CSAT, NPS, or CES? Choosing the Right Feedback Metric for Your JSM Team

Every piece of content about choosing between CSAT, NPS, and CES eventually lands on "it depends on your goals." That's not wrong, but it's not useful. Here's an opinionated take with actual recommendations for JSM teams.

Quick Definitions

CSAT — Customer Satisfaction Score

Asks: "How satisfied were you with this interaction?"

Typically a 1–5 scale, a 1–10 scale, or a thumbs up/down. Sent immediately after a ticket is resolved. Measures how a specific transaction felt. High CSAT on a ticket means that ticket went well — it doesn't tell you whether the customer will churn next quarter.

NPS — Net Promoter Score

Asks: "How likely are you to recommend us to a colleague?" (0–10)

Respondents are classified as Detractors (0–6), Passives (7–8), or Promoters (9–10). Your NPS is the percentage of Promoters minus the percentage of Detractors. Sent periodically — quarterly or after a significant milestone — not after individual tickets. Measures relationship health and loyalty, not transaction quality.

CES — Customer Effort Score

Asks: "How easy was it to get your issue resolved?" (1–7, Very Difficult to Very Easy)

Sent after resolution, especially for complex or multi-touch tickets. Measures friction in your process. Research from Gartner suggests reducing effort correlates more strongly with loyalty than delighting customers does — meaning a low CES score is often a stronger churn predictor than a low CSAT score.

Which Metric Fits Which Team

Use CSAT if: You run a high-volume IT service desk handling transactional requests — password resets, access requests, software installs, hardware orders. You want per-ticket signal you can act on quickly. You want to coach agents based on actual feedback rather than gut feel.

CSAT is the right default metric for most internal IT service desks. The feedback loop is tight: ticket resolved, survey sent, score captured, agent briefed. You can run weekly CSAT reviews and have meaningful data within the first month.

Use NPS if: You run support for an external customer-facing product — a SaaS platform, a B2B service, an external helpdesk. You care about whether customers would recommend your service, not just whether they were satisfied with their last ticket. You want to track loyalty trends over time.

NPS is not useful on a per-ticket basis and shouldn't be sent that way. One NPS survey per customer per quarter is the right cadence. For internal IT support, NPS is usually overkill — internal employees don't "recommend" IT support in the same way.

Use CES if: Your team is getting complaints about process complexity. Tickets take multiple touches, escalations are common, customers frequently reopen tickets or ask for updates. CES surfaces where the friction lives — and friction, not rudeness, is usually what drives poor satisfaction in IT service desks.

CES is particularly useful for incident management and complex service requests where even a good outcome involves an exhausting journey. A customer who waits 4 hours for a resolution, talks to three different people, and gets transferred twice may give a reasonable CSAT score (the outcome was fine) but a terrible CES score (the process was awful).

The Comparison Table

Note on benchmarks: These ranges are for IT and B2B service desk contexts, not consumer support. B2C benchmarks run higher. Don't compare your IT service desk NPS to a consumer app's NPS — they measure different relationships under different expectations.

What JSM Natively Supports

JSM ships with a basic CSAT survey — thumbs up/down plus an optional comment field. It's low-friction and easy to enable, but limited:

• You can't change the scale (no 1–5 stars natively)
• You can't run NPS or CES from the native survey tool
• Reporting is basic — aggregate counts, no filtering by agent, ticket type, or date range in the native view

Atlassian shipped expanded native survey functionality in early 2026, adding multi-question surveys and additional scale types. It handles simple CSAT use cases well. NPS and CES still require a marketplace app or custom setup.

If you need NPS for quarterly sends, or CES for complex incident tickets, you're looking at a Forge app. Myra supports all three natively in JSM, writing responses back to Jira fields so they're queryable and available in dashboards.

A Note on Benchmarking Across Metrics

CSAT, NPS, and CES each have distinct scales and calculation methods. You cannot compare them directly. A CSAT of 4.2/5 and an NPS of +35 are both good scores for an IT service desk — but you can't say one is "higher" than the other. They're measuring different things.

What you can do: benchmark each metric against itself over time. An NPS trending from +20 to +40 over two quarters is meaningful progress. A CSAT trending from 3.6 to 4.1 is meaningful progress. Don't mix the signals.

The Actual Recommendation

Start with CSAT. It's the simplest to act on, the easiest to explain to agents, and the most immediately useful for a service desk. Enable it on every resolved ticket, give it 6–8 weeks to accumulate data, and use that data to identify your worst-performing ticket types and agents before trying to optimise anything else.

Add NPS quarterly if you manage an external customer relationship — real customers who could churn, not internal employees. Run it separately from the support flow; don't attach it to ticket resolutions.

Add CES when complaints about process complexity are recurring. If your CSAT scores are reasonable but you're still seeing escalations and reopens, CES will tell you where the friction is. Apply it selectively to tickets that exceeded your SLA or required three or more agent touches.

Running all three simultaneously from the start is usually too much. The data is only useful if someone is acting on it — and acting on three different feedback streams at once, before you've established a baseline, tends to produce analysis paralysis rather than improvement.

One metric, well-measured, is worth more than three metrics ignored.