AI admissions calls: handling the intake peak

At intake, months of volume compress into weeks

Call volume at an educational institution is not spread evenly across the year. When applications open, when results are announced and as the deadline approaches, the number of calls rises many times above an ordinary day — and it stays there for weeks.

The size of the team does not change in those weeks. The result repeats every intake: lines fill, waiting stretches, some applicants never get through at all, and the same questions arrive again tomorrow.

This article is about managing the peak: splitting calls into three groups, which questions can be automated, how status enquiries are answered, the outbound side, and planning the peak before it starts.

What is lost at the peak

Loss at a peak takes three forms, and none of them appears in a report as a lost applicant.

• Unanswered calls: the line is busy and the caller does not try again
• Lost in the queue: the call is answered, the wait stretches, the person hangs up
• Repeat calls: the same question asked three times because the answer was incomplete
• Wrong information: an operator under pressure gives an imprecise answer
• Missing documents: discovered late, with no time left before the deadline

The last line is the most expensive, because it is not a technical problem but a lost opportunity for the applicant. Learning that a document is missing on the final day is a problem for the family and a reputational matter for the institution.

Splitting calls into three groups

Managing a peak starts with a split. Admissions calls fall into three groups by content, and each has a different route to an answer.

1. 1Repeated information questionsDocument lists, deadlines, addresses, opening hours, payment details, programmes. This group is most of the volume and the answer is the same every time — which means it can be automated fully.
2. 2Personal status questions«Has my application been accepted?», «what stage am I at?». The answer depends on data in a system; if the agent can read it the call is answered, if not it goes to a person.
3. 3Decisions and exceptionsAppeals, disputes about document validity, individual circumstances. This group always goes to a person, and keeping a line free for it is the main purpose of managing the peak.

The practical value of the split: automating the first group creates operator time for the second and third. The goal is not to replace operators but to put them only on the conversations that need a person.

Which questions can be automated

Answers in the first group are reliable on one condition: they must come from an approved source. Admission rules change, and giving out old information is worse than giving none.

• Document lists — by category, with the exceptions
• Dates: when applications open and close, exam and announcement dates
• Address, opening hours, queueing arrangements
• Payment amounts and details
• Programmes, modes of study, language streams
• The exact answers to the twenty most frequent questions

Status questions: connecting to the system

The second group — personal status questions — eats the largest share of operator time at a peak, because the answer is short while the lookup is long.

Automating this group depends on the agent being able to read the system. Technically that is solved with a tool that talks to the institution's own systems: the agent reads the status by application number or identity details and answers within the conversation.

One condition matters: the rule for giving out personal information by phone has to be settled in advance. Who may ask what, which details are said on a call and which are not, is fixed by the institution's own policy — a decision that comes before the script.

Outbound: the other half of the peak

An intake period is not only inbound calls. The institution has calls of its own to make, and those are usually the ones the peak pushes aside.

• Warning applicants with incomplete documents — before the deadline
• Confirming acceptance and explaining the next step
• Booking a time for document submission, and a reminder
• A deadline reminder for those who have not completed registration
• Keeping a list of those interested in the next intake

The first line is the most valuable on the list: told in time, an applicant can fix a missing document. That single call affects both the intake numbers and the impression the institution leaves.

Planning the peak in advance

A peak is not a surprise — its dates are in the calendar. Even so, preparation usually begins after it has started.

• Last year's call statistics are broken down by hour and day — the shape repeats
• The list of most frequent questions is refreshed and written into the knowledge base
• Concurrent call capacity is checked beforehand
• Operator shifts are built around the hourly profile of the peak rather than spread evenly
• A separate line or queue is kept for the third group

The general overflow model — the part that is not sector-specific — is in call overflow at peak volume. This article is about what an intake period adds to it. The logic of holding quality across locations is in call quality across branches.

Limits

What is not automated in an admissions process is clear, and it protects both the institution and the applicant.

Formal decisions come first: acceptance, rejection and exceptions are not the subject of an automated answer. Second are appeals and disputed cases. Third are individual circumstances — health, social status, special conditions — where the answer comes from a person and from documents.

How admissions questions are handled on the chat channel is in the training centre chatbot article; follow-up in the steady flow is in course enquiry follow-up.

How Vexvon supports this flow

Vexvon's AI voice agent answers and places calls from the institution's own scenario, approved information and escalation rules. The parts a peak uses:

• Inbound calls are separated by time policies and prioritised routing rules, and the rules can be tested by simulation before they meet live calls
• The knowledge base holds the admission rules, dates and document lists; the same base can be targeted separately at the chatbot and the voice agent
• With webhook tools the agent can read status data from the institution's own system
• Outbound campaigns are built from a CRM filter — applicants with incomplete documents are their own segment
• Decisions and appeals are handed to an operator by a condition in the scenario

How hourly load and reporting are produced is on the analytics page, and building the knowledge base on the knowledge base page.

Measurement

A peak is read in four numbers, and they have to be measured separately from ordinary periods.

• Answer rate — measured separately for peak hours
• Share of calls answered automatically, and how many of those produced a repeat call
• Average wait for calls handed to an operator
• When missing documents were discovered — how many days before the deadline

The second is the most honest: the share of applicants who call again after an automated answer shows whether the answer was actually enough. To plan an intake period, get in touch.