Use this guide to run a structured weekly performance review for your support team in Fin Operator, the same approach our own customer support team uses. You'll go from high-level conversation volume through to individual conversation investigation, covering handling time, time to close, CX score, and topic breakdowns, all in a single thread.

This guide is for support managers and team leads. Whether you're putting together an RCA (Root Cause Analysis) or keeping a regular pulse on how the week went, the seven-step sequence below works in whatever depth you need.

Give Fin Operator your scope at the start. It'll ask clarifying questions if anything's missing, but the more context you give upfront, the faster you'll get to the good stuff. A good opening message usually covers:

Start here. Volume context is what makes everything else make sense. A big week-over-week spike often explains why handling time is up, why CX dipped, and why certain inboxes are struggling. It's the foundation of any honest RCA.

Example prompts:

Look for inboxes where volume jumped significantly. That spike is usually the root cause of the performance story, so name it early and let it thread through the rest of your RCA.

After establishing conversation volume, add your efficiency metrics (handling time and time to close), segmented by inbox, with a week-over-week comparison. The relationship between the two tells you a lot.

Example prompts:

The pattern to watch for: handling time is low but time to close is high. That usually means conversations aren't being held up by complexity: they're sitting idle. Waiting on a handoff, waiting on a third party, or just sitting in a queue. That's a different problem from an inbox where both metrics are high, which points more to workload or complexity.

With volume and efficiency metrics in place, add CX score (Intercom's AI-predicted customer satisfaction score) by inbox, with a week-over-week comparison. Make sure you're looking at the volume of rated conversations alongside the score itself. The score without the sample size can be misleading.

Example prompts:

The inboxes that matter most for your RCA are the ones with a meaningful CX drop and high volume. A low score on three conversations is noise. A low score on 200 is a signal worth investigating.

Once you have volume, handling time, time to close, and CX score across your inboxes, ask Fin Operator to synthesize them. It'll identify which two or three inboxes are consistently underperforming across all three metrics.

Example prompt:

From here, narrow the rest of the review to those priority inboxes. No need to go deep on everything. Two or three is a practical limit: drilling into more than three inboxes in a single review tends to surface more data than can be acted on in one cycle.

For each priority inbox, drill into the topic and subtopic breakdown. This is where you start to understand what's actually driving the numbers: whether the problem is concentrated in one specific area or scattered across the inbox.

Example prompts:

Look for subtopics with handling times far above the inbox average, subtopics where time to close is much longer than handling time (parked conversations again), and subtopics with consistently low CX at meaningful volume. Those are your smoking guns.

After identifying the highest-impact subtopics in Step 5, pull the actual conversations behind them. This is where you go from patterns to evidence: understanding what specifically made something hard to handle or why customers were dissatisfied.

Example prompts:

Fin Operator can check conversation attributes, tags, and AI-assigned fields to help you classify what you're looking at. Separating bug-related cases from complex-but-non-defect cases is really useful here. A long handling time caused by a known bug has a very different remediation path than one caused by a process issue or a capacity crunch.

Before you close out the review, check whether any conversations from the week are still sitting open in your priority inboxes. Unresolved conversations don't just affect this week's numbers. They can drag CX into the following week too.

Example prompts:

Prioritize any conversations that are open (not snoozed) or flagged with a high-value customer tag. Reassign or action those before closing out the review.

A few limitations worth knowing before you run your first review:

Here's the full seven-step sequence for running a weekly support performance review in Fin Operator, from conversation volume through to open conversations:

The whole review stays in one thread. Fin Operator holds your context (date range, inbox filters, previous results) so you're not repeating yourself at every step. Once you've run it a couple of times, it starts to feel less like pulling a report and more like thinking out loud with someone who already has all the data.