Can aggregate data identify individual students?
Yes, when the group is small. An average over two students reveals both values to anyone who knows one of them, by simple subtraction. "Aggregate" describes the shape of a calculation, not the size of the group under it, and boutique-practice caseloads are exactly the size where aggregate reports stop being anonymous.
A vendor tells you it only shares "aggregate insights," never individual student records. That sounds safe by definition. Aggregate, in plain language, just means a statistic calculated over a group rather than a row of data about one person, and whether that statistic is actually safe depends entirely on how many people sit inside the group being averaged.
Average a school's entire graduating class together and no single student's number is visible in the result. Average two students together and both numbers are visible, because either one of them can be recovered from the other with simple subtraction. The word aggregate describes the shape of the calculation. It says nothing about whether the group underneath it is large enough to hide in.
A worked example, kept deliberately small
Here is a composite example, not a real report from a real vendor. Say a platform's dashboard shows you "average SAT of students from your practice admitted to one particular university this cycle," and say exactly two of your students were admitted there this cycle. The dashboard reports the average as 1430.
You already know one of those two students, because she is your own client. Say her score was 1400. Two scores averaging 1430 sum to 2860. Subtract the 1400 you already knew, and the other student's score is 1460, exactly, with no guessing involved.
Nothing in that arithmetic touched a name, an email address, or any field the platform would call personal information. It only needed one average and one score you already had. The "aggregate" report handed you the second student's private data anyway, through the back door of subtraction.
Boutique practices are the small-group case
This is not a rare edge case for an independent educational consultant. A solo or small practice might place a handful of students at any given college across several years, and fewer than that within a single admissions cycle. Any report sliced down to "your practice, this school, this year" is very likely a cell of one, two, or three students.
The platform does not need to intend this outcome for it to happen. A dashboard built to reassure counselors that their data is used "only in aggregate" can still produce a subtractable cell the moment a counselor filters by school, by year, or by program. Small caseloads make that filter easy to hit by accident, not just on purpose.
The vocabulary that actually protects a small cell
There is a real, established fix for this, and it has specific vocabulary. If a vendor's aggregate reporting is genuinely built to handle small groups, these are the phrases its answers should contain.
Minimum cell size is a rule that no group gets its own line in a report unless it has at least a minimum number of people in it, set in advance. A group smaller than that threshold does not get published on its own at all.
Suppression is what happens to a cell that falls under that minimum: instead of showing "average of 2 students," the report drops that line entirely, or folds those two students into a broader, safer category until the group is big enough to hide in.
Adding noise means the platform deliberately nudges a reported number away from the true average by a small random amount, so that even a group large enough to publish resists exact recovery. Someone trying the subtraction above lands on a number that is close, not exact.
What to ask a vendor
Three questions test whether a vendor's "aggregate insights" actually use this vocabulary, or only borrow the word aggregate.
What is the minimum group size before any number gets reported at all, and is that minimum written down anywhere you can see it?
Can a report ever be sliced, by school, by year, by program, or by any combination of those, down to fewer students than that stated minimum?
Who receives or buys these aggregate insights, and at what level of detail, a whole industry's worth of averages, or a slice narrow enough to matter to one family?
The practical takeaway
Aggregate is a real, useful way to protect privacy when the group underneath the number is large enough that no one row can be pulled back out of it. It is not a privacy technique at all when the group is two or three students, which is the ordinary size of a per-school, per-year slice from a boutique practice.
Before you take a vendor's "aggregate insights" language as a settled answer, ask what the minimum group size is, whether a report can be sliced below it, and who is actually receiving those insights. The checklist has a question aimed at exactly this gap. Two related reads: what "de-identified" actually means for a record before it ever reaches an aggregate stage, and how a free pricing model shapes what a platform does with your data once it has been collected.
Evaluating a platform this week? Bring the one-page checklist to the demo: the checklist.