Measuring CSAT, DSAT, and SLA in Customer Support: What to Track and How
How to calculate CSAT and DSAT, how to define an SLA, and how to read the three together to see where support is strong and where it isn't, with examples and a checklist.
In this article
Average customer satisfaction is 4.4 out of 5. Is that good? Maybe, but the number says nothing about the 12% who were dissatisfied, or whether they were the same people who waited four hours. To see the real picture, read three metrics together: CSAT, DSAT, and SLA.
If you need the basic definitions, start with What Is CSAT? and SLA in customer support. Here we focus on practical use.
CSAT: how many people were satisfied
After each conversation, ask one simple question and let the customer answer on a scale of 1 to 5:
CSAT = (number of 4 and 5 ratings ÷ total ratings) × 100
If 82 out of 100 ratings were a 4 or a 5, your CSAT is 82%.
DSAT: how many people were dissatisfied, and why
DSAT stands for dissatisfaction and shows the share of unhappy customers, usually ratings of 1 and 2:
DSAT = (number of 1 and 2 ratings ÷ total ratings) × 100
Why read it separately? Because a high CSAT can hide a group of seriously unhappy customers. Good teams treat DSAT like a to-do list: every low rating is a conversation that needs to be read. The follow-up question ("What could have been better?") after a low rating is the most valuable part of the survey. In Hodhod you can ask custom follow-up questions after each conversation.
SLA: what we commit to
An SLA is what you promise, not what usually happens. Three common deadlines:
- First response time: how soon do we give the first human reply?
- Next response time: after each new customer message, how soon do we reply?
- Resolution time: how soon is the problem fully solved?
A practical tip: set a realistic deadline for each channel and each priority, and if your team isn't available 24/7, count the deadlines during working hours only. Hodhod's SLA policy has this option ("only during working hours").
How to read the three metrics together
| Pattern | Likely meaning | Action |
|---|---|---|
| High CSAT, low SLA | Customers like the quality but get replies late | Review capacity or escalation |
| Low CSAT, high SLA | You reply fast but the answers don't help | Fix training and canned responses |
| Both low | A structural problem | Start with the recurring root causes (issues) |
| DSAT suddenly jumps | A new incident or issue | Read the most recent conversations |
Go beyond the average
The overall average hides a lot. Slice the same numbers:
- By agent and team: for coaching, not punishment.
- By channel: chat, Telegram, tickets.
- By topic or issue: which problems create the most dissatisfaction? In Hodhod, for each issue (and each occurrence) you can see customer satisfaction, SLA, first response time, and resolution time.
- By time: satisfaction during busy and quiet hours.
Beyond CSAT: reviewing conversation quality
CSAT only shows the opinion of customers who answered. To look at the conversation itself, supervisors score a sample of conversations against the team's criteria and pass the feedback on to the agent; this review complements the satisfaction survey, it doesn't replace it. In Hodhod you build a scorecard with weighted criteria, make sampling automatic (a percentage you choose, always including low-rated conversations), and agents see their own reviews and can acknowledge or dispute them. Full guide: support conversation QA review.
Common mistakes
- A long survey. One rating and one follow-up question are enough.
- Looking only at the average. Look at DSAT and the distribution of ratings too.
- An unrealistic SLA. A deadline that is always breached loses its meaning.
- Measuring without acting. Make one specific decision from the numbers every week.
Setup checklist
- Enable the satisfaction survey after conversations and add one follow-up question.
- Define the three SLA deadlines for different priorities and set the working hours.
- Open the report weekly: CSAT, DSAT, SLA compliance.
- Read every low rating and categorize its cause.
- Turn recurring causes into issues and hand their fix to the relevant team; read the issues' MTTR and recurrence rate next to these numbers.
- For tickets that pass their deadline, set up auto-escalation so their priority rises or someone more senior is told.
- Review a sample of conversations with a scorecard every month.
The Hodhod ticketing system and issue tracking make this workflow possible in the same chat panel.
Frequently asked questions
What is a good CSAT?
It depends on the industry, but in practice your own trend matters more than the absolute number: are you getting better or worse each month?
How do we lower DSAT?
Read the dissatisfied conversations, categorize their causes (slow replies, wrong answers, a recurring problem), and fix the biggest category first.
Should we use the same SLA for all channels?
Preferably not. Chat carries an expectation of faster replies; tickets and email allow longer deadlines. Define a separate deadline for each.
