The #1 reply quality lever: name a specific detail from the post

The stat. In the agent's reply-quality score, the single biggest swing comes from specificity: mirroring a concrete detail from the original tweet earns a +15 reward, while a reply with no mirrored detail takes a −20 penalty — a 35-point spread on a 0–100 scale, larger than any other single factor (openers, question count, humanizers are all ±5 to ±10).

How it was measured. The quality model was tuned against the engagement corpus: features that correlated with higher real-world engagement (likes + 3×replies + 5×bookmarks) were rewarded, features that correlated with lower were penalized. "Mirrors a specific detail from the post" — naming the actual number, tool, or claim the author mentioned, rather than replying generically — came out as the dominant positive lever, and generic replies as the dominant drag.

The honest caveat. The exact point values (+15, −20) are hand-tuned weights in a heuristic scorer, calibrated on one account's data, not regression coefficients with confidence intervals. The direction — specific beats generic — is one of the most robust findings in the study and matches common sense, but the precise magnitude is a design choice. Specificity also can't save a reply that's specific about the wrong thing or that misreads the post.

The tactic it implies. Before you reply, find the one concrete detail in the tweet and respond to that — the exact metric, the named tool, the specific claim. "This resonates" applies to ten thousand tweets and gets ignored; "the part about churn dropping after you killed the onboarding email is wild — did retention move too?" could only be written about this tweet, and that's exactly why it gets seen. Generic replies are invisible. Specificity is the whole game.