Blog · Tools
August 7, 2025 · 5 min read
Lavender is an AI writing assistant that sits inside Gmail, Outlook, and most major email clients. As you write a cold email, it scores it in real time across several dimensions — email length, subject line effectiveness, reading level, mobile optimisation, and personalisation. The score updates as you type.
The premise is simple: most salespeople write emails that are too long, too formal, and too focused on the sender rather than the recipient. Lavender's real-time feedback nudges writers toward shorter, clearer, more human copy. The question is whether that nudge actually changes results.
Lavender's email score is built from several components. Length is weighted heavily — Lavender consistently pushes toward shorter emails, typically below 75 words for a first-touch message. Reading level favours simple language; Flesch-Kincaid analysis penalises sentences that require a college education to parse. Mobile optimisation checks whether the email reads well on a phone, where the majority of emails are first-opened.
Personalisation scoring checks whether the email contains signals that it was written for this specific person — name, company reference, specific context. Generic emails score low. Emails with specific, researched details score high. The AI coach suggests improvements as you write, and users can see exactly which elements are pulling their score down.
Teams that adopt Lavender consistently report their average email length dropping significantly — from 200+ words to under 100 — within the first month. That change alone tends to improve reply rates. Long cold emails signal low confidence and high effort; short ones signal directness and respect for the recipient's time.
The bigger change is in self-awareness. Lavender makes visible the habits that salespeople develop without noticing: starting emails with "I" (Lavender flags this), using filler phrases ("I wanted to reach out"), or writing emails that focus on features rather than problems. Seeing these patterns scored in real time accelerates the kind of improvement that normally takes months of coaching.
A high Lavender score does not guarantee a reply. It guarantees an email that follows the structural best practices that correlate with higher reply rates across a large sample of emails. Individual emails can score low and still work — because the prospect has a specific context, or the offer is uniquely compelling, or the timing is perfect. Optimise for the score as a signal, not as the goal.
Lavender also does not evaluate the quality of the research behind the personalisation. An email that references a generic fact about the prospect's industry might score well for personalisation while offering nothing specific enough to earn a reply. Use Lavender alongside real research, not as a substitute for it.