A fast third-party risk assessment triage takes about 30 minutes. Check the intake form, ask whether the tool touches sensitive data, is critical to operations or connects to core systems, scan for tripwires and reuse existing evidence. Then set a low, medium or high tier and record why.
Most software requests do not need a 300-question review. They need someone to decide, in half an hour, which few actually do.
In a hospital emergency department, the first person you meet is not a surgeon. It is a triage nurse who decides, within minutes, who needs a doctor now and who can wait. That brief decision keeps the surgeons free for the patients who need them. This guide gives your team the same discipline for software: a 30-minute third-party risk assessment triage that sets the right review depth for every tool and leaves a record an auditor can follow.
Why third-party risk queues stall
Queues rarely stall because reviewers are slow. They stall because every request enters the same funnel. A note-taking app and a payroll platform receive the same questionnaire, while the one request that needs expert attention waits behind twenty that never did.
Three causes recur. Intake forms ask what a tool is called rather than what it touches. Reviewers gather evidence before deciding how much is needed. And nobody has written down when moving quickly is acceptable, so teams buy on a company card instead.
The stakes are real. Verizon's 2025 Data Breach Investigations Report found third parties involved in 30% of breaches, double the previous year. NIST Cybersecurity Framework 2.0 expects suppliers to be known and prioritised by criticality. Triage is how that expectation works in practice.
The 30-minute routine
Run the triage as a fixed routine with a timer, ideally by one analyst using a standard form. If facts are missing, the clock stops and the request returns to its owner.
- Minutes 0 to 5, intake check: confirm the business owner, use case, data, users and integrations are stated.
- Minutes 5 to 15, impact questions: answer the three questions below from the form and public documentation.
- Minutes 15 to 20, tripwires: scan for facts that raise the tier regardless of the count.
- Minutes 20 to 25, evidence reuse: check for a prior assessment, SOC 2 report, ISO/IEC 27001 certificate or completed CAIQ, the Cloud Security Alliance questionnaire.
- Minutes 25 to 30, decision: set the tier, approver and next step, and write the two-line record.
The three impact questions
Write a definition for each question and publish it beside the intake form. Sensitive data means personal or customer data, credentials, source code, financial records or unreleased plans. Critical means one day of downtime would stop revenue, payroll, customer service or a regulated process. Connected means write access to core systems, administrator rights, broad mail or file permissions, or an installed agent.
Count the yes answers. None sets the low tier, one sets medium, and two or three set high.
Tripwires that override the count
Some facts raise the tier by at least one level, whatever the count. Examples include an AI feature with unclear training use of your content, and permissions such as read all mail. Others are a missing sub-processor list, regulated data such as health or payment card data, and personal accounts connected to company data.
What each tier should trigger
The tier saves time only if each level has a fixed, published response. The turnaround targets below are internal service goals, not industry rules.
| Tier | Evidence to collect | Approver | Turnaround target |
|---|---|---|---|
| Low: no yes answers, no tripwire | Terms and privacy notice checked, single sign-on confirmed | Analyst | Same day |
| Medium: one yes or a tripwire | Current SOC 2 Type II or ISO/IEC 27001 certificate, scope checked | Security lead | Within five business days |
| High: two or three yes answers | Full assessment: evidence review, findings, contract terms, monitoring | CISO or risk committee | Planned with the requester |
Reuse evidence wherever possible, but read it first. Check that it is current and that its scope covers the product you are buying. A SOC 2 Type II report, an independent auditor's report on how controls operated over a period, outweighs a point-in-time Type I. If any check fails, record the gap and raise the tier.
A worked example: one Monday's intake queue
Here is a hypothetical queue of six requests received on the same morning, triaged with the routine above.
| Request | Yes answers | Tripwire | Tier |
|---|---|---|---|
| Diagramming tool | None | None | Low |
| E-signature service | Data | None | Medium |
| Survey tool | Data | None | Medium |
| Meeting transcription with AI summaries | Data | Training use unclear | High |
| Payroll platform | Data, critical, connected | None | High |
| Browser extension for sales notes | Connected | Read all mail permission | High |
Now the arithmetic, using hypothetical effort figures. Full reviews of 12 analyst hours each would cost 72 hours. With triage, six triages take 3 hours, the two medium requests need 3 hours each and three full reviews need 36 hours. The total is 45 hours.
The real gain lies in where those hours go. Most now land on the three tools that could cause harm. Two of those, the transcription tool and the browser extension, would have looked harmless on a list of names.
Keeping fast triage sound
Four habits keep the method sound. Requesters supply facts, but the analyst applies the rule. Every decision, especially a quick approval, receives a two-line record of the tier, the deciding facts, the approver and the next review date. A tool that adds an AI feature or new integration is triaged again. And a high tier begins a full review rather than ending one.
Volume will not fall. The 2024 Work Trend Index from Microsoft and LinkedIn found that 78% of AI users bring their own AI tools to work. A triage that clears simple requests the same day keeps people inside the process.
The bottom line
Conclusion
The triage nurse treats nobody, yet that decision keeps the emergency department working. A third-party risk programme works the same way. Written definitions, a few tripwires and a two-line record let simple tools through the same day and send the critical few to people with time to examine them.
This week, take the last ten software requests your team received and apply the three questions and the tripwires to each. Compare the resulting tiers with the effort each request actually received; the gap is your business case.
Frequently asked questions
What is the difference between triage and a full third-party risk assessment?
Triage decides how much review a tool needs and takes about 30 minutes. A full assessment examines evidence, records findings, sets contract terms and plans monitoring for high-tier suppliers.
Which tripwires should raise a tool's risk tier?
Common tripwires include unclear AI training use of company data, broad mail or file permissions, no published sub-processor list, regulated data, and personal accounts connected to company data.
Can I approve low-risk software the same day?
Many teams do, provided the tool has no yes answers and no tripwires, the terms and single sign-on have been checked, and a two-line decision record is kept.
Does NIST support tiering suppliers by risk?
Yes. NIST CSF 2.0 expects suppliers to be known and prioritised by criticality, and NIST SP 800-161 Rev. 1 describes a risk-based approach to supply chain risk management.
How often should a triaged tool be re-triaged?
Re-triage on any material change, such as a new AI feature, a new integration or wider data use, and otherwise at the reassessment date set by its tier.
References
- Verizon, 2025 Data Breach Investigations Report (press release)
- NIST, The NIST Cybersecurity Framework (CSF) 2.0, CSWP 29 (26 February 2024)
- Secureframe, SOC 2 Trust Services Criteria explained
- ISO, ISO/IEC 27001 Information security management
- Cloud Security Alliance, CAIQ v4
- Microsoft and LinkedIn, 2024 Work Trend Index
- NIST SP 800-161 Rev. 1, Cybersecurity Supply Chain Risk Management
Disclaimer: General information only, not legal, security, compliance or financial advice. Check the cited sources and seek professional advice. Names belong to their owners; no endorsement implied.

