Timing is 80% of outbound: when a buying signal is hot vs cold

Timing is 80% of outbound. The rest is mechanics — message structure, channel choice, follow-up cadence, all the things that get the airtime. The reason it gets the airtime is that it's tractable. You can A/B-test a subject line. You can't

Timing is 80% of outbound. The rest is mechanics — message structure, channel choice, follow-up cadence, all the things that get the airtime. The reason it gets the airtime is that it's tractable. You can A/B-test a subject line. You can't A/B-test the laws of half-life. But the half-life is what's actually killing your reply rate.

A perfect message sent four days late lands at roughly 2% reply. The same message sent three hours after the trigger lands at 11–22%. The delta isn't your copy. It's the clock.

A perfect message sent four days late lands at 2%. The same message sent three hours after the trigger lands at 11–22%. The delta is not your copy. It is the clock.

What "hot" means, operationally

A buying signal is hot when four things are simultaneously true. The post is recent. The OP is still actively engaged in the thread or the comments. There aren't already five vendors who've replied. And the post matches your ICP closely enough that you have something specific to say.

Drop any one of these and you're in the warm-to-cold range. Drop two and you're cold. The four-condition definition is strict on purpose — most operators call signals hot when they're really warm, then wonder why their reply rates aren't matching the case studies.

Recency alone isn't enough. A 30-minute-old post where the OP posted and immediately went to bed is colder than a 6-hour-old post where the OP is still replying to comments. The OP's behavior is more diagnostic than the timestamp. A signal where the OP has commented in the last hour is hot regardless of when the post itself went up.

ICP fit is the one most teams want to relax. Don't. A hot signal that's a stretch for your ICP is a warm signal at best. The contextual leverage of timing only works if you have something genuinely specific to bring. If you're hand-waving the relevance, you're sending a templated cold email with a fresh timestamp, and the reader will smell it.

The decay curves are different on every surface

Reddit, HN, LinkedIn, and X each have their own half-life shape. Treating them with one cadence is the unforced error.

Reddit. Posts live in /new for roughly 11 minutes of meaningful visibility before they're buried. The /hot algorithm extends life if the post gains traction in the first 30 minutes. The OP-engaged window — the time during which the OP is actively reading replies — runs about 6 hours from the original post for daytime posts, with a sharp dropoff. Late-night posts get a second life in the morning when the OP wakes up and checks. We covered the visibility mechanics in the high-intent Reddit post anatomy piece; the timing layer sits on top of those mechanics.

HackerNews. Front-page visibility is about 12 hours with a brutal decay after the first 3–4. But comment threads stay live for 48–72 hours, and Algolia indexes everything indefinitely. The hot window for in-thread replies is 6 hours; the warm window for cold follow-ups based on the thread is 48 hours; past that you're cold. The HN intent playbook breaks down the structural surfaces; the timing layer is what makes them actionable.

LinkedIn. The decay curve is non-monotonic. A post can sit dormant for three days then resurface because someone in a different network engaged with it. The signal isn't the post date — it's the recency of comment activity. A 5-day-old post with 20 comments in the last 6 hours is hotter than a 1-hour-old post with no engagement.

X (Twitter). Half-life is the shortest of the four. Visibility is roughly 60–90 minutes for an account under 50K followers, with the lion's share of replies arriving in the first 15. The OP-engaged window is similarly short — they've moved on to the next thing within 2–3 hours.

The implication: one outbound cadence does not work across surfaces. The teams that try get the worst of every world — too late on X, too early on LinkedIn, fine on Reddit by accident.

What makes a signal cold

The mirror image of the hot definition. Past the surface-specific window. Multiple replies already in the thread (especially vendor replies). OP has gone silent — no comments, no upvotes, no edits. Or the post itself was a vent rather than a problem statement, in which case it was never really hot to begin with.

The vendor-reply count is the one most operators underweight. The first vendor in a Reddit thread gets read carefully. The second gets compared to the first. The third gets ignored unless they bring something the first two missed. By the fifth, the OP is done with the thread and has either chosen or moved on. If you arrive sixth, your reply rate is 1–2% regardless of how brilliant the message is.

OP silence is the second tell. If the OP posted, then disappeared without engaging in their own thread for 24 hours, they've either solved it offline or lost interest. A cold message at that point is firing into a void. You're not being heard, you're being archived.

The third tell: stale follow-on context. If the OP commented "actually we ended up going with X, thanks everyone" anywhere in the thread, the signal is dead. Read the whole thread, not just the original post. The decision often shows up in a child comment 18 hours after the original post and changes everything.

The 90-minute rule and the 24-hour rule

Two rules that cover most of the operational decisions.

The 90-minute rule. If the signal is on a fast surface (X, Reddit /new in a high-velocity sub), respond within 90 minutes or don't respond. The window closes that fast. Templated alerts that arrive 45 minutes late are fine; ones that arrive 4 hours late are noise.

The 24-hour rule. If the signal is on a slower surface (HN comment thread, LinkedIn post, Reddit /hot in a smaller sub), respond within 24 hours or substantially change the message. The "I saw your post" framing only works while the post still feels recent to the OP. Past 24 hours, the message has to acknowledge the lateness or it reads as if you've been sitting on the signal.

These rules sound mechanical because they are. The teams that optimize for them are the teams whose reply-rate math doesn't collapse under scale. We worked through the actual reply-rate decay in the reply rate math piece; the punchline is that timing accounts for more variance in reply rate than any other lever, including message quality.

Timing accounts for more variance in reply rate than any other lever, including message quality. The 90-minute rule is not a preference. It is a structural feature of the math.

The daily ritual that catches the hot window

The naive solution is to monitor everything in real time. That's a path to burnout. The structured solution is a daily ritual built around the windows that matter for your surfaces.

The morning sweep. 25–35 minutes, first thing. Catch up on overnight signals from your slow surfaces — HN comment threads, LinkedIn posts, Reddit posts from low-velocity subs in EU/Asia time zones. These were warm overnight; you're catching them in the warm-to-hot tail. Process the ones that match ICP closely enough to write a real reply.

The midday check. 10–15 minutes around noon local. This is the live window for fast surfaces — Reddit /new in your subs, X in your category. The signals that came in over the last 4 hours are hot. Triage and reply.

The end-of-day pass. 15–20 minutes before you stop. Catch the afternoon signals before they go cold overnight. Also: follow up on anything from the morning that got engagement.

That's roughly 60–70 minutes a day of structured monitoring. The teams that try to monitor continuously throughout the day either accomplish nothing else or burn out in three weeks. The ritual works because it batches.

The thing the ritual cannot do alone is filter the firehose. There are too many posts to read every one. This is where a buying-signals monitor like Shadow Inbox earns its place — it runs the relevance and intent classification continuously so the daily ritual operates on a pre-filtered list of maybe 30–60 signals instead of the raw feed of thousands. The ritual stays human; the firehose gets machine-handled. Without the filter, the ritual collapses under volume; without the ritual, the filter just produces a backlog.

Sequencing across the cooling curve

Most teams send the same first-touch regardless of where on the cooling curve the signal sits. That's the second-biggest unforced error in the timing space. The message has to acknowledge the temperature of the moment.

Hot (0–6h): "Saw your post about X. We hit the same wall — happy to send the breakdown if useful." Direct, recent, no apology for the timing.

Warm (6h–window-end): "Saw your post yesterday about X — wanted to make sure you've got what you need. We solved this last quarter and the breakdown is in this doc if you want it." Slight acknowledgment of the lag, still useful.

Cold (past window): "Saw your post from last month about X — figured I'd check in case you're still looking. If you've already solved it, ignore me; if not, here's what worked for us." Explicit acknowledgment, lower-pressure ask, a real off-ramp.

The cold message can still convert — it's just operating on a different premise. The OP is no longer "in the moment." They've either solved it, deferred it, or forgotten about it. Your message has to give them a reason to re-engage rather than just continue an existing conversation. We laid out the contextual mechanics in the contextual cold message piece; the cooling curve is the layer on top of that.

The follow-up cadence also changes with temperature. Hot signals get one follow-up at 48–72 hours. Warm signals get one follow-up at 5–7 days. Cold signals get no follow-up — you've used the one shot you had.

Where the math stops working

It stops working at scale. The whole timing model assumes you have humans in the loop who can write a real reply within the window. If you scale past your team's capacity, signals start aging out before anyone gets to them, and you're just burning intent.

The honest constraint: a single operator can handle 30–50 hot signals a week with the ritual. A small team of three can handle 100–150. Past that you need either more operators or smarter triage — there's no way to template your way out of it without losing the contextual leverage that makes the timing matter in the first place.

It also stops working when your category is too small. If there are only 5–10 hot signals a week in your space, the timing infrastructure isn't worth building — you can monitor manually. The math favors timing infrastructure when you're seeing 50+ signals a week and the manual pass can't keep up.

And finally: it stops working if you confuse "hot" with "in your favor." A hot signal where the OP has already chosen a competitor is not your signal. Hot is about the OP's engagement, not about your closing odds. Read the temperature of the OP, not the temperature of your pipeline.

● FAQ

How long does a Reddit signal stay hot?

Roughly 6–12 hours from the moment of post for a typical thread. Half-life is around 11 minutes for visibility on /new, but the OP's engagement window — when they're still checking the thread — runs about 6 hours, with a long tail of 24–48 hours for late-night posts that get found in the morning.

What's the HN window?

Front-page visibility is about 12 hours with a sharp drop after 3–4. Comment threads stay live for 48–72 hours. Reply within 6 hours for the OP-engaged window; reply within 48 for the lurker-engaged window.

Do these windows apply to LinkedIn?

LinkedIn decays differently — the algorithm resurfaces posts when they get late engagement, so a post from 5 days ago can still be hot if the comment count just spiked. The signal isn't the post date; it's the recency of activity on the post.

What if I find the signal late?

You can still engage, but the message changes. Hot reply: 'I saw your post.' Cold reply: 'I saw your post from last month — wanted to ask if you'd already solved this, because if not we ran into the same thing.' The acknowledgment of lateness reads as honest and changes the dynamic.

How do you actually catch the hot window without sitting on a feed?

A daily ritual at the right time of day plus a monitor that pushes only the hot signals. The ritual handles the manual judgment; the monitor handles the firehose. Doing one without the other either burns you out or lets the hot signals decay before you see them.