Core concepts

Three things decide how work moves through the Review Platform. A review kind says what a review is and how urgent it is, a queue says who works it, and the queue's prioritization strategy says in what order. Playbooks fill the queues; reviewers empty them.

Review kinds

A review kind is a template for one type of review your team performs: what is being evaluated and how urgently it must be completed. Each review kind has:

• Name. Identifies the review in queues and in the playbook builder.
• Base priority. A value from 1 to 10 that sets the review's importance relative to other review kinds.
• SLA. Hours from creation until the review's due date.
• SLA ramp factor. How fast the review's priority rises as it approaches its due date, in queues sorted by hybrid priority.

With separate review kinds, a time-sensitive fraud investigation carries more urgency than a routine document check.

Queues

A queue holds the reviews for one team or workflow. Review kinds define what needs reviewing; queues decide who reviews it and in what order. For each queue you:

• Set the order. Choose one of the four prioritization strategies below.
• Bind review kinds. Choose which review kinds land in the queue. A kind bound to several queues has its new reviews spread across them; a kind bound to none lands in the Default queue.
• Bind reviewers. Choose who can pull from the queue. Any reviewer can still open the queue's list and assign an unassigned review to themselves.

A typical structure:

• General queue. Standard document verifications and low-risk reviews, worked by your L1 team.
• Escalations queue. The review kinds your playbooks raise when a case needs a second look.
• Specialist compliance queue. High-risk, complex review kinds for your most experienced reviewers.

Specialists then spend their time on the cases that need their expertise, and routine reviews move through the general team.

Prioritization strategies

Each queue orders its reviews with one of four strategies, named here as the queue settings show them.

Due at

Earliest due date first. You never miss a deadline, but the order ignores how important one review kind is relative to another. Best for teams where all review kinds have similar importance and meeting deadlines is the main concern.

Priority

Highest base priority first, regardless of when a review was created or when it is due. Best for teams that handle both routine and critical reviews, where certain kinds (such as fraud investigations) always take precedence.

Created at

Newest first, so the most recent review is always at the top of the queue. Best for teams that work the queue continuously and care most about the newest arrivals.

Hybrid (recommended)

Calculates each review's priority from base priority, proximity to the SLA deadline, and creation time together. It is the recommended strategy because it reflects how urgency behaves in a real operation.

The Review Platform recalculates each review's effective priority as it ages:

Priority = Base priority × (1 + Progress ^ SLA ramp factor)

• Base priority. The review kind's starting priority (1-10).
• Progress. The share of the SLA already elapsed, from 0 to 1.
• SLA ramp factor. How fast the escalation happens. At 1, priority rises in proportion to how close the deadline is. Above 1, it rises slowly at first and rapidly near the deadline, which gives newer reviews time before older ones become urgent. Below 1, it rises fast early and gradually later, so older reviews gain priority quickly.

A review can at most double its base priority. Two reviews sitting in the same queue, both with a ramp factor of 1, show why that matters:

Neither simple strategy gets this right:

• Priority puts Review A first, and Review B breaches its SLA in 12 minutes.
• Due at puts Review B first, but would keep doing so even if Review A were far more urgent.
• Created at ignores both signals.

Hybrid weighs them together. Review B has the lower base priority, but it is almost out of time, so it surfaces first.

Reviewers and teams

Reviewers are members of your organization you add in Manage reviewers. For each reviewer you choose which queues they can pull from and set an expected number of daily reviews.

Teams group reviewers under a name and a description. The team view on Queues & Teams lists each member with their assigned and started reviews and the ones completed in the last 24 hours.

Percy agents do not work queues. An agent runs as a step inside a playbook, and the playbook's rules decide whether a review is raised at all.

Next steps

With the vocabulary in place, Getting started sets up your first review kinds and queues. What you configure per reviewer is in Reviewers and teams.