CSAT and DSAT are two readings of the same customer service survey. CSAT (Customer Satisfaction Score) is the percentage of responses in the positive range: (positive responses / total responses) x 100. DSAT (Dissatisfaction Score) is the percentage in the negative range: (dissatisfied responses / total responses) x 100. The critical point most explanations get wrong is that they are not opposites and do not sum to 100%, because neutral responses are counted in the denominator of both but the numerator of neither, and customers who never answered the survey are missing from both. CSAT tells you how well the majority experience is going. DSAT tells you how big and how concentrated the failure tail is.
In short
- Both metrics are calculated from the same survey responses, using the same denominator, with different numerators.
- CSAT counts the top ratings, DSAT counts the bottom ratings, and neutral ratings belong to neither, so the two figures leave a gap.
- Higher is better for CSAT. Lower is better for DSAT. Reporting them side by side means one of your two numbers is always read backwards, which is a common source of confusion in reviews.
- Changing what counts as dissatisfied moves DSAT sharply while leaving CSAT untouched, so the two are not mirror images even mechanically.
- CSAT is the health metric for the bulk of your customers. DSAT is the risk metric for the tail that escalates, churns and generates repeat contacts.
- Neither metric explains itself. Both record what the customer felt, and neither records what happened in the conversation to cause it.
CSAT vs DSAT at a glance
Because the two metrics come from one survey, the differences are easy to blur. This table separates them across the dimensions that actually change how you should read the numbers.
| Dimension | CSAT | DSAT |
|---|---|---|
| What it measures | Share of respondents who were satisfied | Share of respondents who were dissatisfied |
| Formula | (positive responses / total responses) x 100 | (dissatisfied responses / total responses) x 100 |
| Typical numerator on a 1 to 5 scale | Ratings of 4 and 5 | Ratings of 1 and 2 |
| Treatment of neutral ratings | Excluded from the numerator, kept in the denominator | Excluded from the numerator, kept in the denominator |
| Direction | Higher is better | Lower is better |
| What it is good for | Tracking overall service health and the majority experience | Sizing the failure tail and finding where it concentrates |
| What it hides | A growing group of angry customers, absorbed by the average | Everything positive, including whether the middle is drifting up or down |
| Sensitivity to threshold choice | Low, the top two ratings are a near-universal convention | High, including or excluding the neutral rating changes it dramatically |
One worked example: the same survey, two numbers
This is the part worth reading slowly, because seeing both metrics come out of one dataset settles the confusion permanently.
A support team sends a post-interaction survey to 1,000 customers over a month. It uses a 1 to 5 scale. 200 customers respond, distributed like this:
- Rated 5: 82 responses
- Rated 4: 48 responses
- Rated 3: 30 responses
- Rated 2: 24 responses
- Rated 1: 16 responses
Calculating CSAT
Positive responses are the 4s and 5s: 48 + 82 = 130. Divide by the 200 total responses and multiply by 100. CSAT = 65%.
Calculating DSAT
Dissatisfied responses are the 1s and 2s: 16 + 24 = 40. Divide by the same 200 total responses and multiply by 100. DSAT = 20%.
Why 65% and 20% do not add up to 100%
Because 30 customers rated the interaction a 3. That is 15% of respondents who belong to neither camp. They are in the denominator of both calculations and the numerator of neither. So the arithmetic is 65% satisfied, 20% dissatisfied, 15% neutral, and only together do the three reach 100%.
If you had assumed DSAT was simply 100 minus CSAT, you would have reported 35% dissatisfied. That is nearly double the real figure, and it would have put 30 indifferent customers into the same bucket as 16 people who rated the experience the worst possible score. Those two groups need completely different responses.
The 800 customers who never replied
There is a second, larger gap. The survey went to 1,000 customers and 200 answered. Both metrics describe only those 200. Survey response rates are typically low, and the customers who do respond are not a random sample: strong reactions in either direction are more likely to produce a reply than mild ones. So the 800 silent customers are not neutral, not satisfied and not dissatisfied. They are simply unmeasured, and the dissatisfaction among them is invisible in both numbers.
Now change the threshold
Suppose the team decides a rating of 3 after a support contact is not an acceptable outcome and should count as dissatisfied. Nothing about the customers changes. DSAT becomes (16 + 24 + 30) / 200 = 35%. CSAT stays exactly at 65%. One metric nearly doubled and the other did not move, on identical data. That asymmetry is the clearest proof that DSAT is a separate metric rather than the inverse of CSAT, and it is why any DSAT figure you compare against needs its threshold stated alongside it.
When to lead with CSAT and when to lead with DSAT
They answer different management questions, so the useful framing is not which one to adopt but which one should lead a given decision.
Lead with CSAT when
You are reporting on overall service health, comparing periods, or tracking whether a change improved the typical experience. CSAT moves with the bulk of your volume, which makes it the better headline number and the better metric to hold a broad target against.
Lead with DSAT when
You are managing risk, prioritizing fixes, or working to a contractual quality threshold. Dissatisfied customers drive escalations, repeat contacts, negative reviews and churn, so the cost of a support operation concentrates in that tail. DSAT is also more sensitive: because it starts from a small base, a genuine problem in one queue shows up as a sharp move rather than a rounding error in an average.
Use both when the two disagree
The most informative situation is when CSAT holds flat and DSAT rises. That combination means the middle of your distribution is compensating for a growing group of genuinely unhappy customers, and it is invisible if you only report the average. The reverse, DSAT flat while CSAT falls, usually means satisfied customers are drifting toward neutral rather than anything breaking outright, which points at a very different fix.
Where the other CX metrics fit
CSAT and DSAT are both transactional: they judge one interaction. They sit alongside relationship and effort metrics rather than competing with them. NPS measures long-term loyalty to the brand, and CES measures how hard the customer had to work to get resolved, which is often the best early predictor of a dissatisfied rating. The full comparison of the three is in CSAT vs NPS vs CES.
The limitation both metrics share
CSAT and DSAT disagree about which customers to count, but they agree on something more important: neither records what happened. A rating is a verdict with no reasoning attached. A 1 could mean the agent was rude, the answer was wrong, the policy was unwelcome, the wait was long, or the customer was already angry when they arrived. The score is identical in every case and the correct response is different in every case.
The reasoning is in the conversation. Every rating, positive or negative, has a transcript sitting behind it that contains the actual sequence of events. Scoring those conversations against your own standards for accuracy, process and tone gives you the explanatory layer the survey cannot provide, and it covers the interactions that never produced a survey response at all.
That is the gap Kaizo works in. It scores the conversation itself rather than the rating, and traces every score back to the specific evidence in the transcript, so a DSAT movement can be attributed to a handling behavior instead of argued about in a review. It runs natively inside Zendesk and Salesforce, so the score and the conversation stay in one place. The method for turning that into a reduction is in how to reduce DSAT.
Frequently asked questions
What is the difference between CSAT and DSAT?
CSAT is the percentage of survey respondents who rated an interaction positively, and DSAT is the percentage who rated it negatively. Both use the same denominator of total responses received. CSAT describes the majority experience and higher is better, while DSAT sizes the failure tail and lower is better. They are two readings of the same survey rather than two different surveys.
Do CSAT and DSAT add up to 100%?
No, and assuming they do is the most common error with these metrics. Neutral ratings, such as a 3 on a 1 to 5 scale, are counted in the denominator of both calculations but the numerator of neither, so they leave a gap between the two figures. Separately, both are computed only on customers who responded, so anyone who ignored the survey is in neither number.
How do you calculate CSAT and DSAT from the same survey?
Take your total responses as the denominator for both. For CSAT, count responses in the positive range, typically 4 and 5 on a 1 to 5 scale, divide, and multiply by 100. For DSAT, count responses in the negative range, typically 1 and 2, divide by the same total, and multiply by 100. With 200 responses split 82 fives, 48 fours, 30 threes, 24 twos and 16 ones, CSAT is 65% and DSAT is 20%, with the remaining 15% neutral.
Why do BPOs report DSAT instead of CSAT?
Most report both, but outsourced contact centers give DSAT more weight because quality is contractual. A service level agreement often sets a maximum tolerable dissatisfaction rate, not just a minimum satisfaction score, since the client cares about the floor of the experience as much as the average. That makes each dissatisfied response something to be reviewed and categorized individually rather than a percentage to watch.
Can CSAT and DSAT both go up at the same time?
Yes, and it is more common than people expect. If neutral ratings shift outward toward both ends of the scale, the satisfied share and the dissatisfied share can grow together while the neutral middle shrinks. It usually signals that experience quality has become more variable, which points at inconsistency in handling rather than a uniform decline.
Which metric should a support team target?
Target CSAT for overall service health and DSAT for risk. Reporting only CSAT lets a growing group of angry customers hide inside a stable average, and reporting only DSAT tells you nothing about whether the majority experience is improving. The most informative signal is when the two disagree, because that is when the average is concealing a change in the distribution.
Related terms
Find out what your dissatisfied ratings actually say
CSAT and DSAT tell you how customers felt. The conversation tells you what happened. Bring a week of your real tickets and we will show you the quality signal behind both numbers, with every score traced back to the evidence in the transcript.
