When your customer's data goes to a supplier, your business still holds the relationship and usually the legal responsibility. The processor operates the system, but the business that chose it owns the consequence. That asymmetry is the core of third-party data processor risk.
This guide is for the manager who signs the contract or approves the vendor. It sets out the risk categories, a scoring method, the controls that reduce exposure and the questions a board should ask. It is not legal advice.
Why processor risk is your risk
A processor might host your customer database, send your SMS, run your support desk, process payroll, provide an AI assistant or store backups. If it has a breach, you may have to notify customers. If it loses access, your operation stops. If it changes its subcontractor, your data may move to a country you did not approve. The processor's contract may limit its liability, but it does not transfer your accountability.
Eight risk categories
1. Scope creep
The processor receives more data than the service needs: full customer records for a delivery notification, salary data for a scheduling tool, or an entire mailbox for a support chatbot. Every extra field increases the impact of a breach and the difficulty of deletion.
2. Security failure
Weak access control, unpatched systems, shared credentials, missing encryption or no incident response plan. Ask for evidence, not assurances. A current penetration test summary or a completed security questionnaire is more useful than a logo wall.
3. Subcontractor chain
The processor uses a cloud host, a messaging provider, a transcription service and a support contractor. Each is another organisation with access. If the subcontractors are not disclosed and approved, you cannot assess the full path of the data.
4. Jurisdiction and transfer
Data may be stored or accessed outside Bangladesh. That can affect notification duties, lawful access requests and the customer's expectations. Know where the primary processing and support access happen, and which law applies.
5. Breach notification
A processor that discovers a breach but notifies you weeks later removes your ability to respond. The contract should define what counts as an incident, who notifies whom, within what period and with what information.
6. Availability and lock-in
A single region, no export, proprietary formats and no exit assistance turn a supplier problem into a business continuity problem. Test whether you can retrieve your data in a usable format before you need to.
7. Privacy rights handling
If a customer asks to see, correct or delete their data, the processor must be able to find it across backups, logs and derived datasets. Many systems can delete the live record but not the copy in a data warehouse or an AI index.
8. Secondary use and AI training
Some providers use customer data to improve models or shared analytics. That may be acceptable for anonymous usage data and unacceptable for customer records. Get the answer in writing and check whether it can be switched off.
Scoring likelihood and impact
Use a simple 1-to-5 scale for each risk, then multiply. The number is less important than the conversation and the consistency it creates.
- Likelihood: how likely is the failure in the next 12 months, given the processor's controls and your history with it?
- Impact: what would happen to customers, operations, money and reputation if it occurred?
- Score: likelihood multiplied by impact, then ranked.
- Context: volume, sensitivity, access level and substitutability. A small supplier with admin access to your payment data can be riskier than a large one with read-only access to public content.
Set thresholds for low, medium, high and critical risk, and define what approval is needed at each level. Review the thresholds annually; a growing data set changes the answer.
Controls that reduce exposure
- Data minimisation: send only the fields the service needs. Remove identifiers where a pseudonymous reference will do.
- Pre-contract diligence: security questionnaire, evidence review, references and a check of the provider's own subprocessors.
- Contract terms: purpose limitation, confidentiality, security obligations, breach notice, audit rights, return and deletion.
- Access control: named users, least privilege, MFA and a defined offboarding process.
- Subprocessor approval: notice before a change, a right to object and a current list.
- Monitoring: a named owner, an annual reassessment and trigger-based reviews after incidents or ownership changes.
- Exit planning: export format, transition assistance, deletion certificate and a tested restore.
Contract clauses that matter
- The purpose and duration of processing, with a defined data scope.
- The security measures required, described specifically rather than as "industry standard".
- Notification of a personal data breach without undue delay, with the information you need to assess it.
- Assistance with customer requests and impact assessments.
- Subprocessor list, change notice and objection rights.
- Audit or assessment rights, including on-site where the risk justifies it.
- Return or deletion of data at termination, including backups and derived copies.
- Liability, indemnity and insurance, reviewed by your lawyer.
- Governing law and dispute resolution.
Ongoing monitoring
Due diligence is not a one-time form. Keep a register with the processor name, service, data categories, risk score, owner, contract expiry and last review date. Reassess when:
- The processor reports a breach or a significant outage.
- It changes ownership, hosting region or subcontractors.
- You add a new data category or a new integration.
- Its audit report or certification expires.
- Your own regulatory obligations change.
Review the highest-risk processors annually and the rest every two years; a register with no owner is only a list.
Board questions
- Which processors hold customer, employee or student data, and which could stop our operations?
- Who approves a new processor, and what evidence do they see?
- Which processors can access data from outside Bangladesh?
- How quickly would we know about a breach at a processor?
- Have we tested getting our data back from the two most critical providers?
Assessment checklist
- Service description and data flow mapped.
- Data categories and volume documented.
- Security questionnaire completed and evidence reviewed.
- Subprocessors listed and approved.
- Contract clauses reviewed by legal.
- Breach notification path tested with a contact list.
- Exit and deletion plan documented.
- Risk score approved by the business owner.
- Review date added to the register.
Red flags
- No named security contact and no incident history.
- Refusal to disclose subprocessors or hosting locations.
- "We delete everything on request" with no description of backups or derived data.
- Customer data used for model training by default, with no opt-out.
- No export format and no transition assistance.
- A contract with no breach notification timeline.
Where this fits
If you need a repeatable processor review with evidence, start with Managed Technology & Cybersecurity. The pricing page explains how we scope assessment work. For the wider supplier picture, read how third-party security risk is changing.
Sources
- NIST SP 800-161 Rev. 1, Cybersecurity Supply Chain Risk Management Practices for Systems and Organizations, accessed 17 November 2026.
- NIST Privacy Framework, accessed 17 November 2026.
- ISO/IEC 27701:2019, Privacy information management, accessed 17 November 2026.
- DataGuidance, "Bangladesh" data protection overview, accessed 17 November 2026.
Responses (0)
No comments yet. Add the first one below.
Leave a response
Your email address will not be published. Required fields are marked *.