
A project risk register is the one place where a team writes down what could still go wrong, how likely and how damaging it would be, what will warn them early, and who will act. This guide gives you a free template in three formats, a fully worked ten-risk example, a scoring scale you can adopt as is, and the review routine that keeps the register alive.
A risk register is a living table of uncertain future events that could affect a project's objectives. Each row records the cause, the event and its effect, a likelihood and impact rating, a trigger that signals the risk is starting to happen, the planned response, one named owner, a status, a next review date and the risk left over after the response (residual risk). The Association for Project Management describes the register as the place to document risks, their analysis and responses, and to assign clear ownership of actions (APM).
Use one whenever a project has a fixed date or budget, depends on vendors or other teams, or runs longer than a few weeks. Things that have already happened are issues, not risks; they belong in an issue log.
Free downloads (fictional example included, no sign-up):
Paste this into any document or spreadsheet. The downloads split some cells into separate columns (cause, event and effect; residual likelihood and impact) and add date raised, escalation and notes.
| ID | Cause → event → effect | L (1–5) | I (1–5) | Score / priority | Trigger | Mitigation (now) | Contingency (if triggered) | Owner | Status | Next review | Residual L × I |
|---|---|---|---|---|---|---|---|---|---|---|---|
| R-01 | Because … , … may happen, leading to … | Open | YYYY-MM-DD | ||||||||
| R-02 | Because … , … may happen, leading to … | Open | YYYY-MM-DD |
Put three lines above the table: the project name, the register version and data-as-of date, and the role that maintains it. Then paste the scoring scale directly under the table, so a "4" means the same thing to everyone who opens the register later.
| Field | What to write | Rule that keeps it useful |
|---|---|---|
| Risk ID | R-01, R-02 … | Never reuse an ID, even after a row closes. |
| Cause → event → effect | Because [fact], [uncertain event] may happen, leading to [effect on an objective]. | One event per row. The cause is a fact today; the event is uncertain. |
| Likelihood (L) | 1–5 from the agreed scale | Rate the evidence, not the mood of the meeting. |
| Impact (I) | 1–5 from the agreed scale | Rate the worst effect on one objective: date, cost or scope. |
| Score and priority | L × I, then the priority band | A sorting aid, not a measurement (see scoring). |
| Trigger | An observable threshold with a date | "More than 2% of records fail on Nov 20", not "if data looks bad". |
| Mitigation | What you do now to lower the likelihood or the impact | A concrete action with a date and an action owner. |
| Contingency | The fallback you run if the trigger fires | Agreed in advance, with any money or time it needs. |
| Owner | One named role accountable for the row | Individual actions can have other owners. |
| Status | Open, Mitigating, Accepted, Escalated, Materialized, Closed | A fixed list, so the register can be filtered. |
| Next review | A date | Blank or past dates mean the row is stale. |
| Residual risk | L × I expected after the mitigation is done | If it is still above tolerance, escalate. |
The UK government's project delivery guidance describes the same building blocks: each register entry describes the risk "in terms of cause, event and impact", has a named risk owner who plans, implements and monitors the response, and assigns tasks to named action owners (Government Project Delivery Framework, "Teal Book", chapter 20). HM Treasury's Orange Book lists early warning indicators and contingencies as parts of a risk treatment plan, which is why the trigger and contingency columns earn their place (Orange Book, 2023).

The register below is a fictional example written for this article. The bookseller, its store replatform, the roles, dates and numbers are invented; they are not an AFFiNE customer or a real project. It is the same data as the example sheet in the downloads.
Project: online store replatform for a fictional bookseller · Go-live: 2027-02-15 · Register: v1.3, data as of 2026-10-05 · Maintained by: project manager · Scales: the 5 × 5 scale below
| ID | Cause → event → effect | L × I = score | Priority | Trigger | Owner | Status | Next review |
|---|---|---|---|---|---|---|---|
| R-01 | 14 years of legacy order history with inconsistent customer records → the trial migration fails validation → go-live slips 4+ weeks past Feb 15 | 4 × 4 = 16 | Critical | More than 2% of records fail validation in the Nov 20 trial migration | Data lead | Mitigating | 2026-10-12 |
| R-02 | The new payment provider requires a merchant risk review → approval is not granted before the Dec 15 code freeze → checkout cannot take card payments at go-live | 3 × 5 = 15 | Critical | No approval decision from the provider by Dec 1 | Finance lead | Escalated | 2026-10-12 |
| R-07 | Product URLs change format on the new platform → redirects are incomplete at launch → search and bookmarked links break and uncovered categories miss launch | 3 × 4 = 12 | High | Redirect map covers less than 95% of the top 2,000 URLs at the Jan 15 crawl | Marketing lead | Mitigating | 2026-10-12 |
| R-04 | Marketing asks for loyalty-program features mid-build → change requests are accepted without re-planning → the release slips 2 to 3 weeks or testing is cut | 3 × 3 = 9 | Medium | More than two unplanned change requests in one sprint | Project manager | Materialized | Tracked as issue I-03 |
| R-06 | The agency's fixed-price quote excludes data cleanup → cleanup needs more than 15 agency days → the budget overruns by 3 to 5% | 3 × 3 = 9 | Medium | Cleanup estimate above 15 days after the Oct 30 data profiling | Vendor manager | Open | 2026-10-19 |
| R-03 | Only one developer knows the legacy inventory integration → that developer is unavailable during the cutover window → cutover stalls and go-live slips about a month | 2 × 4 = 8 | Medium | Leave, illness or resignation notice before Jan 31 | Tech lead | Mitigating | 2026-10-19 |
| R-05 | Caching on the new platform is untested at peak traffic → the load test at 3x average traffic shows slow pages → launch is postponed by about a month | 2 × 4 = 8 | Medium | p95 page load above 3 seconds in the Jan 10 load test | Tech lead | Open | 2026-10-19 |
| R-09 | The platform vendor freezes releases from Dec 15 to Jan 5 → a patch we need is not released before the freeze → integration testing slips 1 to 2 weeks | 4 × 2 = 8 | Medium | Any open vendor ticket of severity 2 or higher on Dec 1 | Vendor manager | Accepted | 2026-10-19 |
| R-08 | Warehouse and support staff are trained only on the old admin → training is not finished before cutover → order processing slows and overtime adds 3 to 5% to the budget | 2 × 3 = 6 | Medium | Less than 80% of staff complete sandbox training by Feb 1 | Support lead | Open | 2026-10-19 |
| R-10 | The freelance copywriter is booked for only 40 hours in January → rewritten category pages are not ready → 12 categories launch with old copy | 2 × 1 = 2 | Low | Drafts for fewer than 6 categories by Jan 10 | Marketing lead | Closed | Closed 2026-09-30 |
| ID | Response | Mitigation (now) | Contingency (if triggered) | Residual L × I | Escalation / notes |
|---|---|---|---|---|---|
| R-01 | Mitigate | Profile and clean the top five error types before the trial; run a dedupe script on customer records | Migrate the last 3 years of orders at launch; keep older history read-only in the legacy system | 2 × 4 = 8 (Medium) | Re-rated from 3 × 4 to 4 × 4 on 2026-09-28 after a sample export showed 3.1% validation failures and 6% duplicate customer accounts. Sponsor informed 2026-09-30 (Critical) |
| R-02 | Mitigate | Submit all merchant documents by Oct 20; weekly check-in with the provider's account manager | Keep the legacy payment integration behind the new storefront for 60 days | 2 × 5 = 10 (High) | Escalated to sponsor 2026-10-02: approve 60-day legacy gateway extension? Decision due 2026-10-09 |
| R-07 | Mitigate | Build the redirect map from analytics top URLs; automated crawl test every sprint | Hold launch of uncovered categories and serve them from the legacy site | 1 × 4 = 4 (Low) | — |
| R-04 | Mitigate | Change-control rule: every new request must swap out planned work of equal size | Move loyalty features to a phase-2 release in April | n/a | Trigger fired Oct 2 (three requests in sprint 4); contingency invoked; tracked as issue I-03 |
| R-06 | Mitigate | Do the profiling in-house first; cap agency days in a contract amendment | Draw from the 8% contingency reserve; sponsor approval needed above 5% | 2 × 3 = 6 (Medium) | — |
| R-03 | Mitigate | Pair a second developer on the integration and write a cutover runbook by Nov 30 | Integration partner contractor on 48-hour call-off | 2 × 2 = 4 (Low) | — |
| R-05 | Mitigate | Configure the CDN and run an early load test in the first week of December | Keep the legacy site live for two more weeks and switch traffic gradually | 1 × 4 = 4 (Low) | — |
| R-09 | Accept | Raise all known vendor tickets before Nov 30 | Use one week of the schedule buffer for extra testing | 4 × 2 = 8 (Medium) | Accepted by project manager 2026-09-28; within schedule tolerance |
| R-08 | Mitigate | Three sandbox training sessions in January with sign-off per person | Two temporary support agents and extended hours for the first two weeks | 1 × 3 = 3 (Low) | — |
| R-10 | Accept | Prioritize the 6 highest-traffic categories | Launch with old copy and update after go-live | n/a | Closed 2026-09-30: all 12 drafts delivered early, so the trigger can no longer fire |
Read the register the way a reviewer would:
Agree on the scale with the project sponsor before anyone rates a risk. The impact thresholds below are written for a six-month project with a fixed launch date; change the numbers to fit your project, then keep them fixed for the life of the register.
| Rating | Likelihood: evidence that fits | Impact: worst effect on one objective |
|---|---|---|
| 1 | Rare. No precedent on comparable projects and no warning signs. | Minimal. Under 1 week of slip, under 1% of budget, no visible scope change. |
| 2 | Unlikely. Happened once on comparable work; no current warning sign. | Minor. 1–2 weeks of slip, 1–3% of budget, or an internal feature trimmed. |
| 3 | Possible. Happened on several comparable projects, or one warning sign is present. | Moderate. 2–4 weeks of slip, 3–5% of budget, or a planned feature deferred. |
| 4 | Likely. Happened on most comparable projects, or two or more warning signs are present. | Major. 4–8 weeks of slip or a fixed date missed, 5–10% of budget, or a key launch feature deferred. |
| 5 | Almost certain. Trigger conditions are already forming. | Severe. Over 8 weeks of slip, over 10% of budget, or a core objective fails. |
Priority bands and review cadence
| Priority | Score (L × I) | Review | Escalation in this example |
|---|---|---|---|
| Critical | 15–25 | Weekly | Tell the sponsor within 2 working days. |
| High | 10–14, or any impact of 5 | Weekly | Escalate if the residual score stays at 10 or more. |
| Medium | 5–9 | Every two weeks | Owner manages within tolerance. |
| Low | 1–4 | Monthly | Close when the trigger window passes. |
The "any impact of 5 is at least High" rule is deliberate: without it, a rare but project-ending risk (L1 × I5 = 5) would sit in the same band as a frequent nuisance.
The numbers 1 to 5 are ordered labels. They are not measurements, so arithmetic on them has limits:
In practice: publish the scale inside the register, read likelihood and impact separately (not just the product), look at proximity (a Medium risk whose trigger date is next week may need attention before a High one that is months away), and when a decision depends on money or dates, estimate the days or cost directly instead of arguing whether a score is 12 or 15. When the choice is mainly about money over several years, a cost-benefit analysis template puts those estimates into NPV and payback.

Text version of the heat map (open risks only; R-04 materialized and R-10 is closed):
| Open risks | Likelihood × impact | Score | Priority |
|---|---|---|---|
| R-01 | 4 Likely × 4 Major | 16 | Critical |
| R-02 | 3 Possible × 5 Severe | 15 | Critical |
| R-07 | 3 Possible × 4 Major | 12 | High |
| R-06 | 3 Possible × 3 Moderate | 9 | Medium |
| R-03, R-05 | 2 Unlikely × 4 Major | 8 | Medium |
| R-09 | 4 Likely × 2 Minor | 8 | Medium |
| R-08 | 2 Unlikely × 3 Moderate | 6 | Medium |
The UK government's project delivery standard asks that risks be identified, given an owner and assessed for impact, likelihood and proximity, then monitored until they are resolved and closed (GovS 002 Project delivery, clause 7.6). Here is that loop as a working procedure.

A risk might happen. An issue has happened, or is happening now, and needs action rather than monitoring. The Teal Book puts it plainly: "If a risk happens then it becomes an issue".
| Risk | Issue | |
|---|---|---|
| Time | Future and uncertain | Present and certain |
| Main question | How likely, how bad, what will warn us? | Who fixes it, by when? |
| Response | Mitigation now, contingency ready | Resolution actions now |
| Where it lives | Risk register | Issue log |
| Example | R-04: change requests may swamp sprint capacity | I-03: three unplanned requests landed in sprint 4 |
When a risk materializes, keep the register row, set the status to Materialized, and write the issue ID in the notes. That link is how a team learns whether its triggers and contingencies worked.
A RAID log tracks risks, assumptions, issues and dependencies side by side, usually with one or two lines per item. The risk register is the detailed version of its "R": the place for scoring, triggers, contingencies and residual risk. Small projects can keep everything in a RAID log; once you have more than about ten live risks or need scores and escalation rules, keep a separate register and link to it from the R section. That threshold is an editorial rule of thumb, not a standard. See our guide on how to run a RAID log.
The other items in a RAID log feed the register. An assumption that turns out false becomes a risk or an issue. A dependency on another team that might slip becomes a risk row with that team's delivery date as its trigger. Before a project is funded, the top risks usually appear first in the business case, and the register takes them over once work starts.
| Document | Question it answers | Relationship to the register |
|---|---|---|
| RAID log | What risks, assumptions, issues and dependencies are live? | Summary view; the register holds the detail on risks |
| Issue log | What has gone wrong, and who is fixing it? | Receives materialized risks |
| Risk assessment | How exposed is an activity or site at a point in time? | Findings with a project impact become register rows |
| RACI chart | Who is responsible, accountable, consulted, informed? | Helps choose risk and action owners |
| Lessons learned | What should we repeat or avoid next time? | Closed rows and materialized risks feed it; it seeds the next register |
A project risk register is not a workplace hazard assessment or a security risk assessment. Those follow their own legal and industry rules; if that is what you need, start with risk assessment templates for hazards and point-in-time checks.
The register is also one of the core outputs of planning, alongside the schedule, budget and communication plan; see the other project planning phase deliverables. Treat it like any controlled project document, with an owner, a version and a review date, following the same project documentation practices for ownership and version control.
AFFiNE does not offer a ready-made risk register template, and its database columns do not calculate formulas. So the most practical setup is the downloaded workbook for scoring, and AFFiNE for the living register, the evidence behind each row and the discussion around it. The steps below use features described in AFFiNE's documentation, release notes and source code as of v0.27.4; we did not test this setup in the app for this article.
Limits worth knowing as of AFFiNE v0.27.4 (August 2026): database columns have no formulas, so scores do not recalculate; there is no person column, so owners are typed as text; AFFiNE does not send reminders when a review date passes; and its import options include Markdown files but not CSV. If you import the Markdown download, check how its table arrives before you rely on it; we did not test that import.
This page replaces the version published on 2026-08-22. That version used a 3 × 3 scale and leaned on cybersecurity and workplace-safety references. This refresh adds downloadable files, a 5 × 5 scale with a worked score, contingency, escalation and residual risk in the template, and project-delivery sources.
References (all accessed 2026-10-05):
Use a spreadsheet when you want scores and review flags to calculate themselves, and a document or database when the discussion, evidence and decisions behind each risk matter more. Many teams use both: the workbook for scoring and a shared page for the living record.
There is no standard number. Keep only risks that someone will act on or watch, and close or merge the rest. As a rule of thumb, if you have more than about 30 live rows, split the register by workstream and roll the Critical and High rows up to one summary for the sponsor.
The project manager usually maintains the register and runs the reviews, while each row has its own risk owner who is accountable for the response. The sponsor sets the tolerance that decides when a risk must be escalated.
Yes. A risk can be positive: an uncertain event that would help the project, such as a vendor delivering early. Log it with the same cause, event and effect structure, rate it, and plan how to make it more likely or more valuable.