Deduplication means one record per prospect, with the history intact
Student lead deduplication across channels means recognising that a web enquiry, a chat conversation, an SMS thread and a missed call came from the same person, then attaching them to a single CRM record. For an enrolment team the goal is practical: one adviser contacts the prospect, once, knowing what has already happened.
Australian recruitment produces duplicates in a particular way. A school leaver may enquire about a Group of Eight university, a regional university and a TAFE pathway in the same fortnight, using a personal email and a school one. An international student may enquire directly, through an education agent and through a partner portal, and appear three times. Domestic applicants also move through admissions centres and direct applications, each with its own record.
This guide covers where duplicates come from, which matching keys to use, what merge rules to write down and which privacy and spam rules apply. It builds on our pieces on cleaning a student lead database and lead routing and SLAs.
Where duplicates come from: a different identifier per channel
| Channel | Reliable identifier | Typical cause of a duplicate |
|---|---|---|
| Enquiry form on your site | Email, often mobile | Personal email first, school email later |
| Website chat | Nothing, or an email given mid-conversation | Anonymous visitor returning on another device |
| SMS or messaging thread | Mobile number | Number entered as 04xx, +61 4xx or with spaces |
| Inbound call | Caller ID, sometimes withheld | Parent's phone instead of the student's |
| Open days, school visits, agent and partner lists | Email, name | Bulk uploads with no checks |
Three cases need an explicit rule: the parent who enquires for a student, the student with two emails, and the international enquirer who arrives via an agent and later directly. Decide which record is the person and which is the agent's submission about the person.
Choose matching keys from safest to most fragile
- Exact match. The same normalised email (lower case, trimmed) or the same normalised mobile number. Matching can be automatic.
- Probable match. The same first and last name with the same course of interest, or the same mobile with different emails. The system suggests; a person confirms.
- Possible match. Similar names, transliterations, name order differences. Flag, never auto-merge.
Normalise phone numbers to the international format in ITU recommendation E.164: +61, then the number without the leading zero. A mobile stored as 0412… in the CRM and +61 412… in an SMS tool is the same number.
Name order is a real hazard for international prospects: given name and family name are often reversed between forms. Never merge on name alone; namesakes are common and a wrong merge mixes two files.
Write merge rules before the first merge
- Master record. The oldest stays; the others attach to it, keeping the identifier your other systems use.
- Contact details. The most recent verified value wins; the older goes to history.
- Course and intake. Keep the latest stated interest and preserve earlier ones. A shift from a diploma to a bachelor degree is a signal, not noise.
- Owner. The adviser with the most recent human contact keeps the record; otherwise your routing rule decides.
- Consent and opt-outs. The most restrictive setting wins.
- Source. Keep every source, including the agent or partner, so referral arrangements are not lost.
- Audit trail. Log who merged what, when and why so merges can be reversed.
Keep un-merging possible, especially for shared family numbers.
Privacy and spam rules: an opt-out must survive a merge
Merging records is handling personal information. The Privacy Act 1988 and the Australian Privacy Principles apply to most universities and larger private providers, and the Office of the Australian Information Commissioner publishes the principles. Two bear on deduplication: taking reasonable steps to keep personal information accurate, up to date and complete, and correcting it on request. A single record makes both easier. Some public universities are also subject to state privacy law, so confirm with your privacy officer which regime covers you.
Commercial electronic messages are covered by the Spam Act 2003, which requires consent and a functioning unsubscribe, and telemarketing calls are covered by the Do Not Call Register rules. The deduplication consequence is simple: if any merged record carries an unsubscribe, a do-not-call flag or an erasure request, the surviving record inherits it. Have your privacy officer review the approach.
Three habits reduce risk:
- Store the date and source of each consent, channel by channel.
- Apply the most restrictive setting when two records conflict.
- Keep a suppression list separate from the CRM so a later import cannot recreate an opt-out.
For prospects under 18, involve a parent or guardian in conversations that lead to a commitment, and do not merge their record into the student's.
Put the process in place in six steps
- List every entry point: forms, chat, SMS, phone lines, open days, agent and partner lists.
- Normalise on entry: lower-case emails, +61 phone format, cleaned names.
- Turn on CRM duplicate rules, in suggest mode first, then automatic for exact matches.
- Clear the existing database in batches, exact matches first.
- Write the ownership rule: who takes over the merged record and how the other adviser is told.
- Review monthly the rejected suggestions and reversed merges.
If an institution shares its database across campuses, brands or pathway colleges, settle with your privacy officer whether matching across them is permitted before switching it on.
Measure without inventing a benchmark
No public source says what duplicate rate is normal for an Australian institution, so measure your own before and after.
| Indicator | What it shows |
|---|---|
| Share of records with at least one probable duplicate | The size of the backlog |
| Duplicates created per channel per month | The entry point to fix first |
| Reversed merges | The quality of your keys |
| Prospects contacted by more than one adviser | The visible cost to the team |
| Merged records that kept an opt-out | Whether safeguards hold |
Do not credit fewer duplicates for higher yield. Enrolment outcomes have many causes, and any effect has to be tested against your own comparison periods.
What Skolbot contributes, and what stays with the institution
Skolbot Chat is an agent installed on the institution's website. It answers from content the institution has published, qualifies the enquiry, offers an appointment or open day and hands the conversation to the team and the CRM. Capturing an email or mobile number early gives you a matching key anonymous visitors usually lack. The wider logic is in our article on AI agents from first contact to CRM.
Matching and merge rules stay with the institution and its CRM. How the connector behaves with your CRM is confirmed during scoping. To place your tooling, see the CRM comparison for higher education, and for the big picture return to our guide to recruiting more students.



