How to route enquiries across 2 to 20 counsellors without losing any

Every guide to lead routing is written for a SaaS sales team. Territories, quotas, account executives. None of it maps cleanly onto a study-abroad office, where the thing that decides who should take a lead is which country the student wants and which language they are comfortable in.

This is what we learned building routing for consultancies with 2 to 20 counsellors, including the parts that broke.

What is lead routing in a consultancy?

Deciding which counsellor gets a new enquiry, automatically, the moment it arrives.

It sounds like an administrative detail and it is not. Routing is where fast response goes to die. A consultancy can answer an enquiry in ten seconds and still lose the student, because the enquiry then sat in a shared inbox for a day while everyone assumed someone else had it. Speed to first reply and speed to the right human are two different problems and both have to be solved.

Should a small consultancy just use round robin?

Yes, if your counsellors genuinely all handle everything. With two to four people and no real specialisation, round robin is the correct answer and anything cleverer is overhead.

One detail changes whether it works: rotate on who was assigned least recently, not on a stored counter.

A counter is the obvious implementation. Keep an index, assign to counsellor number three, increment. It works until the list changes. Someone leaves, someone joins, someone is marked inactive for a week, and the index now points at a different person than it did yesterday, or past the end of the list entirely. We hit exactly that and had to harden it: recalculate against the live list every time, filter to active counsellors first, and keep the index in range.

Least-recently-assigned avoids the whole class of problem. Store a timestamp on each counsellor, pick whoever has waited longest, stamp them. Nothing breaks when the team changes, and it self-corrects if someone was skipped.

When should you route by destination and language instead?

The moment counsellors specialise, which in most consultancies happens by about six people.

Once you have someone who knows US applications properly and someone who handles the UK, round robin is actively harmful. It hands a Canada student to the person who has to go and ask a colleague, and the student can tell. Two things are worth routing on in this market:

• Destination. The strongest signal. A counsellor who places students in Australia every week knows the intakes, the visa timelines and which universities move fast.
• Language. Not everyone is comfortable in English, and a student who is asked to explain their fee situation in a second language will say less than they mean.

Seniority is a third, but it is not a routing dimension on its own. It is a tiebreak, and only for leads you already believe are strong.

The part everyone gets wrong: what happens when nothing matches

Match on destination and language and you will produce combinations nobody covers. A Hindi-preferring student asking about Germany, on a Sunday, when the one Germany counsellor is on leave.

A naive implementation returns nothing. No counsellor, no alert, no error. The lead sits in the database, correctly captured, and no human is ever told it exists. That is worse than round robin, because at least a bad match gets a reply.

The fix is to degrade in steps rather than fail. Ours, for a lead we already scored as hot:

1. Senior counsellor, destination and language both match
2. Senior, destination matches
3. Senior, language matches
4. Any senior counsellor
5. Destination and language both match
6. Destination matches
7. Language matches
8. Anyone active
9. The manager

For a normal lead the senior steps are skipped and it starts at destination and language. The important step is the last one. The ladder ends at a named human who gets every lead nothing else could place, so the answer is never "nobody".

It is also worth logging which step won. We write the winning pool onto the lead record, and it turns out to be one of the more useful fields anyone reads: if half your leads are landing on step 8, your counsellor coverage does not match your actual demand, and that is a hiring signal rather than a software one.

Should hot leads skip the queue?

Yes, and this is well established outside education too. Leads that look strong go to your most capable people first, rather than waiting their turn.

The qualifier is that you need a real score before you can act on it. "Hot" has to mean something measurable, from what the student actually said: how urgent their intake is, how specific they are about the destination, whether their budget is realistic for it, whether they asked to talk to someone. If hot is a feeling a counsellor has, routing on it just concentrates leads on whoever is loudest.

The second qualifier is fairness. Sending every strong lead to two senior people burns them out and starves everyone else of the leads they learn from. Bypassing the queue should apply to a genuine minority of leads.

What about multiple branches?

If you run more than one office, branch is a filter that sits above everything else, not another matching dimension.

A Mumbai student should not be routed to a Delhi counsellor because the language happened to match. So the branch is resolved first, from the line the enquiry arrived on, which is deterministic: the WhatsApp number, the phone line they dialled, the DID on the call. Everything above then runs inside that branch's counsellors, and the fallback is that branch's manager rather than head office.

The reason to tag from the inbound line rather than infer from the student is that inference is a guess, and a wrong branch is worse than a slow one. The student who walked past your Pune office and saved the number from the board is a Pune lead, whatever their pin code says.

Who else should know about the assignment?

The assigned counsellor, immediately, on the channel they already use. That means WhatsApp or Slack, not an email nobody opens and definitely not a dashboard they have to log into.

A manager copy is worth it and a group blast is not. If everybody is alerted, nobody owns it, and you have recreated the shared inbox you were trying to escape. One owner, one manager who can see it slipping.

The alert should carry enough to act on without opening anything: the student's name, the four qualifying answers, the score if you have one, and a tap-to-call or tap-to-chat link. A counsellor who has to look the student up before replying has lost most of the speed you just bought.

Where routing does not help

Routing distributes work. It does not create capacity.

If your counsellors are genuinely at capacity, better routing changes who is overloaded rather than whether anyone is. The signal is easy to check: if leads are being answered quickly but consultations are not being booked, routing is not your problem.

It also cannot fix a lead nobody answered. Routing decides who should reply, and it runs after something has already captured and qualified the enquiry. If enquiries are sitting unanswered overnight, fix the answering first. That is the whole point of WhatsApp automation for a study-abroad consultancy, and the same applies to calls that ring out, which is what an AI voice agent is for. Routing a lead you never captured is not a routing problem.

And if you have four counsellors who all do everything, the honest advice is round robin with a good fallback, and spend the effort you saved somewhere it matters.

The short version

• Route on what your counsellors are actually specialised by. For most consultancies that is destination first, language second.
• Under about six counsellors with no specialisation, round robin is correct.
• Rotate on least recently assigned, not on a stored counter. A counter breaks the moment the team changes.
• Degrade in steps instead of failing, and end the ladder at a named human. A routing system that can return nobody eventually will, silently.
• Log which step won. A pile of leads landing on the last step is a coverage problem, not a software one.
• Let genuinely hot leads skip the queue, but only if hot is measured rather than felt.
• Branch filters above everything else, tagged from the line the enquiry arrived on rather than inferred.
• Alert one owner plus a manager, on WhatsApp or Slack, with enough detail to reply without opening anything.