I built an email scorer in one HTML file with zero dependencies. Here is every threshold and where it came from.
A developer created a lightweight, framework‑free email scoring tool that evaluates cold outreach based on measurable criteria. The tool assigns a 0‑100 score using deterministic thresholds for subject length, body length, readability, calls to action, personalization, and spam phrases. It emphasiz…
In the world of B2B sales, the quality of a cold email can mean the difference between a quick reply and a silent inbox. Yet most advice on crafting the perfect outreach is vague: "Keep it short," "Be personal," or "Add a call to action." These guidelines are hard to measure and even harder to test. To bring clarity, one developer built a tiny, zero‑dependency email scorer that turns those abstract rules into concrete numbers.
Why a Deterministic Scorer?
Modern solutions often rely on machine‑learning models that require an API call, a network connection, and a hefty codebase. While powerful, they lack the reproducibility that salespeople need. If you tweak a single word and get a different score, you can’t tell whether the change mattered. The developer’s goal was threefold:
- Determinism: The same input must always produce the same output.
- Privacy: Users can paste real, sensitive emails without fear of data being sent to a server.
- Transparency: Every threshold is documented, not hidden behind a black‑box algorithm.
How It Works
The tool is a single index.html file—about 220 lines of vanilla JavaScript. No frameworks, no build step, no external API calls. Paste a subject line and body, hit the button, and receive a score out of 100 across eight checks. The scoring logic is intentionally banded rather than linear, mirroring how email filters treat spam indicators.
Here’s a breakdown of the key metrics and their thresholds:
- Subject Length: 28–55 characters. Shorter subjects often indicate automation; longer ones risk truncation on mobile clients.
- Body Length: 50–150 words. Under 50 words is too brief to persuade; over 150 words tends to lower reply rates.
- Reading Grade (Flesch‑Kincaid): Target 8 or below. The algorithm uses a heuristic syllable counter to estimate grade level, accepting a margin of error.
- Call to Action (CTA): Keyword matching against a list of common phrases (e.g., "are you open", "quick call"). A question mark at the end of the email also triggers a CTA hit.
- Personalization: The email must include the prospect’s company name. The tool counts occurrences of the exact company name, ensuring it’s not just a generic placeholder.
- Pronoun Ratio: The ratio of second‑person pronouns (you, your) to first‑person pronouns (I, we) must be reasonable. A low ratio indicates the email is too focused on the sender.
- Spam Phrases: A list of 20 known spammy words or patterns (e.g., "no obligation", "risk free"). Each hit reduces the score heavily.
- Subject Shouting: All caps in the subject line are penalized, as they often trigger spam filters.
Each metric carries a weight—spam phrases are worth 20 points, subject length 15, body length 15, CTA 15, and so on. A perfect email scores 100; anything below 60 is a red flag and should not be sent.
Why Rule‑Based Beats AI Here
One might argue that a language model could score an email more accurately. However, models are inherently stochastic: the same prompt can yield slightly different outputs each time. For salespeople who need to tweak a single word and see the impact, that uncertainty is unacceptable. Moreover, a deterministic tool respects privacy—no data leaves the user’s browser. Finally, the transparency of rule‑based thresholds allows the author to justify each decision, fostering trust among users who may otherwise be skeptical of opaque AI systems.
Limitations and Next Steps
The scorer deliberately omits factors that cannot be assessed from a pasted string, such as DNS records, sending reputation, or the actual value of the offer. It also does not generate the email itself; for that the developer recommends a separate proposal generator that drafts outreach based on the sender’s and prospect’s websites.
Future enhancements could include optional integration with a reputation API for advanced users, or a visual heatmap that highlights problematic sections in the email. For now, the tool serves as a quick, privacy‑respecting audit that helps sales professionals refine their messaging before hitting send.
Why it matters
A reliable, transparent scoring system empowers sales teams to craft more effective outreach, reduce spam flagging, and protect sensitive data—all without the overhead of complex AI models.
Key points
- Deterministic, privacy‑first scoring in a single HTML file
- Eight measurable metrics with documented thresholds
- Band‑based scoring mimics real spam filters
- Rule‑based approach offers transparency over black‑box AI
- Tool excludes reputation checks and offer quality
- Ideal for quick pre‑send audits in B2B sales
Frequently asked questions
Can I use this tool for personal email outreach?
Yes, the tool is generic and can score any email, but it was designed with B2B cold outreach in mind.
Does the tool send my email data to a server?
No, all processing happens locally in the browser; no network calls are made.
What if my email uses a company name with special characters?
The script escapes special characters in the company name before matching, ensuring accurate personalization detection.
Can I modify the thresholds?
The thresholds are defined in the single HTML file; you can edit them if you prefer different criteria.
