What actually moves the needle on retention? Our 2026 Student Impact Report has the numbers. Read the report
Platform Solutions All features Pricing Research 2026 Impact Report Blog Customer stories About Careers FAQs Sign in Book a demo
Security & data protection

Your students’ data is not our asset.

We act as a data processor under your institution’s instruction. That is not a marketing line — it determines what we are architecturally able to do, and it is the reason we cannot offer you a cross-sector benchmark.

Four commitments

What is architectural, not contractual.

Each of these is a property of how the platform is built rather than a policy we could quietly revise. That distinction matters when you are assessing a vendor.

PoolingNoneYour data is never combined with another institution’s, for any purpose including benchmarking.
Model trainingYours onlyYour data trains your model. It does not improve a model that serves anyone else.
ResidencyIndiaDefault region, in line with the DPDP Act. Other regions on request.
Sale of dataNeverNo sale, no sharing with advertisers, no data product built on your records.
Regulatory posture

Which law applies, and what we do about it.

FrameworkOur roleWhat we provide
DPDP Act, 2023 (India)Data Processor; your institution is the Data FiduciaryProcessing agreement, India residency, breach notification within 72 hours, assistance with data principal requests
FERPA (United States)School Official under the institution’s direct controlWritten agreement, use limited to the stated purpose, no redisclosure
GDPR (EU/EEA students)Processor under Article 28DPA with SCCs where required, records of processing, DPIA support
Institutional policyWhatever your policy definesWe enforce your retention, visibility and access rules rather than substituting ours

We are not a law firm and this table is not legal advice. It describes our operating posture so your counsel has something concrete to assess. Our DPA and security questionnaire responses are available on request to info@eduplatter.com.

Controls

The unglamorous part.

Encryption and access control are table stakes, and any vendor will claim them. The controls that actually differ between vendors are the ones about who inside the company can see student records, and under what circumstances.

Access

Named individuals, time-boxed, logged

Production access to institutional data requires a ticket, expires automatically, and is logged in a record your institution can request. There is no standing engineering access to student records.

Encryption

In transit and at rest

TLS 1.3 in transit; AES-256 at rest with keys managed in a hardware security module. Per-tenant key separation on dedicated deployments.

Isolation

Per-institution tenancy

Logical isolation as standard, dedicated tenant or on-premises on System engagements. No shared analytical layer across institutions.

Deletion

On instruction, verified

Deletion on your instruction or at contract end, with written confirmation and backup expiry within 35 days.

Subprocessors

Listed, and you are notified

The current list is in the DPA. Material changes come with 30 days’ notice and a right to object.

What we do not claim

Honest limits.

Written here because a security page that only lists strengths is not useful to the person evaluating it.

We are not certified to ISO 27001 yet

We operate to its control set and can share our internal assessment, but we do not hold the certificate. If your procurement requires one, say so early — it may be a blocker and we would rather you know in week one than month three.

We cannot fix upstream data quality

If attendance is recorded on paper in half your departments, our models will be weaker for those students. We report that in the validation phase rather than imputing quietly.

Shared notes create a real privacy tension

Cross-department visibility is useful for coordination and risky for students. We default to the narrowest sensible setting and require an explicit institutional decision to widen it, but the call is yours.

Predictions can be wrong about individuals

A model with good aggregate precision still misclassifies people. That is why we refuse to wire a risk score to an automatic consequence, and why every prediction shows its factors.

Assessment

Send us your security questionnaire.

We answer institutional questionnaires directly and will flag gaps rather than working around the question.