Certa is named a Leader in the 2026 Gartner® Magic Quadrant™ for Third-Party Risk Management Tools. Get the report

Blog/Vendor Risk Management

How to tier vendors by risk: a criticality model for regulated enterprises

Two questions decide every vendor's tier

Criticality asks what breaks if the vendor fails. Inherent risk asks what the vendor can touch. Score both, map them to four tiers, and let the tier set the depth of everything after it.

  • 50% Lower Costs at Uber50% Lower Costs
  • 130+ Integrations130+ Integrations
  • Gartner® MQ Leader, 2026Gartner® MQ Leader
  • Onboarding up to 3x Faster3x Faster Onboarding
CertaNovember 19, 2026 · 15 min read
Two colleagues pointing with pens at printed charts spread across a desk

Ask ten people in your company which vendors are critical and you get ten lists. Procurement ranks by spend. InfoSec ranks by data. The business names whoever it called last week. Then an examiner asks for the list, with the reasoning behind it, and the program has to pick one.

Regulators leave the method to you and hold you to the result. The Interagency Guidance on Third-Party Relationships, issued by the OCC, FDIC and Federal Reserve in June 2023, opens its risk management section with the principle that drives every tiering model: "Not all relationships present the same level of risk, and therefore not all relationships require the same level or type of oversight or risk management." It names no tier count, no scoring scale and no template.

What follows is a working model in six steps: inventory by service, assess criticality, score inherent risk, map both into tiers, attach diligence and monitoring depth to each tier, and retier when the facts change. Every threshold below is a starting point to tune against your own risk appetite.

The Short Answer

Tier vendors on two separate measures: how critical the service is to your operations, and how much inherent risk the engagement carries through data, access, regulation and geography. Any vendor supporting a critical activity goes to Tier 1. Everyone else lands in Tier 2, 3 or 4 by inherent risk rating, and the tier sets due diligence depth, monitoring and reassessment cadence.

What you need before you start vendor risk tiering

Four inputs the model can't run without

Most vendor risk tiering efforts stall on inputs, long before anyone argues about scoring. The model below assumes you can put your hands on four things. If one is missing, build it first. A tier assigned against a guessed list of critical activities will not survive its first challenge from internal audit.

A list of critical activitiesThe business services your firm has already named critical, from resilience, BCM or the risk committee. Under DORA, your critical or important functions.
A board-approved risk appetiteThe statement that tells you how much third-party exposure is acceptable, so thresholds trace to something the board signed.
Named business ownersOne accountable person per vendor relationship who can answer what the service does and what stops without it.
Source systems for the inventoryAccounts payable, contract repository, SSO and procurement intake, so the list reflects who you actually pay and connect to.
Factor scales from the specialistsInfoSec, privacy, legal, finance and continuity each define what low, moderate and high mean in their domain.

One scoping note. Banking organizations supervised by the OCC, FDIC or Federal Reserve work to the Interagency Guidance. EU financial entities in scope of DORA (Regulation (EU) 2022/2554, applying since January 17, 2025) carry a separate, binding test for ICT services. Covered entities under NYDFS 23 NYCRR Part 500 have their own third-party service provider rules in section 500.11. If more than one applies to you, the model has to satisfy the strictest of them, and counsel should confirm which entities and services are in scope.

Who owns the vendor tiering model

One owner for the method, one for each answer

A vendor tiering model fails when the people who benefit from a low tier are the same people who assign it. Split the ownership. The third-party risk function owns the method: the questions, the scales, the mapping rules and the override process. The business owner supplies the facts about the service. Neither should be able to change the other's part alone.

The Interagency Guidance draws the same line. The board "provides clear guidance regarding acceptable risk appetite, approves appropriate policies," and management develops and implements the practices. A footnote adds that each business line "can provide valuable input into the third-party risk management process, for example, by completing risk assessments, reviewing due diligence information, and evaluating the controls over the third-party relationship." Input, in other words. The methodology sits elsewhere.

RoleOwnsDoes not own
Board or risk committeeRisk appetite, policy approval, the definition of criticalIndividual tier decisions
Third-party risk functionTiering method, scales, mapping rules, overrides, the Tier 1 listFacts about how the business uses the service
Business ownerCriticality answers, scope of service, notice when scope changesThe final tier
InfoSec, privacy, legal, finance, BCMThe factor scale for their domainFactors outside their domain
Internal auditIndependent testing that tiers match the methodDesigning the method it tests

Write the override rule down before you need it. A tier can move by exception, with a named approver, a reason and an expiry date. Anything looser, and the Tier 1 list becomes a negotiation.

Step 1 of 6

Inventory every third party & the service it supports

Tier the engagement, not the logo

You can't tier what isn't on the list, and you can't tier a vendor in the abstract. The same cloud provider that hosts your core ledger may also host a marketing sandbox. Those are two engagements with two different answers, so the unit of the inventory is the vendor and the service together.

The Interagency Guidance ties inventory and tiering in one sentence: "Maintaining a complete inventory of its third-party relationships and periodically conducting risk assessments for each third-party relationship supports a banking organization's determination of whether risks have changed over time." Its documentation section then asks for a current inventory "that clearly identifies those relationships associated with higher-risk activities, including critical activities." DORA Article 28(3) is stricter for ICT services. Financial entities must keep a register of information on all ICT contractual arrangements, "distinguishing between those that cover ICT services supporting critical or important functions and those that do not."

For the fields, borrow from the NIST supply chain risk guidance. SP 800-161 Revision 1 (May 2022, updated November 1, 2024) adds control SR-13, Supplier Inventory, which records for each supplier the procurement instrument, a "description of the supplied products and/or services," the program, project or system that uses them, and an "assigned criticality level that aligns to the criticality of the program, project, and/or system." It was written for federal systems, and aerospace and defense suppliers will recognize it. The field list works in any sector.

  • Legal entity, parent and a unique ID that matches accounts payable.
  • Each service or engagement as its own row, with its contract reference.
  • The internal business service or process that depends on it.
  • A named business owner and the business unit.
  • Data types handled and systems accessed.
  • Delivery locations and known subcontractors.
  • Tier, criticality flag, inherent rating and the date each was last set.
Reconcile against money & accessCompare the inventory to twelve months of accounts payable and to your SSO application list. Vendors that appear in either source and not in the inventory are the ones nobody tiered. Identifying critical vendors starts with finding the ones you didn't know you had.

Step 2 of 6

Run a vendor criticality assessment on each service

What stops if this vendor stops?

A vendor criticality assessment answers one question about your own operations: what happens to you if this service fails? It says nothing yet about whether the vendor is well run. Keep those apart. Criticality is a property of your dependency, and no SOC 2 report changes it.

Start from the regulator's wording. Under the Interagency Guidance, critical activities may include those that could "cause a banking organization to face significant risk if the third party fails to meet expectations," "have significant customer impacts," or "have a significant impact on a banking organization's financial condition or operations." The guidance adds that "an activity that is critical for one banking organization may not be critical for another." DORA Article 3(22) defines a critical or important function as one whose disruption "would materially impair the financial performance of a financial entity, or the soundness or continuity of its services and activities," or whose failure would materially impair continuing compliance with its authorization or other financial services law.

Turn those tests into a short vendor tiering questionnaire that the business owner completes at intake, before any vendor-facing questionnaire goes out. Eight questions are enough:

  1. Which business service or process does this engagement support, and is that service on the critical activities list?
  2. If the vendor stopped delivering today, how many hours or days until customers, patients or production feel it?
  3. Is there a tested substitute or manual workaround, and how long would the switch take?
  4. Does the vendor interact with your customers, or act in your name?
  5. Would a failure cause you to miss a regulatory obligation, filing or license condition?
  6. How many business units or critical services rely on this same vendor?
  7. Does delivery depend on a subcontractor the vendor could not easily replace?
  8. What would a 30-day outage cost in revenue, penalties and recovery?

Set the rule in advance. One workable version: an engagement is critical if it supports a service on the critical activities list and either the time to impact is shorter than the time to substitute, or failure would breach a regulatory obligation. Everything else is non-critical, however large the contract. Certa's guide to vendor criticality assessment steps covers the wider assessment this feeds.

Don't let the answer set itselfThe business owner supplies answers. The third-party risk function applies the rule. If owners can tick a box marked "critical" or "not critical," the tier follows whoever wants the shorter onboarding.

Step 3 of 6

Score inherent risk with a vendor risk rating matrix

Exposure before any control is counted

Criticality told you what you depend on. Inherent risk tells you what the engagement exposes: the data it holds, the systems it reaches, the laws it triggers. Score it before considering the vendor's controls. A vendor risk rating that already nets off the vendor's encryption or its audit report is a residual rating, and you haven't done the diligence that would justify it yet.

A vendor risk rating matrix keeps the scoring consistent across reviewers. Each factor gets a scale written by the specialist team that owns the domain, with definitions concrete enough that two analysts reach the same score. The example below uses six factors and three levels.

FactorLow (1)Moderate (2)High (3)
Data sensitivityPublic or no dataInternal or limited personal dataRegulated data: NPI, PHI, CUI, export-controlled
System accessNoneStandard user or API accessPrivileged or persistent network access
Regulatory exposureNo regulated activitySupports a regulated processPerforms a regulated activity for you
Customer contactNoneIndirectDirect, in your name
Location & subcontractingDomestic, no key subcontractorsOffshore delivery or disclosed subcontractorsSanctions-sensitive region or opaque sub-tiers
Financial exposureLow value, easy exitMaterial prepayment or exit costHandles funds or payments

How you combine the factors matters more than the scales. Averaging hides the one thing you need to see. A payroll processor that scores 3 on data and 1 on four other factors averages out to low, which is the wrong answer. Use a highest-factor rule instead:

  • High: any single factor scores 3.
  • Moderate: no 3s, and two or more factors score 2.
  • Low: everything else.

Keep the factor scores on the record after the rating is set. They decide which domain assessments run in the next step, and they give ongoing scoring something to compare against. Certa's vendor scorecards guide picks up from there, on tracking performance once the vendor is live.

Step 4 of 6

Map critical vs high risk vendors into tiers

Critical is a flag. High is a score.

The two measures answer different questions, which is why the critical vs high risk vendors distinction trips up so many models. Critical vendors are the ones your operations can't run without. High-risk vendors are the ones that could hurt you through what they hold or reach. Some are both. Many are one and not the other.

Consider two engagements. A market data feed that your trading desk cannot price without holds no customer data and has no access to your network: critical, low inherent risk. A survey firm holding 200,000 customer email addresses could vanish tomorrow without interrupting a single service: non-critical, high inherent risk. One tier label has to carry both facts without blurring them.

CriticalityInherent ratingTier
CriticalAnyTier 1
Non-criticalHighTier 2
Non-criticalModerateTier 3
Non-criticalLowTier 4

Criticality overrides. Any engagement that supports a critical activity is Tier 1, whatever its inherent rating, because that matches what the regulators ask for: "more comprehensive and rigorous oversight and management of third-party relationships that support higher-risk activities, including critical activities." The inherent rating still travels with a Tier 1 vendor. It decides whether the deep review is a cyber assessment, a resilience and exit review, or both.

Where one vendor has several engagements, tier each engagement and report the vendor at its highest tier. Then read the Tier 1 list aloud to the risk committee. If nobody in the room would lose sleep over a name on it, the criticality rule is too loose.

Step 5 of 6

Set due diligence & monitoring depth for each tier

The tier decides what happens next

Tiers exist to ration attention. Vendor segmentation that changes nothing downstream is a labeling exercise, and a program that runs the same 300-question assessment on every vendor has no tiers at all, whatever the spreadsheet says. The Interagency Guidance sets the standard: "The scope and degree of due diligence should be commensurate with the level of risk and complexity of the third-party relationship." Monitoring follows the same logic, and "more comprehensive or frequent monitoring is appropriate when a third-party relationship supports higher-risk activities, including critical activities."

The table shows one way to turn risk-based vendor segmentation into operating rules. No US banking regulator prescribes these cadences. NYDFS section 500.11(a)(4) asks only for "periodic assessment of such third-party service providers based on the risk they present." Treat the frequencies as defaults to adjust.

TierDue diligenceMonitoringReassessment
Tier 1Full assessment in every relevant domain, evidence validated, exit plan and subcontractor reviewContinuous signals plus quarterly performance reviewAnnual, with board-level reporting
Tier 2Full assessment in the domains that scored 3, standard set elsewhereContinuous signals in the high-scoring domainsAnnual
Tier 3Standard questionnaire, attestation or certification acceptedScreening and adverse media alertsEvery two years
Tier 4Sanctions and watchlist screening, contract standardsScreening alerts onlyOn trigger or at renewal

For Tier 1 under DORA, add the items the regulation requires for ICT services supporting critical or important functions: the concentration assessment in Article 29(1), covering a provider "that is not easily substitutable," and the pre-contract steps in Article 28(4). Confirm the full list against the text with counsel.

Running four different paths by hand is where tiered programs slide back to one-size-fits-all. A platform such as Certa's TPRM orchestration suite routes each intake to the workflow for its tier, with due diligence that is risk-tiered across domains and periodic re-reviews tuned to each vendor's risk tier. Quantcast, a Certa customer, reports vendor onboarding 3x faster, cut to a 7-day cycle for high-risk vendors and 5 days for low-risk ones. The point of the split is visible in those two numbers: low-risk vendors stop waiting in the high-risk queue.

Step 6 of 6

Retier vendors when the relationship changes

A tier is a snapshot with an expiry date

The vendor you tiered at onboarding is rarely the vendor you have three years later. A reporting tool picks up a customer data feed. A regional supplier becomes sole source after a competitor exits. Scope creep moves vendors up the tiers quietly, and the tier on record stays where it was.

The guidance anticipates this: "Because both the level and types of risks may change over the lifetime of third-party relationships, banking organizations may adapt their ongoing monitoring practices accordingly." DORA goes further and makes one change reportable. Under Article 28(3), financial entities must inform the competent authority in a timely manner "when a function has become critical or important." That only works if something in your process notices.

Build retiering on events as well as the calendar. Rerun the criticality questions and the rating matrix whenever one of these fires:

  • A change order, renewal or new statement of work that alters scope, data or access.
  • Your own critical activities list changes, or a new business unit adopts the vendor.
  • The vendor is acquired, merges, or changes a key subcontractor or delivery location.
  • A security incident, service outage or material audit finding at the vendor.
  • A sustained drop in a security rating or financial health score.
  • A new regulation, or a new jurisdiction, brings the service into scope.

The first two triggers come from inside your company, so wire them to procurement intake and to the resilience team's review cycle. The rest arrive as external signals. Certa's primer on continuous monitoring explained covers how those feeds work. Whatever the source, the retier itself should take minutes: same questions, same matrix, a new date and a recorded reason.

Common vendor risk classification mistakes

Where sound models go wrong in practice

Most vendor risk classification problems trace back to a shortcut that looked reasonable when the program was small. These are the ones that surface in audits.

  • Tiering by spend. Contract value measures procurement effort. A $40,000 identity provider can be more critical than a $4 million facilities contract.
  • Tiering the vendor instead of the engagement. A single tier for a multi-service vendor is either too high for most of its work or too low for the part that matters.
  • Averaging the factor scores. One 3 among five 1s is a high-risk vendor in one domain. The average says low.
  • Rating on residual risk. If the vendor's SOC 2 Type II lowers its tier, the tier now decides whether you ever check the SOC 2 again.
  • Tier inflation. When owners can self-declare, a large share of the portfolio ends up critical and Tier 1 stops meaning anything. The reverse happens when Tier 1 means slower onboarding.
  • Undocumented overrides. A tier moved by email, with no approver or reason on the record, is the first thing an examiner samples.
  • Set once, never revisited. Covered in the step above, and still the most common finding.

A quick self-test covers all seven. Pull five Tier 1 vendors and five Tier 4 vendors at random, and ask whether the record shows who tiered each one, on what answers, and when. If it takes more than an afternoon to produce, fix the record before you tune the scoring.

Vendor risk tiering FAQ

How Many Tiers Should a Vendor Risk Model Have?

Three or four is enough for most programs. No US banking regulator sets a number: the Interagency Guidance says some organizations "assign a criticality or risk level to each third-party relationship, whereas others identify critical activities and those third parties that support such activities." Add a tier only when it would change what you do, such as a different assessment, cadence or approver.

What Is the Difference Between Critical & High Risk Vendors?

A critical vendor supports a service your operations can't run without, so the measure is the impact of its failure on you. A high-risk vendor carries high inherent exposure through the data it holds, the systems it reaches or the regulated activity it performs. A vendor can be either or both, which is why the model scores them separately and sends every critical engagement to Tier 1.

How Often Should You Retier Your Vendors?

Whenever a trigger fires, and at every scheduled reassessment as a backstop. Triggers include scope changes at renewal, incidents, vendor ownership changes and changes to your own critical activities list. NYDFS section 500.11(a)(4) requires periodic assessment "based on the risk they present," and the Interagency Guidance expects periodic risk assessments for each relationship without fixing an interval.

Can AI Automate Vendor Tiering?

AI can do the preparation and should not make the call. Certa's Pre-Fill Agent pulls third-party data from websites, news and documents to pre-populate forms and DDQs, and its Risk Agent runs InfoSec risk assessments and surfaces only what needs your review.

The tier decision, and any override, stays with a named person. Certa describes its AI as working "as an assistant to provide insights, not replace human decision-making," with the rationale and data sources shown so users can audit and adjust them.

Vendor Risk Management Articles

Keep reading

Third-Party OSBook a Demo
Get In Touch

By clicking “Accept”, you agree to the storing of cookies on your device to enhance site navigation, analyze site usage, and assist in our marketing efforts. “Reject” keeps only the cookies the site needs to work. View our Privacy Policy for more information.

We use cookies for analytics and advertising, which stay on unless you choose “Reject”. “Accept” also allows session analytics and company identification. You can change this at any time under Your Privacy Choices. View our Privacy Policy for more information.