Skip to content
⌘K
InboxAsk a human

Step 5: Handling complex routing and triage

Updated by Dustin

With your historical threads migrated and all of your data enrichment in place, the next step is setting up the building blocks that define your triage strategy. In Plain, those building blocks are labels (what a thread is about), teams (who owns it), business hours (when your team is working), and SLA policies (how fast you promise to reply).

When you're running a multi-region or tiered support org, routing breaks down if those primitives aren't structured cleanly before you touch the workflow builder.

What good looks like

Ideally, you’ll want your global routing matrix — labels, regional/tiered teams, shift schedules, tier + contract SLA policies, and escalation ladders — built and tested in isolation:

  • Orthogonal label taxonomy: Topic labels (Product > Compute, Billing) are separated cleanly from team labels (T1 Triage, Compute — EMEA, Compute — AMER, T3 Engineering) so you can filter and route on any combination of product × region × tier

  • Capacity-aware regional queues: Frontline and regional T2 team labels have auto-assignment and Maximum threads per user configured, while escalation-only queues (T3 Engineering) have auto-assignment turned off

  • Regional shift schedules: Each operating region has its own business-hours schedule (AMER Support Hours, EMEA Support Hours, APAC Support Hours) so an EMEA customer's SLA clock pauses at 18:00 Berlin, not 18:00 San Francisco

  • Tier-wide + contract-override SLA policies: Standalone SLA policies are linked to those regional schedules for both standard tiers (Enterprise — EMEA) and custom contract addendums (Contract — 15m First Response)

  • Predictable escalation ladders: Escalation paths step cleanly from regional T2 → regional lead → T3 engineering

Three design rules for multi-region & tiered ops

1. Keep topic labels and team labels strictly separate

In Plain, a team is a label with members and auto-assignment rules attached to it.

  • Use hierarchical topic labels (up to 3 levels, e.g. Product > Compute > Data) for reporting and categorization. Give every routine topic label a crisp one-sentence Description (the AI uses that description to decide when to apply the label in workflow actions), and toggle off Apply with AI on restricted operational labels (i.e. incidents).

  • Use team labels strictly for queue ownership. Because adding a team label to an unassigned Todo thread triggers auto-assignment immediately, applying the team label is routing the thread.

2. Understand how capacity-aware auto-assignment and handovers interact

Plain's auto-assignment isn't blind turn-by-turn round-robin — when a team label is applied to a thread, Plain evaluates every member of that team and:

  1. Skips unavailable agents: Anyone whose personal status is Offline (between shifts/lunch) or Away (PTO or a break) is excluded, and so is anyone who can't view the thread

  2. Skips agents at Maximum threads per user (5 by default): Anyone already holding their Maximum threads per user of open Todo threads is excluded

  3. Assigns to the least-loaded eligible agent: Among everyone left, the teammate with the fewest open Todo threads gets the thread (and if everyone is at capacity or offline, the thread waits unassigned and backfills automatically the moment an agent frees up or comes online)

Two handover rules your leads need to know:

  • Auto-assignment only fires on unassigned Todo threads. If T1 already owns a thread (or the Customer Agent is handling it live) and someone adds the Compute — EMEA team label, Plain will not overwrite the existing primary assignee. When handing a thread from T1 to T2 or across regional shifts in a workflow, add an Unassign thread step first, then apply the team label or use the Assign to team action.

  • Use Primary vs. Additional Assignees for T2 ↔ T3 pairing. Every thread has one primary assignee (who owns the customer relationship and SLA) plus optional additional assignees. When T3 engineering helps investigate a complex bug while T2 still owns customer comms, have T3 join as an additional assignee rather than transferring primary ownership back and forth.

3. How per-contract SLA overrides work ("last Set SLA policy wins")

Unlike legacy tools where SLA rules sit in a giant condition table inside the SLA settings page, Plain's SLA policies are standalone definitions (up to four targets: First response, Next response, First resolution, and Total resolution, plus an Imminent breach warning window and a linked Business hours schedule). A policy attaches to a thread when a workflow runs the Set SLA policy action, or when someone sets it in the thread's SLA policy field.

A thread can hold one SLA policy at a time. If multiple workflows run Set SLA policy on the same thread, the last workflow to run overwrites the earlier policy.

That mechanic is how you handle custom enterprise MSAs cleanly without creating 40 tiers:

  • Baseline workflow (runs first): Your baseline tier × region workflow sets Enterprise — EMEA (30m first response / 2h next response on EMEA Support Hours)

  • Contract-override workflow (runs immediately below it): Your contract-override workflow checks Account field "sla_contract" equals "acme-15m" and overwrites the policy with Contract — 15m First Response

Build sequence (in dependency order)

For the basic click-by-click setup of teams, labels, business hours, and SLA policies, see Step 5: Getting your threads to the right person in our small-team series.

Build these four objects in order so each one exists when the next step references it:

1. Configure Away mode and build your team matrix

  1. Under Settings → Away mode, set your workspace PTO behavior: choose whether an agent switching their status to Away unassigns all of their open Todo threads immediately (Unassign immediately, the default and recommended for regional queues so open threads re-enter rotation right away) or only when new activity arrives on a thread (Do not unassign immediately).

  2. Under Settings → Labels, create your topic hierarchy and your team labels:

    • Frontline / T2 regional teams: Toggle Auto-assign threads for this team ON and set Maximum threads per user (the default is 5 open Todo threads per support engineer — raise it to 10 based on thread complexity).

    • Escalation-only teams: Leave Auto-assign OFF so escalated threads land in that team's view to be triaged by the on-call lead rather than auto-assigned blindly (this is team/org dependent)

2. Create regional business-hours schedules

  1. Head to Settings → Business hours and click Add business hours

  2. Create one schedule per regional shift (AMER Support Hours, EMEA Support Hours, APAC Support Hours) and, if you offer 24/5 or 24/7 enterprise coverage, a 24/5 Global Coverage schedule

  3. Set each schedule's timezone and weekly operating windows, then save

(Note: Business-hours schedules don't include a built-in regional bank-holiday calendar. Ahead of a regional holiday, add an override to the schedule (or call createBusinessHoursOverride in the public business hours API). The override replaces that schedule's weekly hours on those dates, and the weekly hours apply again afterwards.)

Add a holiday override to a schedule

3. Create baseline and contract-override SLA policies

  1. Head to Settings → SLA policies and create your regional tier policies (e.g. Enterprise — AMER, Enterprise — EMEA, Scale — EMEA):

    • Configure First response time and Next response time (plus resolution targets if your contracts require them)

    • Set Send a notification before target (e.g. 10 minutes or 15 minutes) so threads transition to Imminent breach early enough for a workflow to alert a team lead

    • Select the matching regional schedule

    • Leave Track 24/7 unchecked on each target that should follow it

  2. Create a separate named policy for each custom contract addendum (Contract — 15m First Response, Contract — 24/7 Dedicated).

Operational SLA gotchas to keep in mind:

  • Only a human reply stops a response timer: Automated replies from the Customer Agent or a workflow do not achieve First Response or Next Response SLA targets.

  • Waiting for customer cancels the response clock; Paused for later does not: When the ball is in the customer's court, agents must set status to Waiting for customer. Using Paused for later (snoozing for an internal reminder while you still owe the customer an answer) keeps the SLA clock ticking.

4. Build regional escalation paths

  1. Head to Settings → Escalation paths and click New escalation path

  2. Build an ordered ladder for each product/region (e.g. Compute — EMEA Escalation: Level 1 Compute — EMEA → Level 2 EMEA Duty Lead → Level 3 T3 Engineering)

  3. Point each level at a team label rather than individual users wherever possible so shift rotations and PTO never break the path

Create an escalation path

Verification checklist

  1. Capacity + regional auto-assignment: Apply a team label to a normal unassigned Todo thread (auto-assignment skips test threads created by the “test thread” tool) and confirm it assigns to the least-loaded online member of that team. Set that teammate to Away and confirm a second unassigned thread skips them.

  2. Regional SLA pause check: Apply the Enterprise — EMEA tier from the sidebar in any thread and confirm the SLA countdown appears in the header.

  3. Escalation ladder: Set an escalation path on a test thread (with a Set escalation path workflow step or from the command palette), then use Escalate and verify ownership steps through the path cleanly.

Troubleshooting/FAQ

A thread got the regional team label, but it wasn't assigned to anyone (or stayed with its T1 owner).

Auto-assignment only triggers when a team label is added to a Todo thread with no primary assignee, and it skips anyone who is unavailable or at capacity. Check these three causes:

The thread already had a primary assignee when the new team label was added

  • When a T1 agent (or a Customer Agent running live) already owns a thread and adds a T2 team label like Compute — EMEA, Plain preserves the existing primary assignee.

  • Try this: When escalating or handing over a thread between teams in a workflow, add an Unassign thread step before applying the new team label or the Assign to team action.

Every member of that regional team is Offline, Away, or at max capacity

  • Agents who are Offline, Away, or holding their Maximum threads per user of Todo threads are skipped, leaving the thread unassigned in the team queue until someone frees up or starts their shift (at which point Plain backfills their queue automatically).

  • Try this: Check whether Maximum threads per user is set too low for morning backlog spikes, and confirm agents are setting Away (with immediate unassignment enabled in Settings → Away mode) when on PTO so their open Todo threads don't artificially cap the rotation.

Auto-assignment wasn't toggled on inside the team label's settings

  • Creating a team label and adding members is separate from enabling auto-assignment.

  • Try this: Open Settings → Labels, click into the team label, and confirm Auto-assign threads for this team is toggled on.

An enterprise customer with a custom contract SLA received the standard tier SLA instead.

Because a thread can only hold one SLA policy at a time, the last workflow on your Workflows page that runs Set SLA policy overwrites any policy set by an earlier workflow. Here are the two causes:

Your per-contract SLA workflow sits above your tier-wide SLA workflow

  • If the contract-override workflow runs first (setting the 15-minute policy) and the general Tier equals Enterprise workflow sits below it, the general workflow runs second and overwrites the custom policy with the standard 30-minute policy.

  • Try this: On the Workflows page, drag your per-contract SLA override workflows below your baseline tier-wide SLA workflows so the most specific rule runs last.

The thread's Account was attached after creation, and the SLA workflow only triggers on Thread created

  • If the Account was attached to the thread after it was created, a workflow that only listens to Thread created won't re-evaluate.

  • Try this: Add Thread account updated and Thread tier updated to the automatic trigger on your SLA workflows alongside Thread created.

SLA timers are still counting down overnight, on weekends, or on a regional bank holiday.

An SLA policy runs 24/7 unless it is explicitly linked to a business-hours schedule, and a schedule follows its weekly hours unless you add an override for a specific date. Check these three causes:

Business hours isn't enabled on the SLA policy (or points at the wrong regional schedule)

  • Each SLA policy needs that region's schedule (EMEA Support Hours) selected, and Track 24/7 unchecked on each target that should follow it.

  • Try this: Open Settings → SLA policies, inspect the policy, and confirm the right regional schedule is selected.

Agents paused the thread as "Paused for later" instead of "Waiting for customer"

  • Setting status to Waiting for customer stops the response SLA timer; Paused for later is an internal reminder and keeps the SLA clock running.

  • Try this: Train your team to use Waiting for customer whenever you've replied and are waiting on the customer's next message.

Today is a regional public holiday

  • Weekly business-hours schedules don't automatically close for date-specific holidays.

  • Try this: In Settings → Business hours, open the schedule and add an override for the holiday dates, or create one with createBusinessHoursOverride in the business hours API. The schedule returns to its weekly hours on its own afterwards.