← all ideas

AI Cold Email Fundraising Fail

The idea

The idea was to use AI to send personalized cold emails to alumni of a fraternity to solicit donations. The plan was to scrape their LinkedIn profiles to gather information and write custom emails. However, the AI system made numerous mistakes, including incorrect information and awkward formatting, which led to confusion and anger from the recipients. Additionally, the lack of a follow-up email for three days further compounded the issue, causing a follow-up email that was perceived as overly salesy and inappropriate.

Why it works

• The AI system was able to scrape and personalize emails, saving time and effort.
• The AI's ability to find and use alumni profiles on LinkedIn made the process scalable.
• However, the AI's lack of thorough testing led to a significant number of errors and misunderstandings.
• The follow-up email's poor quality and salesy tone further exacerbated the negative responses.
• The failure serves as a cautionary tale about the importance of thorough testing and understanding edge cases.

The playbook

1. Start by identifying the target audience: Clearly define the alumni you want to reach out to and ensure you have their explicit consent to send them personalized emails. This step is crucial to avoid legal issues and maintain a positive relationship with your alumni community.

1. Develop a robust AI model for email generation: Use a proven AI model like Claude or Qwen to generate personalized emails. Tailor the AI model to your specific use case by training it on a dataset of successful alumni emails and continuously refining the model to improve accuracy and relevance.

1. Implement a robust email automation system: Integrate your AI-generated emails into an email automation tool like Mailchimp or Sendinblue. Set up automated follow-up emails to re-engage recipients who did not respond initially. Ensure your system has proper tracking and analytics capabilities to monitor the effectiveness of your outreach.

1. Test extensively and iterate: Before sending out a large volume of emails, conduct thorough testing to identify potential issues such as missing data, incorrect information, or technical glitches. Iterate on your AI model based on feedback from test recipients to improve the quality and effectiveness of your emails.

1. Implement a robust feedback loop: Establish a system for collecting feedback from recipients and incorporating it into your future outreach efforts. This includes setting up mechanisms to handle complaints, request clarifications, and address any concerns raised by recipients. Continuously improving your approach based on real-world feedback will help you build stronger relationships with your alumni community.

Where people get it wrong

1. Failure Mode: Ignoring Privacy and Consent. The founder's AI system scraped LinkedIn profiles without consent, which is illegal and unethical. What to Do Instead: Always obtain explicit consent from the individuals before collecting or using their personal information. Use a clear opt-in process where people are made aware of how their data will be used.

1. Failure Mode: Ignoring Personalization Limits. The AI system personalized emails based on scraped LinkedIn data, but it likely failed to capture the nuances of each individual. What to Do Instead: Develop a more sophisticated AI that can analyze and understand more complex data points, such as the alumni's career paths, interests, and previous interactions with the organization. This can provide a more tailored message that resonates better with each recipient.

1. Failure Mode: Not Testing Thoroughly. The AI system failed to identify edge cases and produced emails with hallucinated information, which confused recipients and led to negative responses. What to Do Instead: Conduct extensive testing, including beta testing with a diverse sample of recipients. This includes testing for various edge cases and feedback from recipients to ensure the AI can handle a wide range of scenarios gracefully.

Do this next

1. Start with a small, controlled test email campaign to a small, specific group of alumni. This will help you identify any issues with your AI model and ensure it is ready for a larger rollout.

1. Use this test data to fine-tune your AI model. Make sure to thoroughly test all possible edge cases and ensure the AI is generating accurate, personalized content.

1. Once you're confident in your AI's performance, schedule a follow-up email campaign. However, ensure you have a robust plan in place for handling any negative responses or issues that arise.

1. Monitor your campaign closely, and be prepared to make quick adjustments based on feedback and performance data. This proactive approach will help you avoid the pitfalls of the past and ensure a smoother fundraising process.