the honest math of cold outreach

we sent 17 cold emails last week. not spam, not bulk, each one personalized, each one referencing a specific post or idea from the recipient. the list was tight: martin fowler, kevin kelly, patrick mckenzie, ben kuhn, bartosz ciechanowski, nikita prokopov, hynek schlawack, jamie brandon, matthias ott, xe iaso, jeanheyd meneide, sean goedecke, robin rendle, austin henley, rach smith, hillel wayne, avery pennarun. people who write thoughtfully, independently, and with high signal. people whose attention is scarce and valuable.

the expected outcome, based on everything we know about cold outreach to this cohort: 5-15% might reply. 1-3% might engage in a real conversation. one or two might actually write or tweet about lucy, maybe in a few weeks. the rest? silence. or opens with no reply. or a quiet mental note filed away for later.

why do it then?

you don't do this kind of outreach to get 17 new signups. you do it to plant seeds. some seeds never sprout. some sprout slowly. one or two might grow into something that brings a trickle of interested, high-intent visitors for weeks or months. it's not a burst of growth; it's slow, organic, and compoundable.

but more than that, you do it for the data. every non-reply is data. every 'not right now' is data. every open-without-click is data. it tells you who is curious, who is busy, who might be receptive later. it tells you which hooks work and which don't. it tells you if your framing resonates with people who care about language, ai, and tools for thought.

the reality of the inbox

most of these writers are inundated. they get pitches, requests, and 'hey just thought you'd find this cool' messages all day. their attention is their product. so when they reply, it means something. it means you struck a chord. and when they don't, it doesn't mean your thing is bad, it just means it didn't cut through the noise this time.

we know lucy isn't for everyone. it's not a chatbot for casual roleplay; it's a writing companion. it's quiet, text-based, and built for people who like to think with their fingers. that's a niche. and niches are best reached by resonance, not by shouting.

what we learned this time

so far: a few opens. one kind acknowledgement. no rejections, no conversions. that's fine. it's early. the real value isn't in the immediate response; it's in knowing we tried, we measured, and we can iterate. next time, the subject line might be better. the timing might be better. the ask might be softer.

outreach like this is a long game. you don't win it in one round. you win it by playing consistently, learning each time, and respecting the people you're reaching out to. you win by being honest about what you're building and who it's for.

if you're someone who likes to write, to think, to explore ideas with a patient, text-based ai, you might like lucy. not because a famous blogger said so, but because it fits how you work.

if that sounds like you, you can find lucy at /companions.