Leaving this course?

Lesson 1 of 12

Why a supplier is an attack path

Supply-chain risk is usually introduced as a list of controls. It is better introduced as a piece of arithmetic: an attacker who compromises one supplier reaches all of that supplier's customers, so the return on a single intrusion is multiplied by the size of the customer base.

The structure, defined. ENISA's analysis is precise about the shape: "A supply chain attack is a combination of at least two attacks. The first attack is on a supplier that is then used to attack the target to gain access to its assets." Two attacks, two victims, and the first victim is often not the one the attacker cares about.

That matters for how you read an incident. A supplier that reports "we were breached, no customer data was affected" may be describing the first attack while the second is still in progress.

What gets targeted at the supplier. Of the attacks ENISA analysed, the supplier assets involved were "Code (66%), Data (20%) Processes (12%)". Code dominates because code is what gets distributed: a change made once at the supplier is installed by every customer, usually automatically, usually with high privilege, and usually without anyone reading it.

This is worth sitting with. The mechanism that keeps you secure — applying updates promptly — is the same mechanism that distributes a compromise. There is no version of good practice that removes this, which is why the answer lies in provenance and blast radius rather than in slowing updates down.

Four routes, and they need different controls:

  • Software you install. A compromised update, a tampered package, a dependency with a maintainer who is no longer the maintainer.
  • Services you depend on. A provider that holds your data or runs your process. Here the attacker does not need to reach your network at all.
  • Access you granted. A support contractor, a managed service provider, an integrator with standing credentials into your systems.
  • Hardware and devices, where the supply chain is long, opaque and largely undocumented.

Where this course sits. The security of software as it is built — weaknesses, memory safety, bills of materials from the developer's side, and the Log4j event — is covered in the application security course. This one is about the supplier as a counterparty: knowing who they are, what you have agreed, what happens when they fail, and what the law now requires of you rather than of them.

One category deserves singling out. A managed service provider holds administrative credentials into many customers' environments at once, by design, because that is the service. It is the only supplier type that combines deep privileged access with the one-to-many property, which makes it the highest-value target in this whole subject and the one where a single compromise has produced the largest numbers of downstream victims.

If you use one, the questions in lesson four are not proportionate for them. They need the treatment in lesson three's top tier, whatever they cost.

The sentence to carry through all twelve lessons. You can outsource the work and you cannot outsource the obligation. Every regime in lesson eleven says a version of that, and every enforcement action turns on it.

Lesson 2 of 12

Knowing who your suppliers are, and who theirs are

Every programme in this field fails at the same first step, and it is not a sophisticated one: the organisation cannot say who it depends on.

The requirement is explicit. The NIST Cybersecurity Framework 2.0 added a whole category for this, and its first substantive outcome is blunt: "GV.SC-04: Suppliers are known and prioritized by criticality." Known, and then ranked. Most organisations have the first half partially and the second not at all.

A vendor list is not a dependency map. Procurement holds a list of everyone who sends an invoice. That list is simultaneously too long — the stationery supplier is on it — and too short, because it omits:

  • free and open-source dependencies, which send no invoice and may be in everything you run;
  • services procured on a card by a team that needed something quickly;
  • anything inherited through an acquisition, where the contracts came with the company;
  • the provider behind the provider, which is the subject of lesson six.

What you need is not a longer list but a different question: for each business service you must keep running, what does it actually depend on, and what happens if each of those stops?

The register, as regulation now defines it. DORA requires financial entities to "maintain and update at entity level, and at sub-consolidated and consolidated levels, a register of information in relation to all contractual arrangements on the use of ICT services provided by ICT third-party service providers." Two details in that sentence do real work: it is all contractual arrangements, not the material ones, and it is maintained at group level as well as entity level, which defeats the common arrangement where each subsidiary knows its own suppliers and nobody knows the overlap.

The implementing regulation that specifies the templates also explains why the data is wanted: the register's information "is essential for the financial entities' internal ICT risk management, for the effective supervision of the financial entities by their competent authorities". It is a supervisory instrument as well as an internal one.

What an entry has to answer. The name of the supplier, what service they provide, which business service depends on it, who owns the relationship internally, whether the service supports a critical or important function, where the data is processed, who the known sub-contractors are, when the contract ends, and what the exit arrangement is.

And keep it where the work happens. A register maintained once a year by a risk function, from questionnaires, is out of date on the day it is published. One updated by procurement at the point of contract, and by architecture at the point of integration, has a chance.

Lesson 3 of 12

Criticality: deciding which suppliers matter

Treating every supplier alike is the most expensive mistake available here. It spreads limited assurance effort evenly across a population where the risk is wildly uneven, and it produces a programme that reviews the stationery contract to the same depth as the core banking platform.

Criticality is about impact, not spend. The two correlate badly. A small monitoring tool with an agent on every server, bought for a few thousand a year, has deeper access than a large outsourced print contract. The question is what the supplier can reach and what stops working without them — not what they cost.

A definition worth adopting. The Financial Stability Board's toolkit defines a critical service as "A service provided to a financial institution whose failure or disruption could significantly impair a financial institution's viability, critical operations, or its ability to meet key legal and regulatory obligations." Three limbs, and the third is the one organisations forget: a supplier whose failure stops you meeting a regulatory obligation is critical even if the business could trade through it.

The two axes to score, separately. They get conflated and they call for different controls.

  • Access. What can this supplier reach? Administrative credentials, an agent on every endpoint, a network connection, customer data, the ability to push code. A supplier with deep access is a security concern even if losing them would cause no disruption at all.
  • Dependency. What stops if they stop? How quickly, and is there an alternative? A supplier with no access but no substitute is a resilience concern rather than a security one.

A supplier high on both is where assurance effort belongs. One high on access only needs its permissions cut. One high on dependency only needs an exit plan and a tested alternative.

Tier, then differentiate the work. The output of this exercise is not a score; it is a different process per tier. The top tier gets full due diligence, contractual audit rights, continuous monitoring and a tested exit plan. The bottom tier gets a standard clause set and nothing else. Without that differentiation, the tiering is decorative.

Beware "critical" as a negotiation. Business owners understate criticality when the assurance process is painful, because a low rating means a faster purchase. The tell is a supplier rated low whose service appears in the business continuity plan, or whose outage would be reported to a regulator. Reconcile the supplier tiering against the continuity plan once a year and the inconsistencies are the finding.

And review the rating on a trigger. Criticality changes without anyone revisiting the form: a pilot becomes the production system, a second product is added, a supplier acquires a competitor you also use — which quietly creates concentration where there was diversity.

Lesson 4 of 12

Due diligence that tells you something

Most supplier due diligence produces a completed questionnaire and no information. The questions are closed, the answers are self-reported, and nobody reads them against what they already know.

Evidence, not assurance. The distinction is the whole lesson. "We have an information security policy" is assurance. A current certification, naming the scope, with a date, is evidence. An independent test summary with findings and their remediation status is better evidence. The shift from the first to the second changes the conversation from a form to a review.

Certifications have a scope, and the scope is the point. An organisation certified to a management-system standard has had a system audited, not every product and location. The question is always what was in scope: which entities, which services, which sites, over what period, and what exclusions are recorded. A certificate covering the head office does not cover the development centre where your code is written.

The same applies to audit reports: read the scope section and the exceptions, and if neither is available, you have been shown a logo rather than a result.

What to actually ask for, in rough order of how much it tells you:

  • The scope and exceptions of their most recent independent audit or certification.
  • How they manage their own suppliers, which is lesson six and the question most likely to be answered badly.
  • Their incident history and how they notified customers, including an incident they handled badly. A supplier who cannot name one is either very lucky or not answering.
  • Named security contact and escalation route, tested before you need it.
  • Where data is processed and stored, by country, and whether that changes with load.
  • What access their staff have to your data, and how it is granted, reviewed and removed.
  • Their patching and vulnerability response commitments, in time rather than in adjectives.

How far down to go. The NCSC's supply chain principles put the judgement plainly: "You will need to think about how far down your supply chain you need to go to gain understanding and confidence in your suppliers." There is no universal depth. For a critical supplier it is at least their critical sub-contractors; for the rest it is the contractual obligation to pass requirements down, and nothing more.

Proportionality is not a loophole. A small supplier cannot produce what a global provider can, and demanding it either excludes capable small firms or produces documents written to pass rather than to describe. The workable position is to accept smaller evidence from smaller suppliers and to reduce what they can reach instead — a supplier with narrow access needs less assurance, which is a control you hold rather than a question you ask.

A note on questionnaire fatigue, because it is the reason the answers are poor. A supplier serving two hundred customers may receive two hundred different questionnaires asking the same things in different words, and will answer them with a copy-and-paste team rather than the engineers who know. That is a rational response to an irrational demand, and it is why shared and standardised assessment schemes exist in several sectors.

Using one costs you some specificity and buys an answer somebody actually thought about. Where a scheme exists for your sector, the sensible pattern is to accept its assessment as the baseline and spend your own questions on what is specific to you: what this supplier can reach in your estate, and what breaks if they stop.

Lesson 5 of 12

What the contract has to carry

The contract is where supplier risk management either becomes enforceable or stays aspirational. Everything you want from a supplier during an incident has to have been agreed before one.

The framework states it as an outcome. NIST CSF 2.0: "GV.SC-05: Requirements to address cybersecurity risks in supply chains are established, prioritized, and integrated into contracts and other types of agreements with suppliers and other relevant third parties." The operative words are prioritized and integrated: a standard clause set applied to everyone, including the stationery supplier, is how organisations end up negotiating audit rights they will never exercise while omitting the one thing they needed.

The clauses that earn their place:

  • Security requirements, specific enough to test. Patch timescales in days, encryption of data at rest and in transit, multi-factor authentication for anyone with access to your environment, and the standards they will maintain.
  • Breach and incident notification, with a clock. Not "promptly". A number of hours, counted from the supplier's awareness, with a named route that does not depend on your account manager being at their desk. The reason is in lesson eleven: your own notification deadlines start running whether or not your supplier has told you.
  • The right to look. DORA requires, for ICT services supporting critical or important functions, "unrestricted rights of access, inspection and audit by the financial entity, or an appointed third party, and by the competent authority." Note the third limb — the supervisor's own right of access, which a supplier must accept rather than negotiate.
  • Sub-contracting consent and disclosure, covered in the next lesson.
  • Requirements passed down. The NCSC principle is specific: "Include your minimum security requirements in the contracts you have with suppliers and in addition, require that your suppliers pass these down to any sub-contractors they might have." Without that sentence your requirements stop at the first tier.
  • Data location and transfer, because a change of processing location can create a cross-border transfer you have not assessed.
  • Return or deletion of data at termination, with evidence, and a stated format for the return.
  • Exit assistance, which lesson eight explains is the clause most often missing.

The right to audit is worth having and rarely used. That is not an argument against it. Its value is partly that it exists: a supplier who knows you may inspect behaves differently from one who knows you cannot. But a right exercised never is a right that has not been tested, and the first time you use it should not be during an incident.

And read what you already signed. Most organisations' real exposure sits in contracts agreed years ago, before any of this was standard, renewed automatically ever since. The renewal date is the only moment of leverage you get, and it usually passes unnoticed by anyone in risk.

Lesson 6 of 12

Sub-outsourcing: the chain you did not sign

You contracted with one company. The work may be performed by three, and your contract reaches only the first.

The regulators have named it. The PRA's outsourcing statement puts the risk plainly: "Sub-outsourcing, which is also sometimes referred to as 'chain' outsourcing, can amplify certain risks in material outsourcing". Amplify, not create — the risks are the ones you already assessed, now with less visibility and no contractual relationship.

And defined it. The DORA implementing regulation defines an "ICT service supply chain" as "a sequence of contractual arrangements connected with the ICT service being provided by the direct ICT third-party service provider to the financial entity". A sequence of contracts, each between two other parties, delivering something you depend on.

Why it is harder than first-tier risk. Four reasons:

  • You have no contract, so no audit right, no notification clause and no standards obligation, except whatever your direct supplier passed down.
  • You may not know they exist. Sub-contracting can change without notice unless the contract requires notice, which is why the consent clause matters more than the disclosure clause.
  • Your supplier's assessment of them is all you have, and it is as good as their own programme.
  • Concentration hides here. Four suppliers who look independent may all run on one platform, use one authentication provider, or sub-contract to the same specialist. Diversity at the first tier can be an illusion at the second.

What to require, realistically:

  • Notice and consent for material sub-contracting, with the right to object. A right to be told after the fact is not a control.
  • Disclosure of the chain for critical services, naming the sub-contractors and what each performs.
  • Flow-down of your security requirements, as the previous lesson sets out.
  • Your supplier's own due diligence on them, available on request.
  • Notification obligations that survive the chain, so an incident two tiers down reaches you.

The nth-party problem, and one practical handle. You cannot assess a chain of unbounded depth, and pretending otherwise produces a programme that collapses under its own paperwork. The FSB toolkit offers a concrete approach for software: "An illustrative example of the information that service providers may provide is a software bill of materials from software suppliers." A list of components is a partial answer to "what is in this", obtained from the supplier rather than reconstructed by you. The developer's side of bills of materials is covered in the application security course; here it is a procurement artefact — something to ask for, store with the register, and query when a new vulnerability is announced.

Where to stop. For critical services, the first tier in depth and the named critical sub-contractors by disclosure. Below that, contractual flow-down and notification. An organisation that tries to assess the fourth tier of a non-critical supplier has mistaken effort for control.

Lesson 7 of 12

Concentration and substitutability

Diversification is the oldest risk control there is, and technology procurement has spent two decades doing the opposite for good reasons: fewer suppliers mean better integration, lower cost and simpler management. The result is that many organisations, and whole sectors, now depend on a small number of providers.

The two forms, and they need different answers.

  • Concentration within your own estate. Many critical services resting on one provider. The FSB toolkit describes the measure as "The number of critical services supported by a single or closely connected service providers (sometimes referred to as 'aggregation')" — and note closely connected, which catches two suppliers that look separate and share an owner or a platform.
  • Concentration across the market. Everyone depending on the same provider, so its failure is a sector event rather than a company one. No single institution can fix this, which is why supervisors have taken an interest in it directly.

Substitutability is the question that makes it actionable. DORA requires a financial entity to assess, among its risks, "contracting an ICT third-party service provider that is not easily substitutable". That phrasing is useful because it is answerable: could you move, to whom, how long would it take, and what would it cost? A supplier that is not easily substitutable is not forbidden — it is a position you have to take deliberately and be able to justify.

What reduces it, in descending order of realism:

  • A tested alternative for the most critical service. The FSB toolkit names the practical version: "designating a primary and backup service provider for a critical service; and maintaining a list of viable potential alternative service providers." A list of alternatives is cheap; a tested one is not, and only the second tells you anything.
  • Portability by design. Standard formats, exportable data, avoiding features that have no equivalent elsewhere. This is paid for at build time and cannot be retrofitted cheaply.
  • Accepting the dependency, with eyes open. Sometimes there is no alternative of comparable quality, and the honest answer is a documented acceptance at the right level of seniority, with a reduced-service plan for an outage rather than a substitution plan.

The thing not to do. Add a second supplier for appearance, with no traffic, no tested failover and no current integration. It costs money, adds a second set of credentials and a second attack surface, and delivers nothing on the day, because nobody has ever switched to it.

And look for the hidden single points. NIST's supply chain practices put the diversification expectation and the diagnostic together: enterprises "should diversify their supply base, especially for critical ICT/OT products and services", and as part of that exercise "should attempt to identify single points of failure and risk among primes and lower-level entities in the supply chain." Lower-level entities. The single point is usually one tier down, where nobody drew the diagram.

Lesson 8 of 12

Exit, and the plan nobody tests

Exit planning is the part of third-party risk management that is most often written and least often tested, and the gap between the two is where the loss happens.

It is now a requirement, not good practice. DORA: "For ICT services supporting critical or important functions, financial entities shall put in place exit strategies." The implementing standard goes further and names the two things that make a strategy real: "The policy shall contain requirements for a documented exit plan for each contractual arrangement and for the periodic review and testing of the documented exit plan." Per arrangement, reviewed, and tested.

Two kinds of exit, and planning for only one is the common failure.

  • An orderly exit. You chose to leave, there is a notice period, the supplier is cooperative, and there is time to migrate.
  • A stressed exit. The supplier has failed, been breached, lost a licence, been acquired by someone unacceptable, or simply stopped answering. There is no cooperation and no notice.

The PRA's statement is explicit that the second must be covered: "Firms' exit plans should cover stressed exits and be appropriately documented and tested as far as possible." Most exit plans assume the supplier will help, which is precisely the assumption a stressed exit removes.

What a usable plan contains:

  • Where the data is and how it comes back, in what format, and whether you have ever received a full export. An export that has never been tested may be unusable, incomplete, or take three weeks.
  • What runs the service in the meantime — an alternative, a degraded manual process, or an accepted outage with a stated duration.
  • Who does the work. Exit is a project, and in a stressed exit it competes with incident response for the same people.
  • What the contract obliges the supplier to do, which is the exit assistance clause from lesson five and the reason it belongs in the contract rather than in the plan.
  • Knowledge that is not written down. The operational knowledge of a long-standing supplier is a dependency of its own, and it leaves with them.

The framework's version of the same point. CSF 2.0: "GV.SC-10: Cybersecurity supply chain risk management plans include provisions for activities that occur after the conclusion of a partnership or service agreement." After the conclusion — the access that must be revoked, the data that must be returned or destroyed, the credentials and certificates still trusted, the firewall rules still open.

A test worth running once. Pick one critical supplier and ask the owner three questions without warning: where is our data, how would we get it, and what would we do on Monday. The quality of the answers is the state of your exit planning, whatever the document says.

Lesson 9 of 12

Monitoring a supplier after you have signed

Due diligence is a photograph taken at the moment of least risk — when the supplier is trying to win the business. Everything that matters afterwards happens without that incentive.

What changes after signature. The supplier is acquired. The team that served you is reassigned. A sub-contractor is added. A certification lapses. The service is re-platformed. Their own key supplier fails. None of these triggers a conversation unless something in your process is watching for them.

The right to ask has to be in the contract, and used. The FSB toolkit lists among the contractual tools a "Financial institution's right to access, audit and obtain relevant information from the service provider, as appropriate including information on supply chain risk management" — and the last clause is the one worth noticing, because it extends the right beyond their own controls to how they manage their suppliers.

What ongoing monitoring actually consists of, in descending order of value for effort:

  • Evidence renewal with an expiry date. Certifications and audit reports have dates. Track them like certificates and chase them before they lapse, because a lapsed certification is sometimes the first visible sign of trouble inside a supplier.
  • Change notifications you actually read. Sub-contracting changes, ownership changes, location changes. Most contracts oblige the supplier to tell you; few organisations have anyone whose job is to receive it.
  • Re-answering the questionnaire, and comparing. The value is not the new answers; it is the differences from last time. An answer that moved from yes to no, or a scope that narrowed, is worth more than a hundred unchanged rows.
  • Their incidents, and how you heard. Did they tell you, within the contractual window, through the agreed route? Or did you read it in the press? That one data point is the best available measure of the relationship's real quality.
  • Service and security performance against what was agreed, including patch timescales, which are frequently in the contract and never measured.
  • Access reviews. Who at the supplier still has credentials into your systems, and should they. This is the cheapest and most productive review in the whole discipline, and it almost always finds accounts belonging to people who left.

Watch the quiet suppliers. Attention follows noise, so a supplier with outages gets reviewed and a supplier that works silently for six years does not. The second may be running on unpatched infrastructure with four former employees' accounts still active. Review on criticality and on trigger, not on how much trouble a supplier has caused lately.

And treat a refusal as information. A supplier who will not answer, will not allow an audit they contractually agreed to, or whose answers become vaguer over time is telling you something that no questionnaire score captures. Record it, escalate it, and let it count in the tiering.

Lesson 10 of 12

SolarWinds: what one compromise reached

The clearest illustration of the arithmetic in lesson one, documented in a public advisory rather than a vendor report.

What happened, as CISA described it. The advisory records that the compromises "began at least as early as March 2020" and identifies the route: "One of the initial access vectors for this activity is a supply chain compromise of a Dynamic Link Library (DLL) in the following SolarWinds Orion products". A component of a network monitoring product, distributed as a signed update, installed by customers as routine maintenance.

Consider what that product is. Network monitoring software sits everywhere, talks to everything, and holds credentials for the systems it monitors — which makes it close to the ideal place to put an implant. The attacker did not choose it because the vendor was careless; they chose it because of where it sits.

The scale of the response. "On December 13, 2020, CISA released Emergency Directive 21-01: Mitigate SolarWinds Orion Code Compromise, ordering federal civilian executive branch departments and agencies to disconnect affected devices." A national authority ordering an entire government to disconnect a commercial product is a measure of how little could be assumed about which installations were affected.

And the assessment of the actor. "This adversary has demonstrated an ability to exploit software supply chains and shown significant knowledge of Windows networks." Not an opportunist — a capability deliberately developed, because the one-to-many property makes it worth developing.

The five lessons that generalise, and each maps onto an earlier lesson of this course:

  • A signed, authentic update from a legitimate vendor was the attack. Signature verification worked correctly and proved the wrong thing: that the update came from the vendor, which it did. Provenance answers "who built this", not "is it safe".
  • The victims had done nothing wrong. They applied updates promptly, which is what everyone is told to do. No amount of customer-side hygiene would have prevented it.
  • Blast radius was the only control that mattered. What limited the damage at any given organisation was what the product could reach from where it ran — a question of segmentation and privilege, decided long before the incident.
  • Detection came from outside. Most affected organisations learned from a third party, which is the normal outcome and should be planned for rather than resented.
  • The question "which of our suppliers could do this to us" is answerable today. Make a list of every product with an agent, an integration or standing credentials in your estate. That list is your supply-chain attack surface, and almost nobody has written it down.

One caution about reading cases. It is tempting to conclude that the answer is to distrust vendors. It is not available: you cannot run an estate without their software. The usable conclusion is narrower and harder — reduce what each one can reach, and know which ones could hurt you most.

Lesson 11 of 12

What the law now requires

Third-party cyber risk moved from good practice to legal obligation across several regimes in quick succession, and the common thread is that responsibility stays with the customer.

DORA, for financial entities in the EU. Regulation (EU) 2022/2554 applies from 17 January 2025, and its central provision on this subject is unambiguous: financial entities with contractual arrangements for ICT services "shall, at all times, remain fully responsible for compliance with, and the discharge of, all obligations under this Regulation and applicable financial services law." Fully responsible, at all times. Alongside it sit the register of information from lesson two, the exit strategies from lesson eight, and the audit rights from lesson five. DORA also establishes supervision of ICT third-party providers designated as critical — a regime that reaches the provider directly, which is new.

NIS2, more broadly. Directive (EU) 2022/2555 required member states to apply its measures from 18 October 2024, and names supply chain security as a required measure: "supply chain security, including security-related aspects concerning the relationships between each entity and its direct suppliers or service providers." It then says how the assessment must be done: entities must "take into account the vulnerabilities specific to each direct supplier and service provider and the overall quality of products and cybersecurity practices of their suppliers and service providers, including their secure development procedures."

Read that against lesson three. It is a per-supplier assessment, not a policy, and it reaches their development practices.

NIS2 also sets a reporting clock worth knowing: an early warning to the authorities "within 24 hours of becoming aware of the significant incident." Twenty-four hours, which is why the supplier notification clause in lesson five needs a number smaller than that.

Data protection, which applies to almost everyone. Where a supplier processes personal data on your behalf, the relationship is regulated directly. Data protection laws commonly require a written contract between controller and processor, with terms they specify. The Data Protection Act, 2019 in Kenya, for example, requires that the controller and processor "enter into a written contract which shall provide that the data processor shall act only on instructions received from the data controller and shall be bound by obligations of the data controller."

And the notification periods interlock in a way worth reading twice. The controller must "notify the Data Commissioner without delay, within seventy-two hours of becoming aware of such breach" where there is a real risk of harm. The processor must tell the controller "without delay and where reasonably practicable, within forty-eight hours of becoming aware of such breach."

Forty-eight of your seventy-two hours can lawfully be consumed by your supplier before you know. That is the arithmetic that should set your contractual notification window, and it is the clearest practical argument in this course for negotiating a shorter one. A second national example, where that arithmetic does not exist. India's Digital Personal Data Protection Act, 2023 places the duty on the Data Fiduciary and gives the Data Processor no statutory clock at all. Section 8(1) makes the Fiduciary responsible for complying with the Act "irrespective of any agreement to the contrary" for any processing done "on its behalf by a Data Processor"; section 8(2) allows a Processor to be engaged "only under a valid contract"; and section 8(6) requires the Fiduciary to tell the Board and each affected person, with no number of hours fixed. So where one law hands you a supplier deadline to rely on, another hands you none and forbids you to contract the responsibility away. Either way the lesson is the same, and it is the contract that has to carry it.

And the baseline is being measured. The Central Bank of Kenya's 2025 survey of the banking sector's adoption of its cybersecurity guidance reports that "95 percent of commercial banks assess and manage cybersecurity risks posed by third-party vendors, while 5 percent do not." Supervisors are now counting, which tends to precede requiring.

Lesson 12 of 12

Running the programme, and what to measure

What a working third-party cyber risk programme looks like when nothing is on fire.

The discipline has a name and a definition. NIST's practices document defines it: "C-SCRM is the organized and purposeful management of cybersecurity risks throughout the supply chain." Organised and purposeful, which distinguishes it from the common arrangement — a questionnaire triggered by procurement, a spreadsheet nobody owns, and a review when a supplier makes the news.

Where it has to sit. Not in security alone, which has no leverage over contracts, and not in procurement alone, which has no view of technical exposure. The decisions require both, plus legal for the clauses and the business owner who holds the relationship. The practical arrangement that works is a single owner for the process and a named owner for each critical supplier.

Build it into the moments that already exist. A programme that depends on people remembering to engage it will be bypassed. The hooks are: the purchase requisition, the integration or access request, the contract renewal date, and the architecture review. Anything procured without passing one of those is the gap, and the cloud bill is where you find it.

Two requirements worth adopting verbatim. On diversity, NIST: enterprises "should diversify their supply base, especially for critical ICT/OT products and services." On the chain, the same document's notification control: "At minimum, enterprises should require their suppliers to establish notification agreements with entities within their supply chain that have a role or responsibility related to that critical service or product." That is the mechanism by which an incident two tiers down reaches you — an obligation on your supplier to have the same obligation on theirs.

What to measure. Not the number of questionnaires issued:

  • Coverage. What proportion of critical suppliers have a current assessment, and how many are overdue, and by how long. Overdue age is the number that moves.
  • Register accuracy. How many suppliers were found that were not on the register, in the last period. A programme that never discovers one is not looking.
  • Notification performance. For each supplier incident, did you hear from them, inside the contractual window, through the agreed route — or from somewhere else.
  • Access hygiene. Supplier accounts in your estate with no current business need, and how long after a contract ended that access persisted.
  • Exit readiness. How many critical suppliers have an exit plan, and how many of those plans have been tested rather than written.
  • Undisclosed sub-contracting found, which measures whether the clause in lesson six is doing anything.

How far to go, settled. The NCSC's principles decline to give a universal depth and put the judgement where it belongs: "You will need to think about how far down your supply chain you need to go to gain understanding and confidence in your suppliers." Depth by criticality, flow-down by contract, and an honest record of where you stopped and why.

And the sentence from lesson one, which is the whole subject. You can outsource the work. The obligation stays.

Knowledge check

Fifty questions

Answer all fifty questions, then check your answers. You need 45 out of 50 to pass and receive a certificate. If you score less, you will see which answers were right and wrong, and then go through the course again before you retake the check. Your answers, progress and times are kept only in this browser.

Standards and frameworks covered

The published frameworks this course teaches toward. Saolix is not affiliated with, endorsed by or accredited by any of these bodies; naming a standard is a citation, not a claim about its author.

National detail, by market

This course teaches the international standard and one worked example. The national detail for a market sits in its own pack, so it can be kept current without reopening the course: India, Japan.

Sources

The official documents this course relies on. Laws and guidance change, so check the current version.

  1. Threat Landscape for Supply Chain Attacks (July 2021) · European Union Agency for Cybersecurity (ENISA)
  2. The NIST Cybersecurity Framework (CSF) 2.0, including the GV.SC category (CSWP 29, February 2024) · US National Institute of Standards and Technology (NIST)
  3. SP 800-161r1-upd1: Cybersecurity Supply Chain Risk Management Practices for Systems and Organizations (updates as of November 2024) · NIST
  4. Regulation (EU) 2022/2554 (Digital Operational Resilience Act), Articles 28 to 30 · Official Journal of the European Union, via EUR-Lex
  5. Commission Delegated Regulation (EU) 2024/1773 on the policy for ICT services supporting critical or important functions · European Commission, via EUR-Lex
  6. Commission Implementing Regulation (EU) 2024/2956 on the standard templates for the register of information · European Commission, via EUR-Lex
  7. Directive (EU) 2022/2555 (NIS2), Articles 21 and 23 · Official Journal of the European Union, via EUR-Lex
  8. Enhancing Third-Party Risk Management and Oversight: a toolkit for financial institutions and financial authorities (December 2023) · Financial Stability Board
  9. Supervisory Statement SS2/21: Outsourcing and third party risk management (November 2024 update) · Prudential Regulation Authority, Bank of England
  10. Supply chain security guidance: the principles of supply chain security · UK National Cyber Security Centre (NCSC)
  11. Alert AA20-352A: Advanced Persistent Threat Compromise of Government Agencies, Critical Infrastructure, and Private Sector Organizations · US Cybersecurity and Infrastructure Security Agency (CISA)
  12. Data Protection Act, 2019 (No. 24 of 2019), sections 42 and 43 · Office of the Data Protection Commissioner, Kenya
  13. Survey on the Adoption of the 2017 CBK Guidance on Cybersecurity to the Banking Sector (June 2025) · Central Bank of Kenya