nobody believes.
Never state fees automatically
This is the first hard refusal, and it goes against every instinct in an admissions team, because "fees kitni hai" is the most-asked question in the inbox.
Edtech pricing is rarely a single number. It varies by cohort, by intake, by duration, by which scholarship or waiver applies, by whether an instalment plan is involved, and by whatever pricing decision was taken last week. A template holding a figure will go on holding it after the figure changes, and the version a prospect screenshots is the version they will argue from.
Worse, the answer is frequently entangled with a financing arrangement, and a message about instalments or a loan is a message about somebody's money. That is not a thing to automate, and it is certainly not a thing to make claims about. Fee structures, financing and refunds go to a person who can state the current terms and be accountable for them.
A quoted fee is a commitment. A template does not have the authority to make one.
Never state eligibility automatically
The second refusal, and the one with the sharper consequences.
"Can I apply with a diploma?" "Am I eligible with 58 percent?" "Backlog hai, chalega?" "Is this valid for my state's exam?"
Every one of these looks like a lookup and is a determination. It depends on the applicant's actual documents, on the institution or board's current rules, and often on a case-by-case admissions decision. Rules change between intakes, and what counts varies by jurisdiction and by awarding body — we are in no position to tell you what any of them currently require, and neither is a generated sentence.
An automated "yes, you are eligible" is a person who pays a fee, prepares for weeks, and finds out at verification that they were not. That is a refund, a complaint, and a story.
Hard stop on eligible, eligibility, criteria, qualify, admission ho jayega, backlog, gap year, percentage chahiye. Straight to admissions, with the enquiry intact.
The claims that quietly attach themselves to templates
Edtech marketing has a vocabulary that must not leak into automated replies: placement guarantees, assured jobs, salary figures, "100 percent" anything, a promised rank or score, or a claim about how many students achieved something. In a brochure those claims are at least reviewed by somebody. In a template they send themselves, at scale, to individuals who read them as personal promises.
If your marketing does make outcome claims, they belong on a page with whatever substantiation and disclaimers your legal team requires, not in a chat reply that arrives as a friendly answer to a direct question.
Volume brings its own problems
Three things that only appear at scale.
The same person, four times. Ads retarget, people click again, and a prospect who receives three identical opening messages concludes you are a bot farm. Deduplicate on the number.
Quiet hours. Ads run at 2am. Replies at 2am get muted.
Consent drift. A person who enquired about one course has not agreed to hear about your other seven, or about next year's intake, or about a webinar. Promotional messaging on Meta's platform sits in its own category with its own consent and template approval, and the specifics are Meta's and change — read their documentation rather than an article's summary. The durable point: on this channel an unwanted message is blocked and reported by somebody who was in the middle of their day, and blocks are what damage a business number.
What we run today
PostEngage answers Instagram comments and DMs on the official Graph API. That is the whole product — there is no WhatsApp in it, on any plan. This post covers the channel because admissions teams search for it and deserve an accurate picture.
For the channel we do run: seven days to answer a comment, twenty-four hours in a DM thread, restarted only by their own next message. Templated replies are free and unlimited, a credit is spent only when the AI writes new text, the free tier is 100 credits with no card, and packs start at ₹499.
Related: tutors and individual teachers for the small-scale version of this, the education sector on Instagram, and what governs this channel in India.
