Skip to content
EncryptTechBD
Home Industries Our Work Blog Pricing About Contact Discuss Your Project
বাংলা (BN)
Privacy & Compliance 6 min read

Cybersecurity policy for small business: a practical risk guide

How to turn a small business cybersecurity policy into a risk register with owners, controls, evidence and an explicit residual-risk decision.

E
EncryptTechBD · 0 comments
Laptop on a desk with papers and a pen
Photo: Markus Spiske / Rawpixel

A small business does not need a sixty-page security manual to reduce risk. It needs a short set of decisions that people follow when they are busy, tired or under pressure. The policy is the written part; the risk register is the part that proves the decisions were made with the business in mind.

This guide treats policy as a management instrument rather than a compliance artefact. It shows how to record risk, choose controls, assign evidence and decide what residual risk the owner is willing to accept.

What a policy can and cannot do

A policy cannot stop a determined attacker, and it will not compensate for an unpatched server or an untested backup. It can do three useful things: set defaults so people do not improvise, allocate responsibility so decisions have an owner, and create evidence that the business exercised reasonable care.

Be honest about the boundary. If the policy says multi-factor authentication is required but the finance system does not support it, the document is describing an intention, not a control. The gap belongs in the register with a date and an owner.

Building a register that fits on two pages

Use one row per risk and keep the language operational. A useful structure is: risk, the event that would cause harm, likelihood, impact, existing controls, residual risk, owner and review date. Add a reference to the policy section that governs the control, so the two documents stay connected.

Write the event, not the category. "Ransomware" is a category; "a staff laptop is encrypted and the shared drive has no verified restore point" is an event you can test. Specific events produce specific controls.

Scoring without false precision

A three- or five-point scale is enough. Define likelihood by conditions you can observe: has it happened before, is the exposure internet-facing, is there a known vulnerability, does anyone monitor it? Define impact across operations, cash, legal obligation and reputation. A five-by-five matrix with no definitions invites arguments; a short rubric reduces them.

Record the reasoning in one sentence. "Likelihood 4 because the service is public and has no patching owner; impact 5 because dispatch stops without it" is more useful than a colour. The score is a prioritisation aid, not a measurement.

Seven risks that deserve a line each

  1. Credential compromise: reused passwords, shared logins and missing multi-factor authentication on email, finance and remote access.
  2. Phishing and payment change fraud: a supplier email asks for a new bank account, or an employee approves a payment from a lookalike domain.
  3. Ransomware with untested backups: backups exist but restore has never been attempted, or the backup is reachable from the same account that was compromised.
  4. Unpatched internet-facing services: a website plugin, VPN appliance or remote desktop gateway without an owner and a patch window.
  5. Third-party and remote access: contractors, agencies and former staff retain access longer than the work requires.
  6. Offboarding and privilege creep: accounts are disabled late, shared mailboxes remain accessible and admin rights accumulate.
  7. Data leakage through collaboration and AI tools: customer records pasted into unapproved assistants, or files shared through personal accounts.

Add a privacy line if you hold personal data. Bangladesh's data protection framework and sector rules may impose obligations on notice, consent, retention and breach handling. Legal advice should confirm which apply to your organisation; the policy should at least name who is responsible for answering that question.

Controls, owners and evidence

Each control needs an owner, a frequency and an artefact. A control without evidence is a belief.

  • Identity: multi-factor authentication enforced for email, finance and remote access. Evidence: configuration export and a user list review.
  • Access: quarterly review of admin accounts and supplier access. Evidence: signed review sheet with removals recorded.
  • Patching: a monthly window for servers and plugins, with an emergency path. Evidence: patch log and exceptions.
  • Backup: restore test on a schedule, including one full application restore. Evidence: test record with duration and gaps.
  • Phishing: a reporting route and a short exercise for staff. Evidence: report volume and follow-up actions.
  • Incident response: a one-page contact and decision tree. Evidence: the last review date and a walkthrough record.
  • Offboarding: a checklist completed on the last working day. Evidence: signed checklist per leaver.

Residual risk and the acceptance decision

After the control, some risk remains. The owner should decide whether to accept, reduce further, transfer through insurance or avoid the activity. Record the decision, the rationale, the approver and an expiry date. An accepted risk with no expiry becomes invisible.

This is the most important part of the policy because it converts a technical conversation into a business one. The owner is not agreeing that the risk is small. They are agreeing that the remaining exposure is understood and preferable to the cost of further control.

Policy sections staff will actually use

Keep the document short and put the operational detail in appendices. The sections that earn their place are acceptable use, access and passwords, devices and remote work, data handling, incident reporting, supplier engagement, AI tool use and the annual review. Write the incident section as instructions: who to call, what to preserve, what not to do and how quickly.

Publish it where people work, not only on a shared drive. A policy read once during induction and never again is a training record, not a control.

Questions an owner should ask every quarter

  • Which risk has changed most since the last review, and why?
  • Can we restore the system that matters most, and when did we last prove it?
  • Who has admin access today, and who should not?
  • Which supplier could cause a breach, and what evidence do we hold?
  • What would we do in the first hour of an incident, and who decides?
  • Which accepted risk is now outside its expiry date?

How we support this work

We assess and harden small-business environments through our Managed Technology and Cybersecurity division. The website security assessment service covers the public-facing estate. The pricing page explains how assessment and ongoing support are scoped. For the budget side of the same decision, read what a cybersecurity risk assessment costs in Bangladesh.

Sources

Services that deliver this

Want this implemented properly?

Pricing →

Every engagement starts with a written scope. If a requirement is outside it, we quote it separately before starting.

Responses (0)

No comments yet. Add the first one below.

Leave a response

Your email address will not be published. Required fields are marked *.

Keep reading

More field notes