I built an email scorer in one HTML file with zero dependencies. Here is every threshold and where it came from.

The author built an email scorer in a single HTML file without any dependencies. This scorer assesses cold emails and provides a score out of 100 based on eight checks. The entire project is deterministic and does not rely on models or network calls. The thresholds for each check were determined to ensure consistency and privacy.

The subject line length is important, as short subjects (28 characters) and mobile truncation are correlated with automation. The scoring is 15 points for a perfect subject length, 8 points for being too short, and 6 points otherwise. Capital letters and exclamation marks are also considered, with more of them increasing the score. Common spam phrases are matched as substrings against the subject and body, with the score adjusted based on the number of hits found.

In the body, a word count between 50 and 150 is optimal for response rates. The reading grade is calculated using the Flesch-Kincaid formula, aiming for a score of 8 or below. The call to action check uses keyword matching against a predefined list of phrases, with a fallback for unusually phrased requests. Personalization is achieved by counting the occurrences of the company name in the email.