You just resolved a customer's issue. They seem happy. You want to measure that happiness. But which survey do you send?
CSAT (Customer Satisfaction Score) and NPS (Net Promoter Score) are the two most common support metrics. They look similar. Both are short surveys. Both produce a number you can track over time. But they measure fundamentally different things, and sending the wrong one at the wrong time gives you data that misleads rather than informs.
Here is the difference, when to use each, and how to set them up.
What CSAT Measures
CSAT measures satisfaction with a specific interaction. After a support chat ends, you ask: "How satisfied were you with this interaction?" The customer picks a rating, typically 1-5 stars or a smiley face scale.
The score is calculated as: (number of positive responses / total responses) x 100.
If 80 out of 100 customers rate their experience 4 or 5 stars, your CSAT is 80%.
What CSAT tells you:
- Was this particular conversation helpful?
- Did the agent (or AI) resolve the issue?
- Is support quality consistent across your team?
What CSAT does not tell you:
- Will this customer buy again?
- Will they recommend you?
- How they feel about your product overall?
CSAT is a snapshot. It captures a moment, not a relationship.
CSAT Benchmarks
According to the American Customer Satisfaction Index (ACSI, 2024), the average CSAT across industries is 74.4%. Here is how to read your score:
- Below 60%. Serious problems. Customers are leaving interactions frustrated. Audit your most recent conversations to find patterns.
- 60-74%. Below average. There is room for significant improvement in response quality, speed, or accuracy.
- 75-84%. Good. Your support is meeting expectations for most customers.
- 85-94%. Very good. You are above industry average. Focus on maintaining consistency.
- 95%+. Exceptional. Be aware that very high scores sometimes indicate survey fatigue (only happy customers bother responding).
What NPS Measures
NPS measures loyalty and likelihood to recommend. The question is: "On a scale of 0-10, how likely are you to recommend [company] to a friend or colleague?"
Responses are grouped into three categories:
- Promoters (9-10). Enthusiastic customers who actively recommend you.
- Passives (7-8). Satisfied but not enthusiastic. They will not go out of their way to recommend or criticize.
- Detractors (0-6). Unhappy customers who may discourage others from using your product.
The NPS formula: % Promoters minus % Detractors. The result ranges from -100 to +100.
If 60% of respondents are promoters and 20% are detractors, your NPS is 40.
What NPS tells you:
- Overall brand loyalty
- How customers feel about your company as a whole
- Likelihood of organic growth through referrals
What NPS does not tell you:
- Whether a specific support interaction was good
- What to fix in your support process
- Why a customer is or is not loyal
NPS is a relationship metric. It captures the big picture.
NPS Benchmarks
Benchmarks vary widely by industry. According to Bain & Company (the creators of NPS) and Retently's benchmark data (2024):
- Below 0. More detractors than promoters. Your product or service has fundamental issues.
- 0-30. Average for most B2B SaaS companies. Acceptable but not remarkable.
- 30-50. Good. You have more promoters than detractors and a healthy referral potential.
- 50-70. Excellent. Strong brand loyalty. Common among top-performing SaaS companies.
- 70+. Exceptional. Rare and typically seen in companies with cult-like followings.
When to Use CSAT (After Every Chat)
Send a CSAT survey immediately after a support conversation is resolved. The timing matters because you are measuring the experience of that specific interaction.
Best practices:
- Trigger it automatically. When a conversation is marked as resolved, the survey appears in the same chat window. No separate email. No extra click.
- Keep it simple. One question. A 1-5 star rating or three emoji faces (happy, neutral, sad). Optional comment field for those who want to elaborate.
- Do not survey every channel separately. If a customer had an AI conversation and then a human follow-up, send one survey at the end, not two.
When NOT to send CSAT:
- After automated messages (welcome messages, business hours notices)
- When the conversation was not actually resolved (customer abandoned the chat)
- Multiple times in the same day to the same customer
CSAT is your operational metric. Track it weekly. Compare agents. Compare AI vs. human resolution satisfaction. Use it to identify training needs and knowledge base gaps.
When to Use NPS (Monthly or Quarterly)
NPS is a strategic metric. It reflects the full customer experience, not just support. Sending it after a support chat biases the results: a customer who just had a bad experience will rate your entire company low based on one interaction.
Best practices:
- Send it on a schedule. Monthly or quarterly, depending on your customer base size. Monthly works if you have 500+ customers. Quarterly is better for smaller bases to avoid survey fatigue.
- Send via email. NPS works better as a standalone survey outside the support context. The customer evaluates your company broadly, not a specific interaction.
- Include a follow-up question. After the 0-10 rating, ask "What is the primary reason for your score?" The qualitative feedback is often more valuable than the number itself.
- Segment results. Break NPS down by customer tier, plan, tenure, and usage level. A new customer's NPS tells a different story than a 2-year veteran's.
When NOT to send NPS:
- Immediately after a support interaction (that is what CSAT is for)
- More than once per month (survey fatigue lowers response rates)
- To users who have not been active in the last 30 days (their score reflects inactivity, not your product)
How to Set Up Both
Most support tools include surveys as a built-in feature or an add-on. Here is a practical setup.
CSAT Setup
- Enable post-conversation surveys in your support tool. With Helpable, CSAT surveys are included on all plans and trigger automatically when a conversation is resolved.
- Choose a rating scale. The 5-star scale is the industry standard for CSAT.
- Add an optional comment field. Some customers will explain their rating. This feedback is gold for improving your process.
- Set up a dashboard or report that tracks CSAT weekly. Look for trends, not individual scores.
NPS Setup
- Schedule a recurring NPS survey via email. Monthly for 500+ customers, quarterly for smaller bases.
- Use the standard NPS question: "How likely are you to recommend [company] to a friend or colleague?" Scale: 0-10.
- Add a follow-up open-text question.
- Track NPS quarterly at minimum. Compare to your previous quarter and to industry benchmarks.
Helpable includes both NPS and CSAT surveys on all plans. CSAT triggers after resolved conversations. NPS can be sent as a scheduled campaign. Results are exportable as CSV for deeper analysis.
Using Both Together
The real insight comes from combining CSAT and NPS data. Here are three patterns to watch:
High CSAT, low NPS. Your support team is great, but customers are unhappy with the product itself. The problem is not support; it is the core experience.
Low CSAT, high NPS. Customers love your product but find support frustrating. Invest in faster response times, better documentation, or AI chatbot improvements.
Both declining. Systemic problem. Start with CSAT (it is easier to fix support operations) and investigate NPS drivers through follow-up interviews.
Both improving. Keep doing what you are doing. Document what changed so you can replicate it.
FAQ
Can I use just one and skip the other?
If you have to choose one, start with CSAT. It is directly actionable (you can improve specific interactions) and gives faster feedback loops. Add NPS when your customer base exceeds 200 and you want to track long-term loyalty.
What response rate should I expect for CSAT surveys?
Post-chat CSAT surveys typically get 15-25% response rates (Zendesk Benchmark, 2024). Higher if the survey is embedded in the chat widget rather than sent as a follow-up email. Lower if the survey requires opening a separate link.
Do customers actually fill out NPS surveys?
Email NPS surveys get 10-20% response rates on average. You can improve this by keeping the email short (one question visible without scrolling), using a recognizable sender name, and sending at a consistent time (Tuesday-Thursday mornings perform best according to SurveyMonkey data, 2024).
Should I follow up with detractors?
Yes, selectively. Customers who score 0-3 on NPS and leave a comment are telling you something important. A personal follow-up email ("I saw your feedback and want to understand how we can improve") can convert detractors into promoters. It also prevents churn.
Is CSAT affected by the AI chatbot vs. human agent?
Yes. AI-resolved conversations often receive slightly lower CSAT than human-resolved ones, typically 5-10 points lower. This is normal. Customers value the speed of AI but prefer the empathy of a human for complex issues. Track AI CSAT and human CSAT separately to understand the difference.