How to calculate CSAT without reporting a number you can't defend
Learn how to calculate CSAT with the standard formula, worked examples, and guidance on scales, neutral responses, and sample size before you report a score.
The CSAT formula takes about ten seconds to apply. Deciding what counts as a satisfied response, which scale to run, how many responses you need, and whether the number moved for a real reason takes considerably longer. That second set of decisions is where most CSAT programs quietly go wrong.
The formula is the same one used across the industry: divide your positive responses by your total responses, then multiply by 100. If you want the arithmetic handled for you, the CSAT calculator does it directly, and what a CSAT score is and how it's interpreted covers the reading of the result. This guide covers the calculation decisions those pages assume you have already made.
Calculate your CSAT score in five steps
Work through these in order. Steps 2 and 3 are the ones that change your reported number, and they are the ones teams most often skip.
- Collect responses to a single satisfaction question. Use one clearly scoped question about one interaction, product, or touchpoint.
- Define which ratings count as positive. On a 1 to 5 scale, the standard convention is the top two boxes, meaning ratings of 4 and 5.
- Count your total valid responses. This is your denominator. Exclude skipped and partial responses, and hold that rule constant across periods.
- Apply the formula. Divide positive responses by total responses, then multiply by 100.
- Record the sample size alongside the score. A score without an n value cannot be interpreted or compared.
The formula written out:
CSAT = (number of positive responses ÷ total number of responses) × 100
A worked example: you send a CSAT survey after a support interaction and receive 100 valid responses. Sixty respondents rate their satisfaction a 4 and 25 rate it a 5. That gives 85 positive responses out of 100, so your CSAT is 85%. Report it as 85% (n=100), not 85%.
Why the denominator decision matters more than the formula
Two teams can run identical surveys and publish different scores purely through denominator choices. If one team excludes neutral responses from the total and the other includes them, the first team's score will be structurally higher, and neither team is calculating incorrectly. They are answering different questions.
Pick one rule and document it:
- Include neutrals in the denominator. This is the standard approach and produces the more conservative score. Recommended default.
- Exclude neutrals from the denominator. This reports satisfaction among customers who expressed an opinion, and will read several points higher.
- Count only the top box. This is a much stricter bar, useful when 4-out-of-5 ratings are so common that the top-two-box score sits above 90% and stops discriminating.
Whichever you choose, changing it later breaks your trend line. A score that jumps six points because someone redefined the denominator looks exactly like a score that jumped six points because service improved.
Choose the scale before you calculate, not after
Your scale determines what "positive" means, so it has to be settled before the first response arrives. CSAT surveys commonly run on 1 to 5, 1 to 7, or 1 to 10 scales.
- 1 to 5: The most common choice, and the one the top-two-box convention was built around. Positive means 4 and 5.
- 1 to 7: Gives respondents finer gradation. Positive typically means 6 and 7, though some teams include 5, which is a defensible choice you need to state explicitly.
- 1 to 10: Offers the most granularity but invites confusion with Net Promoter Score, which also uses 0 to 10 and a completely different calculation.
Converting between scales after the fact is the trap. There is no clean mathematical conversion from a 1 to 7 top-two-box percentage to a 1 to 5 equivalent, because the underlying response distributions differ. If you change scales, treat it as starting a new trend line and annotate the break in your reporting.
Write the question so the score means something
The calculation inherits every flaw in the question. A vague question produces a precise-looking number about nothing in particular.
Standard CSAT question wording asks respondents to rate satisfaction with one specific thing:
- "Overall, how satisfied were you with your support experience today?"
- "How satisfied are you with the product you received?"
- "How satisfied were you with the checkout process?"
Three wording rules protect the calculation:
- Scope the question to one interaction, not to the company as a whole. Whole-company satisfaction is a different measurement.
- Label every scale point rather than only the endpoints, so respondents interpret the middle consistently.
- Add one open-text follow-up that every respondent can answer regardless of rating, so you have the reason behind the score.
For fuller question banks, see customer satisfaction survey questions and question wording best practices.
Set up the calculation in SurveyMonkey
A few setup choices make the calculation repeatable rather than a monthly manual exercise.
Use the Matrix/Rating Scale question type for your CSAT question and enable the single-row rating scale option. This attaches weights to each answer option, which lets you filter and segment by specific ratings later instead of recounting by hand. If you use the Salesforce integration, available on Enterprise plans, those weights can pass through to Salesforce.
Beyond question setup, several platform capabilities apply directly to CSAT calculation and reporting:
- Crosstab reports compare responses across questions and audience segments, which is how you calculate CSAT by channel or region without exporting.
- Statistical significance testing tells you whether a change between two periods is meaningful or noise.
- Industry and global benchmarks put your score in context against industry averages rather than against your own history alone.
- Multi-survey analysis combines and compares results across surveys in one view, so a recurring CSAT program doesn't require manual data assembly.
- Email and notification automations alert teams when responses meet defined conditions, such as a low CSAT score arriving.
Calculate CSAT by segment without drowning in noise
Segmented CSAT is where most reporting falls apart. A 200-response monthly total looks healthy until it splits into eight agents, four regions, and three channels, at which point several cells hold four responses each and swing wildly month to month.
Guardrails worth setting:
- Establish a minimum cell size below which you report the count only, not a percentage.
- Use rolling windows rather than calendar months for low-volume segments, so each cell accumulates enough responses.
- Compare segments to each other within the same period rather than to their own prior period when volume is thin.
Mistakes that inflate a CSAT score
Most artificially high CSAT scores come from who was surveyed rather than from arithmetic.
- Surveying only resolved cases. If your trigger fires on ticket closure, customers who abandoned the interaction never enter the denominator.
- Sending too late. Satisfaction ratings drift as the memory of the interaction fades, and late sends skew toward customers motivated enough to still respond.
- Letting agents choose who gets surveyed. Any selection controlled by the person being measured will bias upward.
- Rounding away the sample size. Publishing 85% from 12 responses and 85% from 1,200 responses in the same dashboard treats them as equivalent evidence.
- Reading small movements as trends. Without significance testing, a two-point shift is usually noise.
Route responses automatically after the calculation
The score is the beginning of the work. Logic and automation turn a calculated number into a response.
Branching and skip logic let you adapt the survey path based on the rating given, so a customer who selects 1 or 2 receives a different follow-up question than one who selects 5. This produces the diagnostic detail that a bare percentage lacks. Notification automations can then alert the right team as soon as a low rating arrives, rather than at the end of a reporting cycle.
For turning the resulting themes into action, text analysis handles the open-text side, and how to improve CSAT covers what to do once you know where the score comes from.
Questions teams ask about calculating CSAT
- Is CSAT a percentage or an average?
- Should neutral responses count as positive?
- How is this different from calculating CES or NPS?
- Can I compare my CSAT to another company's?
Calculate CSAT you can act on with SurveyMonkey
A defensible CSAT score depends on decisions made before the arithmetic: the scale, the positive-response definition, the denominator rule, and the send trigger. Settle those once, document them, and the calculation becomes the easy part.
SurveyMonkey handles the calculation and the reporting around it, including segmentation through crosstab reports, significance testing on period-over-period changes, and industry benchmarks for context. Customer satisfaction programs run on that foundation, and analysis features handle the reporting once responses arrive.
Discover more resources
Solutions for your role
SurveyMonkey can help you do your job better. Discover how to make a bigger impact with winning strategies, products, experiences, and more.
Customer satisfaction metrics: what to track when your score says everything is fine
Compare every customer satisfaction metric, calculate CSAT, NPS, and CES with worked examples, and choose the right mix for the touchpoints your team owns.
CSAT survey examples: what to ask, when to ask it, and why length is the thing that breaks
Browse CSAT survey examples by touchpoint, see the wording and scales each one uses, and learn why the shortest surveys collect the best data.
CES vs CSAT: what each metric measures and when to use which
Compare CES vs CSAT to see what each metric measures, how each is calculated, which touchpoints suit each one, and what to do when the two disagree.