Setting up CSAT in Salesforce takes a few steps: configure a survey tool, build a Flow to trigger sends, store responses on the Case, and build dashboards. The data pipeline is straightforward; the operational discipline of acting on results is harder.
- Pick a survey tool
Salesforce Surveys (native, lower cost) or a third-party tool (Qualtrics, SurveyMonkey, Medallia, Delighted). Pick based on existing tooling, feature requirements, and budget.
- Design the CSAT survey
One question on a clear scale. Resist the temptation to add five more questions; longer surveys kill response rates. The single question should map directly to CSAT (How would you rate your experience).
- Build the trigger Flow
Create a record-triggered Flow on Case that fires when Status changes to Closed. The Flow sends the survey through the chosen tool. Add a delay of 1 to 24 hours after close to let the customer reflect.
- Store responses on the Case
Configure the survey tool to write back to a Case custom field (CSAT_Rating__c) and the survey response object. This is what dashboards aggregate.
- Build CSAT dashboards
Service Cloud Service Analytics has pre-built CSAT dashboards. Customize for your team. At minimum: overall CSAT, CSAT by agent, CSAT by Case Reason, CSAT time series.
- Operationalize the responses
Build a Flow that flags low-CSAT cases for manager review. Train managers to follow up with both the agent and the customer. CSAT data is only useful if someone acts on it.
- Response rates of 5 to 20 percent are normal. Self-selection bias is real; do not draw conclusions from small samples.
- Mid-program scale changes (1-5 to 1-10, or text to emoji) break time-series. Pick a scale up front and stick with it.
- Tying CSAT to compensation without normalizing for case complexity is unfair to agents handling harder cases.
- Survey fatigue affects response rates. If a customer files 10 cases in a month and gets 10 surveys, they ignore all of them. Cap survey frequency per customer.
- CSAT data has a delay (24 hours to weeks). Real-time dashboards on CSAT lag the source events; account for the lag when interpreting trends.