### Answer Variable extraction issues with emails and names can stem from the LLM misinterpreting the format or context. Based on community discussions, adjusting your description for extraction fields helps the LLM better understand the required output format. [(1) ]() For post-call analysis data extraction, you can improve accuracy by providing clear descriptions and examples for text fields. [(2) ]() When configuring text analysis, include format examples like "Customer called about billing issue" to guide the extraction. If you're extracting variables during the call (using Extract Dynamic Variable nodes in Conversation Flow) , the issue may relate to the LLM version being used, as some versions have known issues with conversation flow logic. [(1) ]() To improve email extraction specifically: - Adjust your field description to specify the exact format expected - Consider which LLM model you're using (GPT-4.1, GPT-4.1-mini, etc.) [(3) ]() [(4) ]() - Ensure your

post_call_analysis_data

configuration includes clear descriptions and examples [(2) ]() [(3) ]() The

post_call_analysis_model

can be set to different options including

gpt-4.1

,

gpt-4.1-mini

,

gpt-4.1-nano

, and others, which may affect extraction accuracy. [(3) ]() [(4) ]() Check out these potentially useful sources: [Define the information you want to extract](

lai.com/features/post-call-analysis-create) , [Variable Extraction Error]()

​ To learn more, see the sources I considered: [[1] Variable Extraction Error](https://discord.com/channels/1186424376300679209/1471904365907542211) [[2] Define the information you want to extract](https://docs.retellai.com/features/post-call-analysis-create) [[3] List Voice Agents](https://docs.retellai.com/api-references/list-agents) [[4] Update Voice Agent](https://docs.retellai.com/api-references/update-agent) ​

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