
An IT risk assessment for SMEs: from list to action
Most IT risk assessments end as a spreadsheet of fifty red cells that nobody opens again. A useful one is shorter and sharper: what you protect, what can go wrong with it, how bad that would be, and who does what about it by which date. This is the order we follow.
An IT risk assessment starts with what you protect (information, processes, systems), not with a list of threats from a vendor’s brochure.
Each risk gets one sentence, a likelihood and impact score on a scale you have described step by step, and an owner with a date.
The assessment is only finished when every risk has a decision: reduce, avoid, transfer or consciously accept, and the board signs off the last one.
What an IT risk assessment is, and what it is not
An IT risk assessment answers one question: what can go wrong with our information and systems, how bad would it be, and what are we doing about it? The output is a decision per risk, not a report.
It is often confused with two other things. A security assessment or audit measures the technical state of your environment today; we explain that difference in our piece on the IT security assessment. A business impact analysis decides which processes must come back first after an outage. The risk assessment uses both, but asks something else: where is the biggest gap between what could happen and what you have in place against it?
For an organisation of twenty to a few hundred people, that is a working document of a few pages, not a project of several months.
Start with what you protect
The temptation is to start with threats: ransomware, phishing, a fire. You end up with a list that fits every company and therefore none. Turn it round.
- Which information is critical? Customer data, quotes, the accounts, designs, HR files.
- Where does it live? In Microsoft 365, in the finance package, on a file server, in a mailbox, on the laptops of field staff.
- Who can reach it? Your own people, administrators, suppliers with remote access.
Fifteen to thirty such items are enough for most SMEs. The asset register and the process list from a business impact analysis, if you have one, are a good starting point.
One risk, one sentence
A risk you cannot write down in one sentence is a risk you cannot treat. The form that works: a threat that exploits a weakness and causes a consequence.
“Cyber attack” is not a risk. “A member of staff clicks a phishing email, and because there is no multi-factor authentication on the finance package, an attacker can redirect payments” is. The sentence tells you straight away where the control belongs: not only with the person who clicked, but with the missing second factor.
Write down the risks that have nothing to do with hackers too: the only administrator handing in his notice, a supplier going bust, a backup nobody has ever restored, a laptop left on the train.
Scoring likelihood and impact without false precision
A score of 1 to 5 for likelihood and for impact is common. The trouble is that everyone means something different by a “3”. So describe each step of the scale in words that fit your organisation before anyone starts scoring.
- Likelihood: from “not expected in the next few years” to “already happening, or almost certain within a year”.
- Impact: described in hours of downtime, the number of customers affected, whether you would have to report a data breach, or damage to your reputation. Not in a pound figure you cannot back up.
Score with a few people at once, from IT and from the business. Where scores differ widely, one side usually knows something the other does not. Multiply likelihood by impact for a ranking, but always read high impact with low likelihood separately: that is where the emergency procedures belong.
From risk to control
This is where most assessments stall. Every risk above your threshold gets a decision, and there are only four:
- Reduce: a control that lowers likelihood or impact. Multi-factor authentication, a tested backup, network segmentation, a second administrator.
- Avoid: stop doing what causes the risk, for example switching off an outdated system instead of continuing to protect it.
- Transfer: place part of the consequence with someone else, through a contract or insurance. The risk itself stays yours.
- Accept: consciously do nothing, because the control costs more than the risk. That is a legitimate decision, provided it is written down and signed by someone with the authority to do so.
Give every control an owner and a date, and estimate the residual risk afterwards. A control without a name is an intention.
What UK GDPR, NIS and ISO 27001 expect
A risk assessment is not just sensible; it is increasingly asked for. UK GDPR requires security appropriate to the risk in Article 32, and you cannot justify that without an assessment. The UK has its own NIS Regulations for operators of essential services and certain digital services, separate from the EU’s NIS2 directive; if you supply customers in the EU, NIS2 may still reach you through their supply chain requirements. ISO/IEC 27001 requires a documented method for assessing and treating risk.
The NCSC publishes risk management guidance and Cyber Essentials sets out a baseline of technical controls; both are useful to hold against your own register.
What an auditor or regulator wants to see is not the prettiest heat map but the thread: risk, decision, control, owner, check. What ISO 27001 asks you to show is set out in our ISO 27001 checklist.
Keeping it current, and the traps we see most
A risk register is a working document. Review it at least once a year, and also at every major change: a new system, an office move, an acquisition, a new supplier with access, or an incident.
- Everything is red. Then the scale has not been described or nobody dares to choose. A register without green tells you nothing.
- Only technical risks. People, suppliers and processes are missing, while that is often where the biggest gaps are.
- IT owns everything. The risk that invoicing stops belongs to whoever runs invoicing. IT owns the control, not the business risk.
- No check on the control. “Backup configured” is not the same as “backup restored and verified this quarter”.
Where our work stops
What we do. We help set up the assessment and deliver the technical part: mapping where information lives and who can reach it, naming technical weaknesses, proposing and implementing controls, and showing that they work. We are ISO/IEC 27001 certified and keep such a register ourselves.
What we don’t do. We do not own your risks: which residual risk is acceptable is for your board to decide and sign. We are not a certification body and do not judge whether your assessment meets ISO 27001 or any regulation; an accredited auditor does that. We are not lawyers and do not tell you whether a regulation applies to you. And we are not an insurer: which risk you transfer to a policy is a conversation with your own insurer.
Questions we get about this
The five that come up most.
How do you carry out an IT risk assessment?
Start with what you protect: the critical information, where it lives and who can reach it. Write each risk as a threat, a weakness and a consequence, score likelihood and impact on a scale you have described in advance, and take a decision per risk with an owner and a date. Then review the register at least once a year.
What is the difference between a risk assessment and an IT audit?
An audit or security assessment measures the technical state of things at a point in time. A risk assessment weighs what could go wrong against what you have in place and ends in decisions. An audit is good input for a risk assessment, but does not replace it.
What are the steps of a cyber security risk assessment?
Identify what you protect, describe the risks to it, score likelihood and impact, decide per risk whether to reduce, avoid, transfer or accept it, assign an owner and a date to each control, and review. Most of the value is in the last three steps, which is also where most assessments stop.
How often should an IT risk assessment be repeated?
At least once a year, and also at every major change: a new system, an office move, an acquisition, a new supplier with access, or after an incident. A register updated only for the audit is out of date for the rest of the year.
Who owns an IT risk?
The business risk belongs to whoever runs the process, and the decision to accept a residual risk belongs to the board. IT, or the IT provider, owns the technical control. If everything is left to IT, you get decisions nobody in the business actually took.
Further reading
Where the risk assessment connects.
A risk assessment that leads to action?
Tell us where you stand; we will map the technical side and tell you honestly what needs doing first.
Practical IT knowledge in your inbox
New guides on management, security and the workplace, written by the people doing the work. No sales talk, and you can unsubscribe in one click.