Implementing CSAT surveys via post-call Architect flows

General Discussion
1

For post-call CSAT surveys, the simplest approach is an Architect disconnect flow that plays a prompt and captures DTMF.

# Query survey results via analytics
body = {'interval': '2024-01-01/2024-01-31',
        'metrics': ['nSurveysSent','nSurveysStarted','nSurveysCompleted'],
        'groupBy': ['queueId']}
result = api.post_analytics_surveys_aggregates_query(body)
2

For digital channels, deploy a web-based survey instead of DTMF.

After a chat ends, send the customer a survey link via the Web Messaging widget. The digital survey can include open-text fields that provide richer feedback than a 1-5 numeric scale. Route the survey responses to a dedicated analytics queue for processing.

3

The disconnect flow survey must be SHORT. Three questions maximum.

I maintain 50+ flows and have tested survey lengths extensively. At 3 questions, completion rate is 45%. At 5 questions, it drops to 18%. At 7 questions, it is under 5%. Keep it to: ‘Rate your experience 1-5’, ‘Was your issue resolved? 1=yes 2=no’, and optionally ‘Would you recommend us? 1-5’.

4

From a statistical validity perspective, the DTMF survey introduces a self-selection bias.

Only customers who feel strongly (very satisfied or very dissatisfied) stay on the line. The moderate majority hangs up. Your CSAT scores will appear bimodal. To correct for this, apply a response bias adjustment factor based on your overall completion rate.