KoltrixA triage matrix for support email: urgency vs impact
Sort support email by urgency and impact with a simple 3x3 matrix, see real examples placed in each cell, and get the right response for every combination.
On this page(8 sections)
- Two questions, asked of every email
- The matrix
- What each cell looks like in practice
- P0: Blocking, many accounts
- P1: Blocking, one account — or degraded, many accounts
- P2: Blocking, one user — degraded, one account — question, many accounts
- P3: Degraded, one user — question, one account
- P4: Question, one user
- Signals that bump an email up a level
- Running triage in a shared inbox
- A one-line triage note
- Calibrating as a team
- Bottom line
Answering support email in the order it arrived is fair in the way a single checkout line is fair: everyone waits the same, including the person whose house is on fire. A triage matrix gives you a quick, shared way to decide what jumps the queue, so the decision doesn't depend on who happens to be looking at the inbox.
Two questions, asked of every email
Every incoming support email can be placed using two questions:
- Urgency: how badly is the customer blocked right now?
- Impact: how many people or how much business does it affect?
Each has three levels.
Urgency
- Blocking: the customer can't do the thing they use you for. Can't log in, core feature broken, data missing, payments failing.
- Degraded: something is wrong but there's a workaround, or a secondary feature is affected.
- Question: nothing is broken. How-to questions, billing clarifications, feature requests, feedback.
Impact
- One user: a single person is affected.
- One account: a whole customer organization is affected, or a single person whose role makes it account-wide (an admin, the billing owner).
- Many accounts: the same problem is hitting multiple customers, which usually means a bug or incident on your side.
The matrix
| One user | One account | Many accounts | |
|---|---|---|---|
| Blocking | P2 | P1 | P0 |
| Degraded | P3 | P2 | P1 |
| Question | P4 | P3 | P2 |
The labels are just shorthand. Use whatever names your team likes, but keep the shape: the top-right corner is an emergency, the bottom-left is routine.
What each cell looks like in practice
P0: Blocking, many accounts
Examples: login failing for multiple customers; emails from several accounts all saying the dashboard won't load; a sudden wave of "my data is gone."
Response: This is an incident, not a support ticket. Alert engineering immediately, post an internal heads-up, prepare one status message to reuse for every inbound email about it, and keep customers updated on a fixed cadence. Individual replies can be short; consistency matters more than personalization.
P1: Blocking, one account — or degraded, many accounts
Examples: a customer's admin can't access their workspace; a whole team's integration stopped syncing; several customers report slow exports with a workaround.
Response: Same-hour response during business hours. A named owner takes it and stays on it. Pull in engineering early. Give the customer a workaround if one exists and a time for the next update.
P2: Blocking, one user — degraded, one account — question, many accounts
Examples: one person locked out after changing their phone; an account's reports showing the wrong time zone; many customers asking the same question after a pricing change.
Response: Within your standard urgent target. The "question, many accounts" cell is easy to miss: it isn't urgent for any one customer, but a pattern of identical questions usually means something confusing happened, like a release note that didn't land or a UI change. Answer them and tell whoever owns the confusing thing.
P3: Degraded, one user — question, one account
Examples: a button misaligned for one user's browser; an account admin asking how to set up roles for their team.
Response: Standard queue, normal target. These are the bread and butter of support. Answer thoroughly; a good answer to an admin's question often prevents several future emails from their team.
P4: Question, one user
Examples: "Can I change my display name?"; a feature suggestion; general feedback.
Response: Standard or low-priority target. Many of these can be answered quickly with a snippet or a help article link. Don't let them pile up for days, but don't let them interrupt higher cells either.
Signals that bump an email up a level
Some details should move an email up, regardless of where it first lands:
- Security or privacy concerns. A report of a vulnerability, a suspected account takeover, data appearing in the wrong account. Treat these as at least P1 and route them to whoever handles security.
- Money moving incorrectly. Double charges, failed payouts, invoices sent to the wrong customer.
- A deadline the customer mentions. "Our launch is tomorrow" makes a degraded issue effectively blocking.
- Repeated contact. A third email about the same unresolved issue deserves more attention than its content alone suggests.
And some details don't change priority, even though they feel like they should: an angry tone, an all-caps subject line, or a threat to cancel. Respond to those with care, but triage by the actual problem.
Running triage in a shared inbox
The matrix only helps if someone applies it as mail arrives. A workable routine:
- One person triages per shift. They scan new mail and place each email, rather than everyone cherry-picking.
- Label the priority. A short label (P0–P4, or Urgent / Standard / Low) makes the queue sortable and visible to everyone.
- Escalate P0 and P1 out loud. Post in your team channel with a link; don't rely on someone noticing a label.
- Look for patterns across "one user" emails. Three separate one-user reports of the same problem in an hour is a many-accounts problem in disguise.
- Re-triage when facts change. A P3 that turns out to affect a whole account becomes a P2. Update the label.
A one-line triage note
When the triager places an email, a short internal note helps the next person:
P1 — blocking, one account (admin can't sign in, SSO error since 9am). Owner: Priya. Next update promised 11:30.
Calibrating as a team
Disagreement is normal at first. Once a week, pick five recent emails and have two people place them independently. Where you disagree, talk about why and adjust your definitions. After a few rounds, triage becomes fast and consistent, and new teammates can learn it from your examples.
Bottom line
- Triage every support email by two questions: how blocked is the customer, and how many are affected?
- A 3x3 matrix gives a shared, fast way to decide what jumps the queue.
- Many identical one-user reports usually signal a many-accounts problem.
- Security, money and stated deadlines bump priority; tone and threats don't.
- Calibrate with real examples so everyone places emails the same way.
Start with Koltrix
Your domain, one inbox, and an API that sends.
A team inbox where AI sorts and drafts (nothing is sent without your click), plus the transactional API and SMTP relay your product sends with. 7 days free, no card.
