Reply craft
Under 1,000 followers, originals are homework. Replies are distribution. A good reply is a mini-post parked on a page the model is already willing to show. You are not networking. You are publishing where distribution already exists.
Keep the ratio, not the raw hours. ClimbX-speed is 50–100 replies and costs two to three hours. This course’s default is 15–25 good replies. Good means it passes the standalone test. Eighty “this!” replies is worse than twelve that could stand as posts.
Three columns, not one list
| Column | Who | Job |
|---|---|---|
| A · Peers | 25 accounts at ~100–2,500 in your niche | Become a known name. They notice. They reply back. Most of the block, every day. |
| B · Borrow | 10 accounts at ~5k–80k who post daily | Borrow a room. Only with an add. First 15 minutes. 5–8 replies. |
| C · Reciprocals | 15 people who already replied to you | Gravity. Talk every week. Protect these names. |
Split time 70 / 30: A and C first, B second. Big-account replies without an add are noise. Luckforest’s public recap is the warning: chasing heroes did not compound. Same-size tribe did.
How to pick the 25 peers
They post about the thing you post about. They are close enough in size that a reply from you is socially cheap for them to answer. They are not a “connect?” account and not a news handle. Write the 25 names down. A list that lives in your head becomes “whoever is on For You,” which is how you waste a month.
Size of the post matters too. Reply where the author will notice you. A reply on a 40-like post from a founder with 800 followers is worth more than a reply on a 4,000-like post from a celebrity. The big account never sees you. The peer does, and their followers are your first distribution. Hallam calls this the “make friends” strategy, and says “reply guy” is the wrong name for it: you are not the person under every big tweet, you are finding your bubble.
Finding warm, small posts. In his one-to-about-198-follower experiment (two days, roughly 200 manual replies at the peak, no follow-for-follow), he used filtered feeds and search to surface posts small enough that the author would see the reply. Use search terms from your niche, open the repliers as well as the original poster, and sort by recent. Caveat he named himself: people found the account partly because they already knew his main account. Do not use 198 in two days as your forecast.
The standalone test
Before you tap Reply, ask: could this stand as its own post? If the sentence only works as “Great point!” it is cheerleading. Delete it. Cheerleading trains two graphs: theirs (you are a fan) and yours (you produce empty text).
Four shapes that pass
- Number + condition. “This works under 500 followers. Past that the same reply looks needy.”
- Specific add. One fact, one counter-example, one tool the author missed.
- Mini-story. Two sentences from this week, not a memoir.
- Precise question. A question only someone in the niche can answer. Not “thoughts?”
Fill-in versions to keep in a note:
Volume, timing, 30-minute window
15–25 good replies a day is the default. Answer the author if they reply to you, 30 minutes, not tomorrow. Mutual-reply weight is one of the largest numbers in Chapter 1.
IST for English-speaking builders: peers late morning. Borrow-reach 8–10pm when US founders post. Friday can be a scoreboard original. Sunday is review, not a new persona.
What a reply is not
A chatbot paste. A quote-tweet dunk. A “let’s connect” under a stranger. A thread of seven compliments. A pitch on first contact. Those train the wrong graph and, at scale, look like the bots Phoenix is built to bury.
If you only have 45 minutes, do not stretch to 50 replies by lowering quality. Stay at 15–25 that pass the test. Quality means “would I publish this on my own profile.”
Write Column A (25), B (10), C (start empty). Notifications on for five Column B accounts. Write four replies today, one of each shape, into the X log before you post them. Post them. Log whether anyone answered.