CSAT & Customer Effort Score Calculator
Customer satisfaction score from survey counts, pasted ratings or a simple satisfied-of-total figure (1–5, 1–7 or 1–10 scale, top-two-box, top-box or your own cut-off), customer effort score on 1–5, 1–7 or 1–10 scales, distribution tables and charts, margin of error and confidence interval, change versus last period, responses needed to reach a target, and benchmarks.
How it works
CSAT % = responses in the satisfied boxes (top two; top three on 1–10) ÷ all responses × 100; CSAT mean = Σ(score × count) ÷ n; top-box % = top score ÷ n; CES = Σ(score × count) ÷ n on the effort question; low-effort share = easy answers ÷ n; margin of error = z √(p(1−p)/n) with z = 1.645, 1.96 or 2.576; Wilson interval = (p + z²/2n ± z√(p(1−p)/n + z²/4n²)) ÷ (1 + z²/n); change test: z = (p₁ − p₂) ÷ √(p̄(1−p̄)(1/n₁ + 1/n₂)); extra satisfied responses for target t = (t·n − satisfied) ÷ (1 − t)
CSAT asks how satisfied a customer is with a specific interaction and is usually reported as the share choosing the top two boxes, with the mean as a secondary view — transactional, fast to move, and the standard for support and post-purchase surveys. Enter it as counts per score, paste the individual ratings from a survey export, or just type how many were satisfied out of the total. A strict top-box score and your own cut-off are available under More options, and the distribution table and chart show where answers cluster. Customer effort score asks how easy the company made it to get the job done; the CES 2.0 version uses a seven-point agreement scale, and older surveys use 1–5 (sometimes with 1 meaning the least effort). Gartner/CEB research found effort predicts repurchase and word of mouth better than satisfaction, because delighted customers stay loyal only slightly more than merely satisfied ones, while customers who had to struggle overwhelmingly defect. Read the two together: high satisfaction with high effort is a hidden churn risk, low satisfaction despite low effort points at the product rather than the process. Small samples and low response rates make both scores noisy, so the calculator reports the margin of error and a Wilson confidence interval, and with last period’s score and sample size it tests whether a change is real or likely noise.
Worked example
820 CSAT responses on a 1–5 scale (18, 32, 90, 260, 420) with 680 scoring 4 or 5 give 680 ÷ 820 = 82.9% satisfied, a mean of 4.26 and a 51.2% top-box score — good, ±2.6 points at 95% (interval 80.2% to 85.3%). 717 CES responses average 5.52 out of 7 with 80.9% low-effort and 9.3% high-effort: satisfaction and effort agree, and the 13.7% response rate is adequate. If last quarter scored 80% on 700 responses, the 2.9-point rise gives z = 1.47 — not significant at 95%. Reaching 85% would take 114 more satisfied responses with no new unhappy ones.
Frequently asked questions
CSAT vs NPS?
CSAT measures a transaction now; NPS measures relationship-level advocacy. Use CSAT/CES after interactions and NPS quarterly or biannually on the whole base.
Should neutrals count as satisfied?
No — the standard definition counts only the top two boxes. Some teams also report a “top box” (5 only) score for stretch targets; choose it under More options.
Is my CSAT change real or just noise?
Enter last period’s CSAT and its number of responses. The calculator runs a two-proportion z-test: if |z| is below 1.96 (at 95%) the change is within what random sampling alone would produce, so wait for more data before celebrating or reacting.
Which CES scale should I use?
The CES 2.0 question (“The company made it easy for me to handle my issue”, 1 = strongly disagree to 7 = strongly agree) is the most common today. Older surveys use 1–5, and the very first version scored 1 as the least effort. Pick the scale your survey used so the low- and high-effort groups are counted correctly.
What is a good CSAT and CES?
Around 75–85% satisfied is typical for many service businesses, with over 90% excellent. On the seven-point CES, 5.5 or more is strong; watch the share of 1–3 answers, because those customers are the most likely to leave. Compare against your own trend first and industry averages second.
Related calculators
Marketing analytics from the figures entered; attribution windows, platform definitions and sampling vary, so compare like with like over time rather than against generic benchmarks alone.