Understanding Cybersecurity Risk
Run a basic risk assessment: list assets and threats, score likelihood and impact, and rank which risks to treat first.
A long list of worries and a team of two
Neha Kulkarni looks after security for 14 stores, an app with 2.4 lakh customers and 130 staff, with one assistant and a small budget. She could worry about a hundred things, but she can fix perhaps three this quarter. A risk assessment is the method for choosing those three: list what you have, list what could go wrong, score how likely and how harmful each problem is, and rank the results. It turns a vague fear ("we might get hacked") into an ordered to-do list that anyone can question and check.
The casualty ward of a busy government hospital works the same way. Doctors do not treat patients in the order they arrive; a nurse first checks who is in most danger, and that patient is seen first, even if a person with a sprained wrist has waited an hour. This sorting is called triage, and a risk assessment is triage for security problems. It accepts that not everything can be fixed at once and makes sure the most dangerous problems are not stuck at the back of the queue.
A risk assessment is not done once. New systems, new staff and new kinds of attack change the list, so the loop starts again.
Step one: list what you have
Asset identification means writing down the things worth protecting, who is responsible for each one, and which of confidentiality, integrity and availability matters most for it. It sounds dull, and it is the step most often skipped, yet you cannot protect something you have forgotten you own. Neha's first list for Tulsi Health looks like this:
| Asset | Responsible person | What matters most |
|---|---|---|
| Customer database (app, in the cloud) | Rohan Iyer | Confidentiality: prescriptions and addresses |
| 14 billing laptops running MedLedger | Store managers | Availability: no bills means no sales |
| Shared drive with invoices and contracts | Sameer Joshi | Integrity: a fake invoice costs real money |
| Staff email accounts | Neha Kulkarni | Confidentiality, and email resets other passwords |
| Nightly backup | Neha Kulkarni | Availability: the way back after a disaster |
Step two: list what could go wrong
Threat identification asks, for each asset, who or what could harm it, and through which weakness. Write each answer as a full sentence that names the threat, the weakness and the asset. "Phishing" alone is too vague to score. "A fake login email tricks a store employee into giving away their email password" can be scored and fixed. (Phishing is exactly that: a fake message designed to trick someone into handing over a password or money.)
Good places to find threats are the organisation's own history (what went wrong last year), news about similar businesses, warnings from vendors such as Rathi Softech, and a walk around the stores with an attacker's eyes. Ask the staff too. Arjun at Nashik Road knows things about his counter that no one at head office does.
Step three: score likelihood and impact
Likelihood is how probable it is that the risk turns into an incident in a given period, usually the coming twelve months. Impact is how bad it would be if it did: money lost, patients harmed, customers gone, legal trouble, days of disruption. Each gets a score from 1 to 5, using words agreed in advance so that two people scoring the same risk land on similar numbers.
| Score | Likelihood over 12 months | Impact if it happens |
|---|---|---|
| 1 | Very unlikely; never seen here or in similar shops | Minor; fixed in an hour, no customer notices |
| 2 | Unlikely; has happened elsewhere | Small; one store slowed for a day |
| 3 | Possible; has nearly happened here | Moderate; several stores hit or over ₹1 lakh lost |
| 4 | Likely; hits similar businesses every year | Serious; customer data exposed or a chain-wide outage |
| 5 | Almost certain; already happening | Severe; harm to patients or the business at stake |
The risk score is likelihood multiplied by impact, giving a number from 1 to 25. Multiplying, rather than adding, means a risk must be fairly likely and fairly harmful to score high. A frequent nuisance (5 × 1) scores 5, while a possible disaster (3 × 5) scores 15. A family preparing for the monsoon reasons the same way: the leaking roof that drips every July gets fixed first, while the rare flood gets a plan rather than a renovation.
Placing each risk on the grid makes the ranking visible at a glance: the further up and to the right a note sits, the sooner it needs attention.
Step four: rank and decide what to treat first
Risk prioritisation means sorting the risks by score and working from the top. Neha scores five of her risks. Ransomware locking the billing laptops and shared drive is possible (3) and severe (5), so it scores 3 × 5 = 15. A nightly backup failing without anyone noticing scores 3 × 4 = 12, a phishing email stealing a staff email password 4 × 3 = 12, the customer database left open by a settings mistake 2 × 5 = 10, and a billing laptop stolen from a store counter 2 × 3 = 6. She then puts each score into a band people can act on: High from 15, Medium from 8 to 14, and Low below 8. When two risks have the same score, the one with the higher impact goes first, because a worse outcome deserves attention sooner, so the backup failure (impact 4) sits above phishing (impact 3).
If you would like to watch a computer do the same arithmetic, the optional cell below works out all five scores. You do not need to know any Python: press Run and read the results. Then change the phishing likelihood, the 4 marked with a comment, to 2 and run it again; that is the effect of the two-step login described a little further on.
Expected output:
Ransomware: 3 x 5 = 15
Phishing: 4 x 3 = 12
Backup fails: 3 x 4 = 12
Database left open: 2 x 5 = 10
Laptop stolen: 2 x 3 = 6
| Line | What it does |
|---|---|
# Each risk: ... |
A line starting with # is a note for people; the computer ignores it |
risks = [ ... ] |
A list of the five risks. Each line holds a short name, a likelihood and an impact, in that order |
for name, likelihood, impact in risks: |
Repeats the two indented lines below once for each risk, picking up that risk's name and its two numbers |
score = likelihood * impact |
Multiplies the two numbers (* means "times") and calls the answer score |
print(...) |
Shows the risk's name, the multiplication and the score on one line |
The computer's scores match the hand calculation: ransomware alone reaches 15 (High), three risks sit between 10 and 12 (Medium), and the stolen laptop at 6 is Low, which tells Neha where her first time and money should go.
Ransomware leads the list. Three risks sit close together in the middle, so a small change in one score could reorder them.
A ranking is the start of a decision, not the end of one. For each risk near the top, the organisation chooses a response: reduce it with a control (backups that are tested, two-step login), avoid it by stopping the risky activity, share it (insurance, or a contract that makes a vendor responsible) or accept it knowingly when fixing it would cost more than the harm. After a control is added, the risk is scored again, because the whole point of treatment is to move a note down and to the left on the grid.
Try it: Neha turns on two-step login for all staff email. Phishing likelihood drops from 4 to 2. What is the new ranking?
Phishing now scores 2 × 3 = 6, which is Low. The ranking becomes: ransomware 15, backup failure 12, database left open 10, phishing 6 and laptop theft 6. Phishing and laptop theft tie at 6 with the same impact of 3, so they share the bottom of the list. One control moved a Medium risk to Low, and the backup failure, which costs almost nothing to check, is now the obvious next job after ransomware.
Now score some risks yourself. The grid below gives you six more Tulsi Health risks. Give each a likelihood and an impact from 1 to 5 using the scale table, press Rank, and then compare your scores with a reference scoring. Differences of one point are normal; a difference of three means you and the reference imagined a different situation, which is worth thinking about.
Scores are judgements, not measurements, and that is acceptable. Two careful people will rarely give identical numbers, but they will usually agree on which risks belong at the top and which at the bottom, and that agreement is what the method is for. Write down the reason behind each score, as the reference does, so that someone can challenge it later. And put a date on the whole assessment: when Tulsi Health opens its fifteenth store or switches billing vendors, the list of assets changes, and the loop starts again from the first step.