Threat modelling a business workflow
Map one workflow, identify where trust changes and turn plausible threats into tested controls and owned decisions.

Threat modelling is a structured way to ask what could go wrong in a system’s design and how to address it. It’s useful before a new workflow is built and when an existing workflow changes.
Microsoft’s threat-modelling training introduces the technique and the use of data-flow diagrams. For a small business, start with one process that has a clear owner.
Define the scope and business outcome
Choose a workflow such as a customer uploading a document or a staff member approving a payment. Identify the expected result and the records involved.
Bring together someone who understands the work and someone who understands the systems. Include suppliers when their part of the design needs clarification.
Record what the review excludes so its conclusions aren’t mistaken for a whole-business assessment.
Draw where information moves
Show people, applications, stored records and connections. Note where information crosses between different owners or permission boundaries.
Include administrative and recovery routes as well as the normal user path. A forgotten export or service account can change the model.
Keep secrets and live personal records out of the diagram.
Ask concrete questions at each boundary
Could someone impersonate the sender? Could someone without permission change a record, read it or make the service unavailable? Could an action occur without useful evidence of who performed it?
Use a specific example for each concern. For an upload, ask what happens if the file is too large, has unexpected content or belongs to another customer.
Distinguish deliberate misuse from ordinary errors while considering both in the design.
Prioritize with stated assumptions
Describe the business impact and the conditions needed for each scenario. Use evidence about the system itself and credible threat guidance.
Don’t turn an industry statistic into a precise probability for your business. Record uncertainty and decide where further investigation is needed.
Assign an owner to the decision about mitigation or accepted residual risk.
Specify and test the control
Write the intended result in testable terms: an ordinary user can’t read another customer’s upload, or a payment change can’t proceed without verified approval.
Confirm technical requirements with the implementer. Test allowed and denied paths, failure behaviour and the evidence needed afterwards.
A threat model proposes protections. You’ll still need code review, configuration review and security testing scoped to the system.
Keep it with the change record
Retain the diagram, assumptions, decisions and test results with the project. Update the model when data flows, suppliers or access boundaries change.
When the workflow crosses several systems, see our application development service. For testing how existing controls interact during an incident, use the security-layers exercise.


