You've been collecting CSAT scores for a few weeks and your rolling average is sitting at 3.8 out of 5. Is that good? Concerning? The industry average? Most JSM admins starting out with feedback collection hit this question and struggle to find a straight answer.
Here's a straight answer, with the caveats that make it actually useful.
The Industry Numbers
The most widely cited benchmarks for IT service desk CSAT come from HDI (Help Desk Institute) and similar analyst groups. The numbers vary by study and year, but the consistent picture looks roughly like this:
| Performance Level | CSAT Score (1–5 scale) | Equivalent % |
|---|---|---|
| Top quartile | 4.5+ | 90%+ |
| Above average | 4.0–4.4 | 80–88% |
| Average | 3.5–3.9 | 70–78% |
| Below average | 3.0–3.4 | 60–68% |
| Needs attention | Below 3.0 | Below 60% |
The "% satisfied" framing (counting responses of 4 or 5 as satisfied, and dividing by total responses) is sometimes more intuitive than the average score. A team with 85% satisfied customers has a clearer picture than "our average is 4.2."
Caveat 1: These benchmarks are mostly from larger organisations with dedicated service desk teams. If you're running IT support for a 50-person company where everyone knows each other, your scores will likely be higher — the relationship context changes expectations.
Caveat 2: B2C support benchmarks (retail, telecoms, banking) run higher than IT/internal service desk benchmarks. Don't benchmark yourself against consumer support scores unless you're running external customer support.
The JSM-Specific Context
A few factors that tend to shape CSAT scores in Jira Service Management environments:
Ticket complexity skews scores down. IT service desks handle a mix of simple requests (password resets, software installs) and complex incidents (outages, data issues). Complex tickets typically get lower satisfaction scores even when handled well — the customer is frustrated by the situation, and that frustration attaches to the rating.
Teams that separate their CSAT reporting by ticket type often find that password reset tickets score 4.7 while incident management tickets score 3.9 — for the same team. Averaging these together understates performance on simple requests and overstates it on complex ones.
First contact resolution is the strongest predictor. Across most IT service desk studies, whether the issue was resolved on the first contact (no escalation, no reopens) predicts CSAT more reliably than response time or resolution time. A ticket that takes 4 hours to resolve but never needs a follow-up typically scores better than one resolved in 30 minutes that gets reopened twice.
Response rate affects the average. If only your most unhappy and most delighted customers respond, your average will be polarised. A low response rate (under 20%) means your average is unreliable. Before comparing your score to benchmarks, check that your response rate is high enough to be representative.
What to Measure Instead of (or Alongside) the Average
The single average CSAT score is a headline number — useful for tracking trends, not useful for diagnosis. More actionable metrics:
CSAT by assignee. Which agents are consistently scoring below the team average? That's a coaching conversation, not a blame conversation — but you need the data to have it.
CSAT by ticket category. Are hardware requests consistently worse than software requests? Is network troubleshooting your weakest area? Category-level data points to training or process gaps.
CSAT trend over time. Is your score improving month-over-month? A 3.7 that's been 3.7 for six months is a different situation from a 3.7 that was 4.2 three months ago. The trajectory matters.
Low-score comment themes. The comments on 1 and 2-star ratings are your most valuable qualitative data. Tag them — "slow response," "fix didn't work," "unclear communication," etc. — and you'll start seeing patterns within a few weeks.
A Realistic Target-Setting Framework
Rather than chasing an industry average, most service desk teams are better served by setting improvement targets relative to their own baseline:
- Establish your baseline — collect at least 4 weeks of data before drawing conclusions
- Set a +0.2 quarterly improvement target — going from 3.8 to 4.0 in 90 days is achievable; going from 3.8 to 4.5 in one quarter is probably not
- Identify one lever — pick the single biggest driver of low scores (response time, first-contact resolution, specific ticket type) and focus the improvement effort there
- Reassess after each quarter — as you improve on one lever, the next bottleneck becomes visible
This approach is more reliable than benchmarking against external numbers from organisations with different team sizes, ticket volumes, and customer expectations.
The Score That Actually Matters
The best CSAT benchmark is your own historical data. An IT service desk that was at 3.5 twelve months ago and is now at 4.1 has improved more than a team that's maintained 4.3 for two years without understanding why.
The point of measuring is to know whether you're getting better. The number itself is secondary.
Myra collects CSAT, NPS, and CES inside your JSM portal and surfaces the breakdowns — by agent, by ticket type, by trend — that make benchmarks actually useful. Try it free on the Atlassian Marketplace.