Topic 2.1 — Work Triage

In one sentence: Triage is the clinical discipline of rapidly assessing each incoming piece of work for urgency, importance, and fit, then routing it — fast-track, standard queue, or a respectful "no" — so that scarce capacity flows to where it does the most good.

Why this matters in health and care#

Every clinician understands triage: when demand exceeds capacity, you do not treat in arrival order, you assess and sort by need. Digital and operational teams face the same reality — far more requests than hands — yet often handle work first-come-first-served, or worst-of-all by who is loudest. The result is that a low-risk convenience feature gets built while a patient-safety defect waits.

The analogy is more than rhetorical. Health and care organizations already trust triage as a fair, defensible way to allocate scarce resource under pressure; applying the same explicit, criteria-based sorting to digital demand inherits that legitimacy. It also protects the things that must not slip: safety-critical fixes, statutory deadlines (for example a regulatory or information-governance obligation), and equity of access. Triage done well means that when you tell a director "not yet," you can show why — against consistent criteria, not politics. It sits between intake (Topic 2.0) and prioritization (Topic 2.2): intake makes demand visible, triage assesses and routes it, prioritisation orders what survives.

Core concepts#

Triage is a quick, structured assessment that answers three questions for each item: How urgent is it? How important is it? Where should it go? It is deliberately fast and coarse — a sort, not a full analysis. The clinical parallel is a triage nurse categorizing arrivals (immediate / urgent / standard / signpost elsewhere), not a full diagnosis.

Urgency versus importance is the foundational distinction, popularized in general management by the Eisenhower matrix. Urgent work is time-sensitive: a statutory deadline, a live safety risk, an expiring contract. Important work has high value or consequence but may not be time-critical. The classic trap is letting the urgent-but-unimportant crowd out the important-but-not-yet-urgent — which in digital terms means firefighting forever while the strategic platform work never starts.

Categorization is the triage output: each item is placed in a lane. A simple, robust scheme is fast-track (safety, statutory, or trivially small and clearly valuable — bypasses the queue), standard (goes forward to full prioritization), and signpost / decline (belongs to someone else, duplicates existing work, or should not be done). This mirrors clinical streaming, as codified in tools like the Manchester Triage System.

Sizing at triage is a rough order-of-magnitude estimate — t-shirt sizes (S/M/L/XL) or a quick effort band — enough to route and to feed later prioritization, never a detailed estimate. Precision here is false economy.

Cynefin (Snowden) helps triage judge how much discovery an item needs: clear/obvious items can go straight to delivery; complicated ones need expert analysis; complex ones need discovery and experimentation (see Topic 3.4 — Discovery Phases); chaotic ones need an incident response. Triage should recognize the domain and route accordingly rather than forcing everything down one path.

Best practices#

1. Define explicit triage criteria and publish them. Agree the factors that determine a lane: clinical-safety risk, statutory or regulatory deadline, size/effort, strategic fit, and dependency on or from other work. Weight them, write them down, and share them. Explicit criteria are what make triage fair, fast, and defensible — and what let you say "no" without it being personal.

2. Put safety and statutory drivers above negotiation. Anything with a genuine patient-safety or legal-deadline driver is fast-tracked or escalated, full stop. Guard this ruthlessly, because everything wants to claim urgency; require the safety or statutory driver to be evidenced (a clinical-safety flag, a named regulation, a hazard log entry), not merely asserted.

3. Make triage a small, cross-functional decision, not a solo act. Triage benefits from a handful of perspectives — delivery, clinical/operational, information governance, and architecture — because a single triager will systematically under-weight risks outside their expertise. Keep the group small enough to be fast; a triage of two or three well-chosen people beats a committee of twelve.

4. Run triage on a predictable cadence. Hold a short, regular triage session — often weekly — so requesters know when their item will be assessed and the queue never silts up. A fixed cadence turns "when will someone look at this?" into a knowable date, reinforcing the acknowledgement promise made at intake.

5. Separate the fast lane and protect it. Reserve explicit capacity for fast-track work (safety, urgent statutory, tiny high-value fixes) so it does not compete in the main prioritization. Equally, cap it: if everything is "fast-track," nothing is, and the discipline collapses. A small standing allocation for genuine urgent work is healthier than heroics.

6. Size roughly, once, and record the basis. Give each item a t-shirt size at triage and note the key assumption behind it. This is enough to route and to inform prioritization without burning discovery effort on things that may be declined. Detailed estimation is earned by surviving triage, not spent before it.

7. Say no — and signpost — with respect. A good triage declines work clearly and kindly: it explains the criteria, names the alternative (another team, an existing product, a later cycle), and records the decision. Signposting — "this belongs to the service desk," "the shared care record programme already covers this" — is often more valuable than a flat refusal. The ability to decline is what makes a "yes" mean anything.

8. Watch for equity and demand bias. Well-resourced services submit polished requests; frontline and social-care voices may not. Triage should actively check that criteria are applied evenly and that quieter but high-need demand is not systematically down-ranked. Fairness of process is a governance requirement, not a nicety.

9. Separate triage from delivery decisions. Triage decides urgency and routing; it should not quietly become the place where scope, solution and funding are settled without the right people. Keep triage fast and criteria-based, and pass genuinely complex or contested items to a proper prioritization forum (see Topic 2.2 — Work Prioritization) rather than resolving them in the corridor. Confusing the two overloads triage and hides real trade-offs.

10. Review triage decisions and calibrate. Periodically re-examine a sample of past triage calls: did the "urgent" items turn out to matter, and did anything you down-ranked later blow up? Feed the pattern back into the criteria and into how triagers are trained, so the system learns rather than repeating the same misjudgements. Calibration across triagers keeps decisions consistent regardless of who is on rota.

Questions to discuss with your team#

1. What evidence will we require before we accept that something is genuinely urgent — and who gets to challenge an urgency claim? This matters because "everything is urgent" is the failure mode that quietly destroys triage: if urgency is self-declared and unchallenged, the fast lane becomes the only lane and the discipline collapses. In health and care the difficulty is that some urgency claims are entirely real — a safety defect that could let a clinician miss a flagged patient, a statutory information-governance deadline — so the answer is not scepticism but a clear evidentiary bar: a named regulation, a hazard-log entry, a clinical-safety flag, not merely a stressed requester's assertion. The team should debate where that bar sits and who holds it, because the person with authority to say "your urgency claim is not evidenced" needs both the standing and the criteria to make it stick. Consider concrete tests: for the last ten items labelled urgent, how many could produce the evidence, and what would have happened if they had waited a cycle? A good answer defines the specific proofs that upgrade an item to the fast lane and pairs them with a hard cap on fast-track capacity, so that genuine safety and statutory work is protected precisely by refusing to dilute the lane. The honest version admits that most "urgent" requests are important-but-not-time-critical, and that naming that distinction out loud is what makes the safety-critical minority visible.

2. Who actually needs to be in the room for a triage decision to be safe, and how do we keep it fast enough to happen weekly? This is the core tension of triage: a single triager systematically under-weights the risks outside their own expertise — a delivery lead misses an information-governance trap, a technologist misses a clinical hazard — yet a committee of twelve is too slow to run on a predictable cadence and turns a coarse sort into full analysis. The debate should identify the minimum viable set of perspectives for your demand, which in most health and care settings means delivery, clinical or operational safety, and information governance, and then defend that group against both shrinkage (back to a solo act) and bloat (a stakeholder free-for-all). Concrete angles: time-box a real triage session and see how many items you can stream in the slot; look at past decisions that went wrong and ask which missing voice would have caught the problem. A strong answer treats the composition as a safety control, not a courtesy — the clinical-safety officer is there so hazard-flagged work is never buried, not to be consulted — while ruthlessly protecting speed by sizing roughly and deferring depth to discovery. The good version also names what happens when a needed expert is absent: does triage proceed and risk a blind spot, or defer the item, and who decides?

3. How will we know whether our triage is quietly down-ranking the quieter, higher-need voices? Equity of demand is easy to endorse and hard to verify, because well-resourced services submit polished, well-evidenced requests while frontline and social-care colleagues may raise real needs in ways that score poorly against criteria written for articulate requesters. This matters in health and care specifically because the public sector equality duty gives fairness of process statutory weight, and because the populations least able to lobby are often those with the greatest need. The team should debate how bias actually enters their criteria: does "strategic fit" implicitly favour national programmes over local ones, does requiring crisp evidence disadvantage those without analysts, does seniority creep back in through who gets heard? Concrete checks include periodically sampling declined and low-ranked items by originating service to see whether a pattern of quieter voices losing emerges, and inviting a frontline or social-care perspective into the criteria review, not just the triage session. An honest answer accepts that a process applied "consistently" can still be systematically unfair if the criteria themselves are tilted, and commits to a specific monitoring mechanism rather than a good intention. The best answers treat equity as something you measure in the outcomes of triage, not something you assume from the fairness of the room.

4. Where is the line between triaging an item and actually deciding it — and what will we refuse to settle in the triage session itself? Triage is meant to be a fast, coarse sort — how urgent, how important, where does it go — but the gravity of the room constantly pulls it towards full analysis: once a few informed people are looking at an item, it is tempting to scope the solution, argue the estimate, and settle the funding then and there. That drift is dangerous because it slows triage below a weekly cadence, and because it quietly resolves scope and trade-off decisions without the right people or the proper prioritization forum in the room. The team should agree explicitly what triage is not allowed to conclude: it can route an item to discovery or to prioritization, but it should not commit delivery, redesign a service, or negotiate a supplier's scope in the corridor. Concrete angles include time-boxing each item to a few minutes and watching where the session overruns, and reviewing recent decisions to see which ones were really prioritization or design calls dressed up as triage. A good answer names the specific decisions that must be handed onward — contested priority, solution choice, funding — and gives triagers permission to say "this is beyond triage" without it feeling like a failure. The honest version admits that pushing depth downstream will sometimes feel slower in the moment, and accepts that as the price of keeping triage fast and keeping real trade-offs visible.

5. How will we know, months from now, whether our triage calls were actually good — and how will that learning change the criteria? Triage decisions are made fast and under uncertainty, so some will be wrong, and without a feedback loop the same misjudgements repeat indefinitely and the criteria never earn their authority. The tension is that calibration takes effort nobody has time for, and revisiting old calls can feel like blame rather than learning, yet a triage system that cannot show it improves is just consistent guessing. The team should debate a concrete review mechanism: periodically sampling past decisions to ask whether the items marked urgent turned out to matter, and whether anything down-ranked or declined later blew up into an incident, a complaint, or a safety event. Health and care raises the stakes, because a mis-triaged safety defect or a missed statutory deadline is not merely an efficiency loss — so the review should weight near-misses on those dimensions especially heavily. Concrete angles include tracking a small number of decisions per cycle, comparing calls made by different triagers on the rota to surface inconsistency, and feeding the patterns back into the weighting of the criteria and into how triagers are trained. A strong answer commits to a specific, low-overhead cadence for this review and treats divergence between triagers as a signal to calibrate, not a person to correct; the honest version accepts that some down-ranked items should have been done, and builds that admission into a better rule rather than hiding it.

6. How much discovery does an item need before we can even route it — and how do we stop genuinely uncertain work being streamed straight to build? Triage has to judge not just urgency and importance but how knowable an item is, because a clear, well-understood request can go straight to delivery while a complex or novel one needs discovery and experimentation first — and confusing the two is how "we want an app" ends up in a sprint before anyone has understood the problem. The Cynefin distinction is the useful lens here: clear items can be streamed, complicated ones need expert analysis, complex ones need discovery, and chaotic ones need incident response, so triage should be routing by domain, not forcing everything down the delivery path. The difficulty is that uncertain work often arrives wearing the confident language of a solution, and a rough t-shirt size can give false comfort that something is understood when only its apparent shape is. The team should debate what signals mark an item as needing discovery — unclear users, contested outcomes, novel clinical pathways, dependencies no one can yet map — and agree that routing such work to discovery is a legitimate triage outcome, not a delay. Concrete angles include looking back at delivered work that failed or was reworked and asking whether triage should have sent it to discovery first, and checking that the rough size recorded at triage carries its key assumption so later stages know how solid it is. An honest answer resists the pressure to look decisive by committing everything to build, and accepts that naming an item "complex, needs discovery" is often the most useful and most protective call triage can make.

In practice: a health & care example#

A community and mental-health trust's digital team faced a backlog it could not see the top of. Everything arrived as "urgent." The team stood up a weekly triage board: a delivery lead, a clinical safety officer, and an information-governance lead, working from published criteria (safety risk, statutory deadline, rough size, strategic fit, dependencies).

Each intake item was assessed in a few minutes and streamed into three lanes. A defect in a risk-assessment form that could let a clinician miss a flagged patient was fast-tracked that day. A request for a cosmetic dashboard change was sized "S" but streamed to the standard queue with low strategic fit. A supplier's proposal to replace a working system was declined and signposted to the existing roadmap, with the reasoning recorded. A vague "we want an app" idea was routed to discovery under a Cynefin "complex" reading, rather than into delivery.

The effect was cultural as much as operational. Because the criteria were visible and applied consistently, a "not now" from the triage board was accepted where a "not now" from an individual manager had previously been fought. The safety officer's presence meant genuine risk items were never buried, and the trust could demonstrate to its board and regulator that safety-critical work was systematically prioritized.

Sector lenses#

Startup#

In a small digital-health company the triage "board" may be two or three people around a laptop, and the dominant pressure is not internal politics but external pull — the customer who threatens to churn, the investor's feature request, the pilot site that wants a tweak. The scarce resource is engineering time against runway, so triage criteria collapse to a sharp few: does this unblock a paying customer or a regulatory gate, is it a genuine safety defect, and is it small? The fast lane exists but must be capped hard, because at this scale a single "urgent" custom build for one client can consume a sprint the whole company needs. The healthiest startups learn early to signpost respectfully — "not in this release, and here's why" — because the alternative is a product bent out of shape by whoever shouted last.

Small business#

An established small provider — a GP practice, a community pharmacy, a care home, or a small health-tech supplier with real patients but only a handful of staff — faces the full weight of clinical-safety and information-governance duties with none of the enterprise's dedicated roles to discharge them. Triage here is rarely a scheduled board; it is the practice manager or clinical lead deciding, between appointments, which of the supplier email, the failing integration and the CQC-driven change gets attention this week. The scarce resource is not runway but the time and attention of one or two people who are also delivering care, so the criteria must be brutally simple and the safety-or-statutory test still non-negotiable even when there is no dedicated officer to apply it. The practical move is to borrow rather than build capability — lean on a supplier's or a commissioning body's published criteria, and reserve the tiny local fast lane strictly for genuine safety and regulatory items — because an SME cannot afford to firefight every request and still keep the lights on. Saying no and signposting matters even more here: with no slack in the system, an un-triaged "yes" to a nice-to-have directly displaces care or compliance work that has nowhere else to go.

Enterprise#

An NHS trust or ICS triages high volume across many services, so triage becomes a standing cross-functional forum with published criteria, a fixed cadence, and named roles — typically delivery, a clinical safety officer, and an information-governance lead. The value of that formality is defensibility: when the board or a regulator asks why a request waited, the criteria and the decision log answer for it, and a clinical safety officer in the room ensures hazard-flagged work is never buried. Scale also breeds duplication and cross-site politics, so triage must de-conflict five versions of the same request and hold the line against seniority-based queue-jumping. The fast lane needs real, ring-fenced capacity here, because a large organization generates a steady stream of genuine safety and statutory items that cannot wait for the next quarterly cycle.

Government#

At NHS England, DHSC, or a local authority, triage operates on demand that is often already mandated, so the question shifts from "should we do this?" to "how urgent and how large, given we must." Statutory deadlines, ministerial commitments, and legal obligations dominate the urgency axis, and the evidence bar for a claimed deadline is a named regulation or a formal commitment, not an assertion. Accountability is public and adversarial — the National Audit Office and select committees may examine why work was or was not prioritized — so the triage decision log is a formal governance record. Equity carries statutory weight through the public sector equality duty, so a government triage process must actively check that thinly-resourced local authorities and frontline services are not systematically out-competed by well-funded national programmes with polished submissions.

Common failure modes#

• Everything is urgent. With no evidenced test for urgency, every requester claims it, and the fast lane becomes the only lane. Fix: require evidence for safety/statutory claims; cap fast-track capacity.
• Triage by seniority. The most senior requester's item wins regardless of merit. Fix: explicit, published criteria applied by a group.
• Triage that becomes full analysis. The session tries to fully scope and estimate each item and grinds to a halt. Fix: time-box; size roughly; defer depth to discovery.
• No one ever says no. Everything is accepted "to keep the peace," and the backlog becomes a graveyard. Fix: empower the triage group to decline and signpost.
• A single triager. One person's blind spots (often information governance or clinical safety) become systemic. Fix: cross-functional triage.
• Invisible criteria. Decisions feel arbitrary and are re-litigated endlessly. Fix: publish the criteria and the decision log.

Maturity model#

Checklist#

• Triage criteria are explicit, weighted, and published.
• Safety and statutory drivers require evidence, not assertion, and are fast-tracked.
• Triage is done by a small cross-functional group, not one person.
• Triage runs on a predictable, published cadence.
• A capped fast-track lane exists and is protected.
• Every item gets a rough size and a recorded routing decision.
• The team can and does decline work, with respectful signposting.
• Decisions are logged so they are not re-litigated.
• Discovery-needing (complex) items are routed to discovery, not straight to build.
• Equity of demand — including quieter frontline and social-care voices — is checked.

Key sources#

• Snowden, D. & Boone, M. — the Cynefin framework, A Leader's Framework for Decision Making (Harvard Business Review).
• Covey, S. / Eisenhower matrix — urgency versus importance.
• NHS England — DCB0129 clinical risk management (evidencing safety-driven prioritisation).
• ITIL 4 (AXELOS / PeopleCert) — service request categorisation and streaming.
• NHS Digital Service Standard & NHS Service Manual — prioritising the most important user needs.
• Manchester Triage Group — the clinical triage categorisation model (as the domain analogy).

References#

1. Triage — Wikipedia — https://en.wikipedia.org/wiki/Triage
2. Time management (Eisenhower method / matrix) — Wikipedia — https://en.wikipedia.org/wiki/Time_management
3. Manchester Triage System — Wikipedia — https://en.wikipedia.org/wiki/Manchester_Triage_System
4. Cynefin framework — Wikipedia — https://en.wikipedia.org/wiki/Cynefin_framework
5. Information governance — Wikipedia — https://en.wikipedia.org/wiki/Information_governance
6. Health equity — Wikipedia — https://en.wikipedia.org/wiki/Health_equity
7. Prioritization — Wikipedia — https://en.wikipedia.org/wiki/Prioritization
8. Snowden, D. & Boone, M. — A Leader's Framework for Decision Making (Cynefin) — Harvard Business Review — https://hbr.org/2007/11/a-leaders-framework-for-decision-making
9. DCB0129: Clinical Risk Management: its Application in the Manufacture of Health IT Systems — NHS England — https://digital.nhs.uk/services/clinical-safety/documentation
10. ITIL 4: service request categorisation and management — AXELOS / PeopleCert — https://www.peoplecert.org/itil
11. NHS Service Standard — NHS England — https://service-manual.nhs.uk/standards-and-technology/service-standard
12. Manchester Triage Group — clinical triage categorisation model — Royal College of Emergency Medicine / Manchester Triage Group — https://www.triagenet.net/