Routers Overview & Concepts

🔀 What are Routers?

Routers are automated systems that help you manage Quality Assurance and AI Products processing at scale. They automatically:

1. Select interactions from your team's conversations based on the rules you set

2. Distribute them to the right people for review or send them through AI Products processing

3. Track the work so you can see what's been assigned and completed

4. Run on a schedule so you don't have to remember to do it manually

Think of a router as your automated assistant that works 24/7 to ensure consistent monitoring and evaluation of your team's interactions with customers.

🧭 Router Types

There are two main types of routers, designed for different purposes.

📝 Scorecard Routers

Purpose: Human review and quality evaluation.

Scorecard routers assign interactions to your QA team or supervisors for manual review using scorecards.

This is ideal when you need human judgment to evaluate quality, compliance, or coaching opportunities.

Example Scenario

Your QA manager wants to review 10 calls per agent each week.

The router automatically:

• Selects the calls

• Distributes them evenly

• Assigns them to your QA analysts

What you need to configure

• Which agents' interactions to sample

• How many interactions per agent

• Who will review them (participants)

• What percentage of work each reviewer receives

• When the reviews should be completed

🤖 AI Routers

Purpose: Automated AI analysis.

AI routers send interactions through automated AI processing such as:

• Transcription

• AI tagging and Sentiment

• AI metrics

• AI Process (AutoQA)

No human review is required.

Example Scenario

Every night:

• All phone calls from the day are processed

• transcripts are generated

• Sentiment analysis runs automatically

Results appear in your dashboards the next morning.

What you need to configure

• Which interactions to process

• Which AI processes to run

• When processing should occur

⚠️ Important:
Scorecard processing and AI processing cannot be combined in the same router.

You must create separate routers for each workflow.

⚙️ How Routers Work

The Basic Process

If you’d like to see the step-by-step instructions for configuring the router, you can read this article here: Creating and Managing Routers.

1. You configure the router rules

2. You schedule when it should run

3. The router activates automatically

4. Interactions are sampled based on your criteria

5. Work is distributed to reviewers or AI processes

6. You track progress through assignments and dashboards

📊 Understanding Routers Sampling

Sampling means selecting a specific number of interactions per agent.

Example:

You have:

• 5 agents

• each handling 100 calls per week

You configure:

10 calls per agent

The router selects:

10 × 5 agents = 50 interactions total

This represents a 10% quality sample.

Why sampling matters

Reviewing every interaction isn't practical.

Sampling allows teams to:

• Monitor quality effectively

• Maintain manageable QA workloads

• Maintain statistical coverage

👥 Users vs Participants

Understanding these roles is essential.

Users

Users are the agents whose work is being evaluated.

Examples:

• Customer support agents

• Sales representatives

• Service specialists

Participants

Participants are the reviewers performing evaluations.

Examples:

• QA analysts

• Supervisors

• Compliance reviewers

Example

A router samples:

5 interactions × 10 agents = 50 interactions

These 50 interactions are distributed among 2 QA analysts.

⏳ Time Windows

Time windows determine which interactions qualify for routing.

Why time windows matter

A wider time window gives routers more interactions to choose from, increasing the chance of meeting sampling targets.

🏷️ Filters

Filters allow routers to focus on specific interaction types.

Examples:

• Phone calls only

• Interactions tagged as "Sales"

• Escalated conversations

• Specific product lines

Filters must be created in the system first, then selected when configuring routers.

🔄 Distribution Strategies

Distribution strategies determine how assignments are distributed among reviewers.

This applies only to scorecard routers (Manual QA).

Oversai currently supports three strategies:

• Round-Robin (recommended for most QA workflows)

• Random

• Sequential

Each strategy changes how interactions are assigned, but does not affect how sampling works.

Sampling determines how many interactions are selected per agent, while distribution determines who reviews them.

⚠️ Recommended default: Round-Robin

⚠️ If you’d like to learn in more detail how each of these distribution strategies works and see practical examples, you can read this article: Understanding Distribution Strategies in Scorecard Routers (Manual QA).

📌 Summary

Routers in Oversai automate how interactions are selected, distributed, and processed across your quality assurance and AI workflows.

Instead of manually assigning evaluations or running AI analysis, routers allow teams to define rules that automatically sample interactions, assign them to reviewers, or send them through AI processes on a schedule.

In this guide, you learned:

• What routers are and how they work

• The difference between Scorecard Routers and AI Routers

• How interactions are sampled and distributed

• Key concepts such as users, participants, filters, and time windows

• The different distribution strategies available for assigning work

Understanding these concepts will help you design routers that scale quality monitoring efficiently across your CX operation