What is personalization at scale?

Spectrum of personalization

Mail merge ({first_name} only): minimum, doesn't move reply rates. Field-level (industry, company size): modest lift. Snippet-level (one specific data point per prospect): meaningful lift. AI-generated paragraph (multi-source synthesis): biggest lift but quality control matters.

Sources to personalize from

Recent LinkedIn posts, company announcements, podcast appearances, mentioned-in-news, hiring activity, mutual connections, shared groups, recent role changes.

Where personalization breaks

Compounding personalization (15 placeholder fields, AI fills none well) and over-stuffed openers ("I noticed your post about X, your company's recent funding, and that we both know Y" reads as creepy/automated).

Infonet's approach

Two-source synthesis: pick the strongest single signal from a max-of-three candidate pool, write one specific line, then move directly to the value statement. No placeholder fields, no compounding.