Building a CRM your counsellors will actually use
Admission teams abandon CRMs for predictable reasons. Five decisions that made ours stick, and the features we deliberately did not build.
Most admission teams are on their second or third CRM. The pattern is always the same: bought with enthusiasm, configured over six weeks, used properly for a month, and then quietly replaced by a WhatsApp group and someone's personal spreadsheet.
The tool is rarely the problem. The problem is that a CRM is usually bought to answer the manager's questions, and then handed to counsellors who have a completely different question — who do I call next? — which the tool makes them dig for.
When we built our own lead-management system for the counselling team, that tension was the whole design brief. Five decisions did most of the work.
1. One screen, one question
The default view is not a dashboard. It is a single ordered list that answers "who do I call next", and it is already sorted by the time the counsellor logs in.
Not filterable-to-that. Not a saved view they have to pick. That list, first, by default.
The ordering is a small piece of logic that weighs how recently the lead came in, how many attempts have been made, whether the student picked a callback time, and how close the relevant last date is. A counsellor should never have to decide who to call. Deciding who to call is a tax that gets paid every single morning, and it is the moment where the spreadsheet starts looking attractive again.
Everything else in the system — search, filters, reports — is reachable, and none of it is on the first screen.
2. Logging a call must take under ten seconds
This is the one that decides whether a CRM survives.
If logging an outcome takes a minute, a counsellor doing forty calls a day is being asked for forty minutes of data entry. They will not do it. They will batch it to the end of the day, then to Friday, then not at all — and then the data is fiction and every report built on it is fiction too.
What we ended up with: one row, one tap for the outcome, optional one-line note, done. The five outcomes that cover almost everything are single buttons — no answer, call back later, interested, not eligible, chose elsewhere. Anything rarer lives behind a "more" link.
"Call back later" opens a time picker with three suggested times and lets them type one. Picking a time is the only thing that takes more than one tap, because it is the only thing that changes what happens next.
Measure this. If the median time from opening a lead to logging an outcome is over fifteen seconds, the system will be abandoned, whatever else it does well.
3. The status list should be short and honest
Every CRM ships with a status list someone designed in a meeting: New, Contacted, Qualified, Nurturing, Proposal Sent, Negotiation, Closed-Won, Closed-Lost. In admissions, "Nurturing" means nothing and "Negotiation" means nothing, so counsellors park everything in whichever status is least likely to get them asked about it.
Ours is deliberately short, and every status maps to something that is either true or not:
- New — nobody has attempted contact
- Attempting — tried, not yet spoken to a human
- In conversation — actually spoken, discussion live
- Applied — form submitted
- Admitted — fee paid
- Closed — with a mandatory reason: not eligible, chose elsewhere, deferred, unreachable
That mandatory close reason is worth more than any report the system produces. Six weeks of honest close reasons will tell you whether you are losing students to fees, to eligibility, or to follow-up — and those three problems have nothing in common except that they all show up as "lost" in a normal CRM.
4. Assignment rules before features
Unassigned leads are dead leads. If a lead lands in a shared pool, everyone assumes someone else has it.
Every lead gets an owner at the moment it is created — round-robin, or by course, or by language, whatever fits the team. Never nobody. The rule runs in seconds because a lead arriving at 11pm and waiting for Monday's manual assignment has already cost you the thing that mattered.
A reassignment path matters too. Counsellors go on leave, leads go stale, someone is better with a particular course. Reassignment should be two taps and it should carry the whole history with it, so the new owner opens the lead already knowing what was said.
5. Build the manager's view last
This is the one people get backwards, and it is why so many CRM projects fail.
Reporting gets built first because the person paying for the system wants the dashboard. So the data model is designed around the report, the counsellor's workflow gets bolted on afterwards, and the counsellors — the only people who put data in — find the thing unusable.
Build in the order the data flows. Counsellor screen first. Make it fast, make it obviously better than a spreadsheet, get it used for a few weeks. The reports then build themselves, because the data underneath is real.
A dashboard built on data nobody entered honestly is worse than no dashboard, because people make decisions from it.
What we deliberately did not build
The list of things we cut is as important as what we shipped:
- Email sequences. Admission conversations in India happen on WhatsApp and on the phone. Email is where documents go after a decision, not where the decision is made.
- Lead scoring. A number out of a hundred that nobody can explain does not help a counsellor decide anything. The ordered call list already encodes the same judgement, in a form that is actionable.
- Custom fields for everything. Every custom field is a box someone has to fill and nobody reads. We kept the set small and fought to keep it there.
- A mobile app. A fast mobile web page that works on a mid-range Android over patchy data was more useful than an app nobody would install and update.
Does it work?
The honest test is not a feature list. It is whether the spreadsheet comes back.
Three signs the system is genuinely being used: the median time to first contact stays under the target without anyone being chased about it; the share of leads with a logged outcome stays high through peak season, when it is easiest to let it slip; and nobody asks for a data export to "just check something".
That third one is the real signal. The moment someone exports to Excel to answer a question, the system has failed at the thing it exists to do — and the shadow spreadsheet has already started.