CSS: Customer Satisfaction Score

A CSS score (Customer Satisfaction Score) shows how satisfied customers are with a service interaction, such as a support ticket, a call or a visit. It usually comes from one question asked right after the interaction ("How satisfied were you with the service you received?") and is reported as the percentage of satisfied customers or as an average rating.

CSS and CSAT are two abbreviations for the same idea. CSS is the name often used in customer service, contact center and service desk reports, where the score is tracked per interaction, agent and channel. In web development, CSS means Cascading Style Sheets, which is unrelated.

How to calculate a CSS score

There are two common methods. Say which one you use, because they give different numbers.

1. Percentage of satisfied customers (the same as CSAT):

CSS = responses of 4 or 5 ÷ all responses × 100% (on a 1–5 scale)

2. Average rating:

CSS = sum of all ratings ÷ number of responses

Example: 100 customers rated their support experience on a 1–5 scale.

• Percentage method: (15 + 25) ÷ 100 × 100% = 40% satisfied.
• Average method: (5×15 + 4×25 + 3×30 + 2×20 + 1×10) ÷ 100 = 315 ÷ 100 = 3.15 out of 5.

Some service teams use a 3-point scale (Good, Okay, Bad) or thumbs up and down. Then CSS is simply the share of positive answers. The CSAT calculator works for CSS too.

CSS vs CSAT, NPS and CES

Where CSS is used in customer service

• After a ticket is closed, by email or in the help desk widget.
• After a call, as an automated phone survey or an SMS.
• At the end of a chat, inside the chat window.
• After a field service visit, such as a repair or installation.

Teams look at CSS by agent, team, channel and issue type, and next to operational metrics such as first response time, time to resolution and first contact resolution. That shows which delays and handoffs actually cost satisfaction.

How to interpret a CSS score

• Compare like with like: the same question, scale and method across months, channels or teams. A percentage CSS and an average CSS can't be compared.
• Look at the low ratings, not only the score. The share of 1s and 2s and their comments show what to fix.
• Watch the response rate. If only a small share of customers answer, the score may over-represent people with strong experiences in either direction.
• Use enough responses before ranking agents. A handful of ratings per agent mostly reflects luck.

Common mistakes

• Sending the survey days later. Customers no longer remember the details of the interaction.
• Mixing scales (1–5 in email, thumbs in chat) and adding them into one number.
• Using CSS as the only agent KPI. Agents start asking for good ratings, and the score stops measuring service.
• Never following up with customers who gave a low rating. A quick call often saves the customer and explains the problem.

How to run a CSS survey

1. Ask one rating question about the interaction: "How satisfied were you with the support you received today?" with labels from "Very dissatisfied" to "Very satisfied".
2. Add an open question for the reason: "What could we have done better?"
3. Optionally add an effort question (CES) or "Was your issue resolved?".
4. Pass the ticket ID, agent and channel into the survey link as hidden variables (from the Basic plan) so each answer is linked to its interaction.
5. Send it right after the ticket closes, and review low scores every week.

In SurveyNinja, a Smile rating question can show a follow-up question only after the faces you choose, for example "What went wrong?" after the two saddest faces.