All posts
AFFiNE
Toeverything·Originally published Aug 4, 2026 · Updated Oct 5, 2026
Four RAID lanes, each holding one entry from a fictional help desk migration (R-02, A-03, I-03, D-02), converging on decision record DEC-04: migrate the last 18 months and archive the rest read-only

RAID Log: Meaning, Worked Example, and How to Use One

A RAID log is a single project register for risks, assumptions, issues and dependencies. Every entry has one named owner, a priority, a status and a next review date. Project managers use it to keep uncertainty, live problems and handoffs from other teams visible between plan updates, and to turn them into decisions before they turn into surprises.

This guide explains what the acronym means and why some teams swap in actions or decisions. It shows the fields each entry needs and works through one fictional cross-functional project in full: 11 entries, a weekly review routine, escalation rules, and a before-and-after snapshot of a risk that became an issue. You can copy the blank template as a Markdown table or download it as a CSV.

RAID log meaning

RAID most often stands for Risks, Assumptions, Issues and Dependencies. Each letter answers a different question about the same project:

LetterStands forTestExample from the worked project below
RRiskCould it happen, and has it not happened yet?If the ticket export does not fit the cutover window, go-live slips
AAssumptionIs the plan treating it as true without proof?November volume will be at most 1.8× the October average
IIssueIs it a problem now?14 contractor agents cannot sign in through SSO
DDependencyDo we need something from another team or supplier by a date?Production API keys from the Platform team by Oct 19

A risk is uncertain and in the future, so it gets a likelihood and a trigger. An issue has already happened, so it gets a resolution target instead. An assumption is a planning belief that needs a validation date. A dependency is a prerequisite someone else controls, so it needs a provider and a need-by date.

Why some teams use actions or decisions instead

The letters are not fixed. The Institute of Risk Management's short guide lists the categories as Risks, Assumptions, Issues and Decisions "(or Dependencies)". It also notes that some teams combine actions and decisions into a "RAAIDD" log.

This guide keeps the four core categories and links actions and decisions by ID instead of storing them as rows, for two reasons:

  • They have different lifecycles. An action is finished when the work is done. A decision stays valid until it is superseded. Risks and issues need watching until they close.
  • Tasks bloat a log. A 2002 PMI paper on project logs argues that tasks belong in the work breakdown structure and schedule, not in a log. Add tasks to a RAID log and the few uncertain items get buried under the to-do list.

If your organization's template uses A for Actions or D for Decisions, keep its letters and define them at the top of the log. Consistency matters more than any one expansion.

RAID is a convention, not a standard

Two widely used references cover the same ground without defining a "RAID log":

  • PRINCE2 uses separate records: a risk register, an issue register, a daily log and a lessons log.
  • The Association for Project Management (APM) glossary defines issues, assumptions and dependencies, but it has no RAID entry.

Public RAID logs therefore vary. One example is the RAID attachment ELEXON published in 2019 for an Ofgem settlement reform. It held 3 risks, 12 assumptions, 3 issues and 4 dependencies. Its last column changed by type: "Mitigation in place" for risks, "How/when to validate" for assumptions, "How to progress" for issues and "Status of dependency" for dependencies. The template in this guide uses the same idea: shared core fields plus one field that depends on the type.

When to use a RAID log

A RAID log earns its upkeep when at least two of these are true:

  • Several teams or an outside supplier must deliver pieces of the project.
  • A fixed date, launch gate or contract milestone is at stake.
  • The plan rests on numbers or approvals nobody has confirmed yet.
  • The project runs long enough for conditions to change, typically more than a few weeks.

Start it during planning, alongside the other planning-phase deliverables.

A two-person task list rarely needs one. A checklist and a calendar will do. In regulated programs with formal registers, use the RAID log as a summary view that links to those registers, so each item has one authoritative record. That is also where it sits in a wider project documentation set: one of a few coordinating records, not the place every detail lives.

Who owns what in a RAID log

"The PM owns the RAID log" is only half true. Separate four roles:

RoleIn the worked exampleOwnsDoes not own
Log ownerMaya, Support Ops lead and project leadFormat, scales, review prep, the meeting, publishing changesEvery entry's content
Entry ownerDev, IT lead, for R-02 and I-03One row: evidence, status, next step, escalationThe tasks other people do
Decision ownerRosa, VP Customer Experience and sponsorDecisions outside the project lead's authority: scope, date, budgetDay-to-day updates
ProviderPlatform team, vendor professional servicesDelivering a dependencyThe dependency row, which stays with an internal owner

Name a person, not a department, in every Owner cell. A provider outside your team cannot update your log, so an internal owner chases them. A RACI chart settles who is responsible or consulted for deliverables; the RAID log records who is watching each uncertainty.

Anatomy of a RAID entry

Each entry needs nine fields. Field 5 changes with the type of entry, and that is what keeps a risk from being tracked like a task.

Annotated RAID entry R-02 with nine numbered fields, from ID and if-then entry to owner, priority, trigger, response, status, next review and links, plus the one field that changes by entry type
Entry R-02 as it stood at the Oct 6 review. Field 5 is the one that changes with the type of entry.
FieldWhat to writeR-02 at the Oct 6 review
ID and typeA stable ID that is never reused; the prefix shows the typeR-02 · Risk
EntryRisks: "If [condition], then [effect]". Issues: evidence and effect. Assumptions: "We are planning on the basis that…". Dependencies: "[Work] needs [deliverable] from [provider]"If the full ticket export does not fit the 20-hour cutover window, go-live slips
OwnerOne named personDev, IT lead
PriorityFrom your written scale (below)Likelihood 2 × impact 3 = 6, so P1
Type fieldRisk: trigger. Assumption: validation method and date. Issue: resolution target. Dependency: provider and need-by dateTrigger: a sample export projects more than 16 h
Response or next stepA concrete action, with any task IDRun a 24,000-ticket sample export on Oct 8
StatusOne of the status words for that typeOpen
Next reviewA date, so "no change" is an explicit decisionTue Oct 13
LinksRelated IDs in other recordsD-02; raised Sep 15

Status words by type

Use a short, fixed set of status words per type so that a filter can find every open item:

TypeOpen statesExit states
RiskOpenTriggered (link the new issue), Closed (the window passed or the risk was avoided)
AssumptionUnvalidatedValidated, Invalidated (link the resulting risk or issue)
IssueOpen, In progressResolved, then Closed at the next review once the evidence is filed
DependencyRequested, CommittedDelivered, Missed (link the resulting issue)

A priority scale you can apply the same way every time

The worked example uses this scale. Copy it, then change the thresholds to fit your project:

  • Impact: 1 = one workstream slips by less than 3 business days. 2 = a workstream milestone is missed. 3 = go-live, a launch gate or customer data is threatened.
  • Likelihood (risks only): 1 = no current sign. 2 = plausible, with some signs. 3 = expected without action.
  • Risk priority: likelihood × impact. A score of 6–9 is P1, 3–4 is P2 and 1–2 is P3.
  • Other types: P1 if the entry threatens go-live, a gate or customer data. P2 if it threatens a workstream milestone. P3 for everything else.

A written scale stops everything drifting to "High" and lets priority, not volume, set the review order.

Copy-ready RAID log template

Paste this into a doc or spreadsheet and add one row per entry. The four starter rows show the wording for each type; replace them with your own.

| ID | Type | Entry | Owner | Raised | Priority | Status | Trigger / validation / need-by | Response or next step | Next review | Links | Last updated |
|---|---|---|---|---|---|---|---|---|---|---|---|
| R-01 | Risk | If [condition], then [effect]. | [Name] | YYYY-MM-DD | P1 / P2 / P3 (L × I) | Open | Trigger: [observable signal] | [Preventive step and contingency] | YYYY-MM-DD | [IDs] | YYYY-MM-DD |
| A-01 | Assumption | We are planning on the basis that [condition]. | [Name] | YYYY-MM-DD | P1 / P2 / P3 | Unvalidated | Validate: [method] by YYYY-MM-DD | [What happens if it proves false] | YYYY-MM-DD | [IDs] | YYYY-MM-DD |
| I-01 | Issue | [What is wrong now]; evidence: [fact]; effect: [impact]. | [Name] | YYYY-MM-DD | P1 / P2 / P3 | Open | Target: resolve by YYYY-MM-DD | [Resolution step or escalation] | YYYY-MM-DD | [IDs] | YYYY-MM-DD |
| D-01 | Dependency | [Our work] needs [deliverable] from [provider]. | [Name] | YYYY-MM-DD | P1 / P2 / P3 | Requested | Need by YYYY-MM-DD; provider: [team] | [Confirm, chase or fallback] | YYYY-MM-DD | [IDs] | YYYY-MM-DD |

Prefer a spreadsheet? Download the same columns:

Both are plain comma-separated files, so spreadsheet apps such as Excel, Google Sheets and Numbers can open them. They are free and need no sign-up.

How to set up a RAID log in seven steps

  1. Set the scope and rules. Name the project, then write the letters you use, the priority scale, the status words for each type and the review slot at the top of the log.
  2. Collect candidate items at kickoff. Read the charter, schedule, estimates, contracts and open approvals. Then ask each workstream lead four questions: what could go wrong, what does the plan assume, what is blocked now, and what are you waiting for from someone else? If the list is long and messy, group the notes with an affinity diagram before you write entries.
  3. Classify and write each entry precisely. Use the pattern for its type from the field table above. Split a row in two when its parts need different owners or responses.
  4. Score impact and priority. Apply the written scale, and score likelihood for risks only.
  5. Assign one owner. The owner keeps the row current and escalates. They do not have to do every task themselves.
  6. Set the response and the next review date. Fill in the type field (trigger, validation, resolution target or need-by date) and book the weekly review.
  7. Close, link and archive. Close an entry with its evidence and links. Archive it rather than deleting it, so the history survives.

Worked example: a help desk migration RAID log

The project below is fictional. It does not describe an AFFiNE customer, and all of its numbers are invented to show the method. It uses real 2026 calendar dates so you can check every weekday.

Project setup and assumptions

  • Project: An online retailer moves customer support from a shared inbox to a help desk tool before the holiday peak.
  • People: Maya (Support Ops lead) is the project lead and log owner. Rosa (VP Customer Experience) is the sponsor. IT, Identity, Legal, Training, Ecommerce engineering and the vendor's professional services team each contribute.
  • Dates: Kickoff Mon Sep 14, 2026. Cutover from Sat Nov 7, 20:00 to Sun Nov 8, 16:00. That is a 20-hour window that leaves Sunday evening for checks. Go-live Mon Nov 9. Black Friday is Nov 27.
  • Volumes: 216,000 historical tickets. The 98,000 tickets created since May 1, 2025 cover the last 18 months.
  • Cadence: Owners update their rows by Monday 17:00; the review is Tuesday at 10:00 for 30 minutes.

The filled RAID log at the Oct 13 review

The log is split into two tables here so it reads on a phone. The CSV keeps all 12 columns in one sheet.

What and who

IDTypeEntryOwnerPriority
R-01RiskIf agents are not fluent with the new macros before peak, first replies may miss the 4-hour target.Lena, Training leadP2 (2 × 2 = 4)
R-02RiskIf the full ticket-history export does not fit the 20-hour cutover window, go-live slips.Dev, IT leadP1 (2 × 3 = 6)
R-03RiskIf order-lookup calls slow down under peak load, agents fall back to manual lookups.Sam, Ecommerce engineerP2 (1 × 3 = 3)
A-01AssumptionNovember volume will be at most 1.8× the October average, so the staffing plan holds.MayaP2
A-02AssumptionLegal can accept the vendor's standard data processing agreement (DPA) unchanged.Omar, Legal counselP1
A-03AssumptionAgents rarely need tickets older than 18 months, so a read-only archive is enough.MayaP1
I-01IssueThe vendor DPA lacks the 30-day sub-processor change notice our policy requires; signature is blocked.OmarP1
I-02Issue14 contractor agents cannot sign in through SSO because their accounts sit in a separate directory group.Kai, Identity adminP2
I-03IssueA sample export ran at about 7,000 tickets an hour, so the full history would take 30.9 hours.DevP1
D-01DependencyThe order-lookup load test needs production API keys from the Platform team.SamP2
D-02DependencyCutover needs the vendor's migration engineer for Nov 7–8, working to the scope in DEC-04.MayaP1

Control

IDStatusTrigger / validation / need-byNext stepNext reviewLinks
R-01OpenTrigger: under 80% pass the macro check on Oct 30Two practice shifts per team; macro buddiesOct 20—
R-02Triggered Oct 8Trigger: sample projects more than 16 hTracked in I-03—I-03, D-02
R-03OpenTrigger: p95 lookup above 1.5 s in the Oct 21 load testLoad test; cache recent orders if triggeredOct 20D-01
A-01UnvalidatedValidate against Nov 2024 and Nov 2025 actuals by Oct 16If false, raise a staffing riskOct 20—
A-02Invalidated Sep 29Validate: Legal review by Sep 30Closed; follow-up in I-01—I-01
A-03Validated Oct 12Validate: check 500 recent tickets by Oct 129 of 500 (1.8%) cited older tickets; recheck after launchNov 16DEC-04, I-03
I-01Closed Oct 13Target: Oct 9Vendor signed an addendum; filed in contract folder—A-02
I-02In progressTarget: Oct 16Add the contractor group to the SSO app (ACT-31)Oct 20ACT-31
I-03In progressTarget: archive tested by Oct 30Escalated Oct 8; DEC-04 made Oct 12; ACT-37Oct 20R-02, A-03, DEC-04, ACT-37, D-02
D-01CommittedNeed by Oct 19; Platform committed to Oct 16Check keys Oct 16; load test Oct 21Oct 20R-03
D-02RequestedNeed by Oct 23; provider: vendorNew scope sent Oct 12; sponsor calls vendor if unconfirmed Oct 20Oct 20DEC-04, R-02
RAID board for a fictional help desk migration with 11 entries in risk, assumption, issue and dependency columns, each showing owner, priority, status and next step; converted entries are dashed
The same 11 entries as a board. Converted and closed entries stay visible, dashed, so the history is not lost.

Two patterns are worth copying. First, nothing is deleted. R-02 and A-02 stay in the log with a status that points to the record that replaced them. Second, every open entry has a next review date, even when nothing has changed.

Before and after: what changed between Oct 6 and Oct 13

IDAt the Oct 6 reviewAt the Oct 13 reviewWhat changed it
R-02Open, P1; sample export scheduled for Oct 8Triggered, linked to I-03The Oct 8 sample projected 30.9 h, above the 16 h trigger
I-03Did not existP1, In progress; DEC-04 made; ACT-37 due Oct 30Opened Oct 8 and escalated to the sponsor the same day
A-03Did not existValidated: 9 of 500 sampled tickets (1.8%) cited older historyRaised Oct 9 so the decision would rest on evidence
I-01P1, In progress; addendum under negotiationClosed, with the signed addendum filedThe vendor signed on Oct 9
D-02Requested; need by Oct 23Requested; new scope sent; escalation date Oct 20DEC-04 reduced the migration scope
R-01, R-03, A-01, I-02, D-01Next review Oct 13No material change; next review Oct 20"No change" recorded with a new date

The meeting spent its time on the first five rows. The unchanged rows took one line in the minutes.

The decision record: DEC-04

I-03 needed a choice only the sponsor could make, because every option changed scope or the cutover plan. The decision lives in the project's decision log, not in the RAID log, and each record links to the other.

The arithmetic behind DEC-04:

  1. Measured rate: a 24,000-ticket sample took about 3.4 hours, roughly 7,000 tickets an hour.
  2. Full history: 216,000 ÷ 7,000 = 30.9 hours. That is above the 16-hour trigger, which is 80% of the window, and above the 20-hour window itself.
  3. Last 18 months only: 98,000 ÷ 7,000 = 14.0 hours. That leaves 6.0 hours, or 30% of the window, as buffer.
  4. Archive: 216,000 − 98,000 = 118,000 older tickets go to a read-only export.

Options considered:

  1. Extend the window to at least 31 hours, which is 30.9 hours with no buffer. Rejected: the export alone would run until about 03:00 on Monday, leaving no time for checks before the office opens.
  2. Migrate in two waves. Rejected: it adds a second cutover during peak preparation.
  3. Migrate the last 18 months and archive the rest. Chosen, because it fits with buffer and A-03 showed that 1.8% of sampled tickets cited older history.

The record:

DEC-04 fieldValue
DecisionMigrate tickets created from May 1, 2025 onward; export the older 118,000 to a read-only archive
Decided byRosa, sponsor
DateMon Oct 12, 2026
LinksI-03, R-02, A-03, D-02, ACT-37

When the options are closer than this, a weighted decision matrix makes the trade-offs explicit before the sponsor decides.

The arithmetic has limits. It assumes the production export runs at the sample's rate, and attachments or API throttling could change that. That is why I-03 stays open until a full rehearsal fits the window, rather than closing when the decision is made.

How a risk becomes an issue

A risk turns into an issue when its trigger fires or the event happens. Do not edit the risk row into an issue. You would lose the record of what was foreseen and when.

Lifecycle in which risk R-02 triggers on October 8 and becomes issue I-03, decision DEC-04 creates action ACT-37, and I-03 closes with evidence on October 30, above a weekly review loop
R-02 becomes I-03, which leads to DEC-04 and ACT-37. Each record keeps its own ID and home; the links carry the history.
  1. Confirm the trigger with evidence. For R-02, the evidence was a measured projection of 30.9 h against a 16 h trigger.
  2. Open a new issue the same business day. Give it a new ID, copy only the context it needs and link the risk ID.
  3. Mark the risk "Triggered" and link the issue. The risk drops out of the risk review, but its history stays.
  4. Re-prioritize the issue. Likelihood no longer applies; use impact and how soon the damage lands.
  5. Escalate if the fix exceeds the owner's authority. Record any decision in the decision log and any work in the task tracker, then link both IDs.
  6. Close with evidence. Close the issue when the evidence exists, not when the action is assigned.

The same conversion applies to the other types. An invalidated assumption becomes a risk or an issue, as A-02 became I-01. A missed dependency becomes an issue.

Linking dependencies and decisions without mixing records

  • One fact, one home. A decision belongs in the decision log, a task in the plan or tracker, and an uncertainty in the RAID log. Each points to the others by ID.
  • Update, don't merge. When DEC-04 changed the migration scope, D-02 was updated with the new scope and a link. The decision text was not copied into the dependency row.
  • Log only dependencies that need watching. Finish-to-start links between your own tasks belong in the schedule. The RAID log keeps the cross-team and supplier dependencies that someone has to chase.

How to review a RAID log every week

A weekly rhythm keeps the log current without turning the meeting into a row-by-row reading:

  1. Monday by 17:00: owners update. Each owner changes status, adds evidence and sets the next review date. The log owner then filters for new entries, changes, P1s, triggered items, overdue or stale rows, decisions needed and closure candidates.
  2. Tuesday 10:00: 30-minute review.
    • 5 minutes: new entries. Confirm type, owner and priority.
    • 10 minutes: P1 and triggered items.
    • 5 minutes: overdue and stale rows. Reassign, escalate or close.
    • 5 minutes: decisions needed. Name the decider and the deadline.
    • 5 minutes: closures and conversions.
  3. Same day: escalate or decide. Take anything the meeting could not settle to its decider, following the escalation rules below.
  4. Within two hours: publish changes. Send a short list of what changed and feed the P1 items into the progress report.

Not every type needs the same rhythm:

TypeReview rhythm in the example
P1 issuesDaily check-in until resolved; daily stand-up for all P1s in cutover week (Nov 2–9)
RisksAt their next review date, weekly by default
AssumptionsBy the validation date, and again at each phase gate
DependenciesWeekly, plus a check 3 business days before the need-by date
Whole logFull sweep monthly and at each phase gate, including P3s

Escalation rules

Write the escalation rules down before you need them. These are the ones the example uses:

ConditionWho actsBy when
A risk trigger firesThe entry owner opens the linked issue and re-prioritizes itSame business day
A P1 issue needs a scope, date or budget changeThe log owner takes it to the sponsor with optionsSame business day
An assumption is not validated by its dateThe owner validates it or raises the risk that it is falseAt the next review
A dependency is unconfirmed 3 business days before need-byThe owner escalates to the provider's manager, then the sponsor3 business days before need-by
A row has had no update for two reviewsThe log owner reassigns it or closes it with a reasonAt the review
A decision is outside the project lead's authorityThe sponsor decides and the log owner records a DEC entryBefore the affected need-by date

RAID log vs project plan vs risk register

RecordQuestion it answersTypical contentsHow it relates to the RAID log
Project plan or scheduleWhat work happens, when, and who does it?Tasks, milestones, task-to-task dependencies, baselinesRAID entries link to plan tasks; actions live in the plan
Risk registerHow exposed are we to each uncertain event?Detailed scoring, triggers, responses, residual riskIn formal settings the register is authoritative and the R lane summarizes it
Issue logWhat is wrong now, and how will it be resolved?Evidence, owner, target date, resolutionOn smaller projects, the I lane is the issue log
Decision logWhat did we decide, who decided and why?Options, rationale, decider, dateDecisions are linked by ID, not stored as RAID rows
RAID logWhat could change delivery, and who is watching it?All four types with owners, status and review datesThe working view that connects the others

A small project can run on a plan and a RAID log alone. A larger program usually keeps the formal registers and uses the RAID log as the weekly working view.

Common RAID log mistakes

  • Calling an active problem a risk. If it has happened, it is an issue.
  • Owner = "IT". Name the person who will update the row.
  • No next review date. Without a date, "no change" is indistinguishable from "nobody looked".
  • Everything High. Use a written scale and let it set the review order.
  • Tasks in the log. Link to the plan and keep the log short.
  • Deleting or overwriting. Convert and close instead, so the history survives.

How to keep a RAID log in AFFiNE

AFFiNE has no ready-made RAID log template, as of October 5, 2026. You build one from a database block in a doc:

  1. Create a "RAID log" doc. Write the rules at the top: the letters you use, the priority scale, the status words and the review slot.
  2. Add a database block with these properties:
    • Type: select, with Risk, Assumption, Issue and Dependency
    • Owner
    • Priority: select, P1 to P3
    • Status: select
    • Trigger / validation / need-by: text
    • Next review: date
    • Links: text, for the IDs
  3. Add views. Add a Kanban view grouped by Type for the four-lane board. Add a table view sorted by Next review for Monday prep.
  4. Give heavy entries their own doc. Open the row and create a linked doc for its evidence and history, so the row itself stays short.
  5. Keep decisions as separate docs in a decision-log doc, and @-mention them from RAID entries. Backlinks then show which entries reference each decision.
  6. Draw dependencies in Edgeless mode with connectors, when a map is easier to read than the Links column.

Know the limits before you rely on it:

  • AFFiNE does not send date reminders or run automations on database rows, so the Monday update and Tuesday review stay a human routine. A recurring calendar invite does the reminding.
  • AFFiNE does not import CSV into a database, and a pasted Markdown table becomes a simple table block rather than a database. Use the CSV files with spreadsheets, and recreate the columns as properties in AFFiNE.
  • Comments need a cloud or self-hosted workspace.

FAQ

What is a RAID log in project management?

It is one register where a project tracks its risks, assumptions, issues and dependencies. Each entry has an owner, priority, status and next review date. The team reviews it on a fixed rhythm so that uncertainty and blockers turn into decisions early.

What does RAID stand for in a RAID log?

Usually Risks, Assumptions, Issues and Dependencies. Some teams use A for Actions or D for Decisions; state your expansion at the top of the log.

What is the difference between a RAID log and a risk register?

A risk register goes deeper on uncertain events only: scoring, triggers, responses and residual risk. A RAID log is broader and lighter, adding assumptions, live issues and dependencies. Where both exist, keep the register authoritative and link to it from the R lane.

How often should a RAID log be updated?

Owners should update an entry whenever it materially changes, and before each review. A weekly review suits most active projects. Check P1 issues daily, and run a full sweep monthly or at each phase gate.

Who should own the RAID log?

The project lead usually owns the log: its format, scales and review meeting. Each entry has its own owner, a named person who updates the row and escalates. Decisions beyond the project lead's authority go to the sponsor.

Can a RAID log include actions and decisions?

Yes, but only if your organization defines its letters that way, and then you should say so at the top of the log. Otherwise keep actions in the plan or task tracker and decisions in a decision log, and link them by ID.

Sources and method

The worked example, its people and its numbers are fictional and were built for this guide. External sources were checked on October 5, 2026:

  • Institute of Risk Management, Infrastructure Risk Special Interest Group, Short Guide to RAID (PDF; no date printed, file created October 2023). Source for the expansion, the "RAAIDD" variant and agreeing the level of detail.
  • M. L. Love, Risks, issues, and changes: help, I'm drowning in logs!, PMI Seminars & Symposium, 2002. Source for keeping tasks in the schedule and cross-referencing related log items.
  • D. Sankararajan and N. K. Shrivastava, Risks vs. issues, PM Network, June 2012. Source for the risk and issue distinction and their review rhythms.
  • J. Kinser, Don't make an ass out of you and me: using assumptions effectively, PMI Global Congress, 2010. Source for validating assumptions.
  • Association for Project Management, Glossary (no date shown). Source for the definitions of issue, assumption and dependency; there is no RAID entry.
  • ELEXON for Ofgem, Attachment C: Risks, Assumptions, Issues and Dependencies, version 1.1, February 14, 2019. Source for the public RAID log with type-specific columns.
  • PRINCE2 management products (independent reference, not the official manual). Source for PRINCE2's separate registers and logs.
  • AFFiNE product behavior was checked against the v0.27 release notes and AFFiNE's open-source code on October 5, 2026.