### Answer Based on community reports, this "Error in parsing JSON for call analysis" issue has been experienced by other users [(1) ]() [(2) ]() . One user identified that the root cause was Unicode smart quotes (U+201C, U+201D, U+2019) in the transcript that broke JSON parsing in post-call analysis [(2) ]() . They resolved it by adjusting the prompting for the post-call analysis . To troubleshoot your specific case, check the detailed call logs for [(1) ]() : - Any error messages during the analysis phase - The exact time when the analysis process fails - Whether the issue happens every time or only sometimes (consistently or intermittently) If you're seeing blank summaries in call history with this error, you may need to modify your post-call analysis prompting to handle formatting issues, and consider sanitizing any special characters that could break JSON parsing [(2) ]() .

​ To learn more, see the sources I considered: [[1] Post-call analysis "Error in parsing openai JSON forn call anaylisis"](https://discord.com/channels/1186424376300679209/1395844709015883847) [[2] call history error in parsing json](https://discord.com/channels/1186424376300679209/1449983415532257280) ​

Feel free to tag @AI Evy with additional questions. I also appreciate feedback, it helps me improve.

### Answer Adding an 'address' field to your post-call analysis could definitely trigger JSON parsing errors, especially if addresses contain special characters, quotes, or unexpected formatting that breaks JSON structure [(1) ]() . Based on similar cases, one user resolved this issue by switching from Gemini 3 Flash to GPT-4.1 or GPT-4.1 mini for post-call analysis [(2) ]() . This suggests the LLM model choice affects JSON parsing reliability. To troubleshoot: 1. Check your post-call analysis configuration - Review how the 'address' field is defined in your

post_call_analysis_data

array [(3) ]() . Ensure the description and examples are clear about expected format. 2. Review the model setting - Check your

post_call_analysis_model

configuration . Consider switching to

gpt-4.1-mini

or

gpt-4.1

if you're using a different model [(2) ]() . 3. Examine call logs - Look at the detailed logs for calls with this error to identify: - Error messages during the analysis phase - The exact time when analysis fails - Whether it happens consistently or intermittently [(4) ]() 4. Add conditional prompting - Use the

conditional_prompt

field in your address extraction configuration to guide the LLM on handling edge cases like multi-line addresses or special characters [(3) ]() . The timing correlation with deploying a new agent suggests the issue is in the agent's post-call analysis configuration rather than a platform-wide problem.

​ To learn more, see the sources I considered: [[1] Post Call Analysis Overview](https://docs.retellai.com/features/post-call-analysis-overview) [[2] Post-call extraction error](https://discord.com/channels/1186424376300679209/1465667130187059382) [[3] Get Agent Versions](https://docs.retellai.com/api-references/get-agent-versions) [[4] Post-call analysis "Error in parsing openai JSON forn call anaylisis"](https://discord.com/channels/1186424376300679209/1395844709015883847) ​

Feel free to tag @AI Evy with additional questions. I also appreciate feedback, it helps me improve.