Key finding
Across 14,752 keyword comments on one team's LinkedIn lead-magnet posts, connection invites sent to commenters were accepted at 56.29%, against a 27.11% cold benchmark on the same platform, a different-population comparison rather than a controlled test, and 32.05% of the 6,063 leads created went on to reply in a DM conversation.
What we measured
"Post content, get leads" is the most repeated and least measured claim on LinkedIn. This study measures it, on our own pipeline, stage by stage: a post offers a resource, a reader comments a keyword, the platform detects the comment, delivers the asset by DM, connects with the commenter if needed, and tracks what happens next.
The dataset is one workspace, our own: about 10 Reachium team accounts publishing through the platform's content engine. This funnel is not a demo. It is how our own calendar gets filled, and these are the actual counts at every stage, including the stage where most of the volume disappears.
Companion studies cover the posts themselves: what the lead-magnet format earns compared to standard posts, and what reach looks like across the whole series.
Methodology
Window. Every keyword comment processed between 21 February and 6 August 2026, pulled 6 August 2026. 172 posts had an active comment trigger during the window.
Stages. A processed comment creates a lead and enrolls it against the post's DM campaign. Commenters already connected to the author receive the asset by DM directly; commenters not yet connected receive a connection invite first. From there the platform records replies and booked meetings at the lead level.
Deduplication. People comment on several posts. All funnel outcomes below are people-level: a person who commented on three posts and replied once counts once. Enrollment-level counts are given only where labelled.
Maturity. Every one of the 636 connection invites in this study was sent at least 30 days before the pull, so the 56.29% acceptance figure is a matured rate, computed the same way as our outreach benchmarks.
Bookings are tracked bookings only: meetings booked through the platform's own booking detection. A commenter who replies in a DM and later books over email is invisible to this study. The tracked booking numbers are a floor, not a total. Where the study quotes our own booking calendar instead, it says so explicitly, because that count is ours rather than the platform's.
The findings
| Stage | Count |
|---|---|
| Keyword comments processed | 14,752 |
| Distinct people behind them | 6,012 (2.45 comments each) |
| Leads created | 6,063 |
| Immediate DMs at detection | 830 |
| Connection invites sent (not-yet-connected commenters) | 636 |
| Invites accepted | 358 (56.29%) |
| Leads that replied in a DM conversation | 1,943 (32.05% of the 6,063) |
| People with a tracked booking | 30 |
A handful of commenters resolve to more than one lead record, which is why 6,012 people yield 6,063 leads. Counting comment events rather than people: 4,008 arrived from someone already connected to the posting account, 794 from someone not yet connected, 729 from someone with an invite pending, and 9,221 arrived without a resolvable connection state at detection; the pipeline works through those over time.
The acceptance rate is the headline
Invites sent to people who had just commented on a post were accepted at 56.29% (358 of 636, all matured 30+ days). Our outreach benchmark across 180,155 matured cold connection requests on the same platform is 27.11%.
That is 2.08 times the cold rate. The comparison carries a real caveat, spelled out in the limitations: these are different populations, not a controlled test. But the direction matches what the mechanism predicts. A person who just asked you for something is not cold, and LinkedIn's acceptance data reflects it.
Conversations and meetings
1,943 leads, 32.05% of the 6,063 created, went on to reply in a DM conversation. For scale: our cold outreach data shows 27.55% of accepted connections replying, which is 27.55% of the roughly quarter who accept. Here, 32.05% of the leads created from commenters ended up in a conversation.
Thirty people booked a meeting through tracked booking flows between 22 March and 1 August 2026 (31 enrollment-level bookings; one person booked through two campaigns). Against the top of the funnel that is 0.20% of processed comments, and roughly one tracked booking for every 5.7 posts that carried a trigger.
The tracked number is the floor, and the floor sits a long way below the room. Over the same months our own booking calendar recorded more than 250 meetings: demos, setups and strategy calls booked with the team. That count is ours, read from the calendar rather than measured by the platform, and it spans everything this system produces together, because content and outreach feed the same pipeline and a booking link does not say which one sent it. We attribute at least nine in ten of those calendar bookings to the content funnel, and that attribution is our judgment, not a measurement. The 30 is what this dataset can prove end to end, comment to booking, which is why the table reports it.
What this means
The funnel's economics live in the middle, not the bottom. The tracked 30 bookings is understated by design, since it counts only bookings the platform can see, and our calendar's 250+ meetings over the same period says how much escapes: most meetings book through calendar links and email threads that no tracking can attribute back to a comment. The middle is not understated: 1,943 DM conversations with prospects who self-selected by asking for a resource is the pipeline output, and sales teams know what a reply rate of nearly a third is worth.
The 56.29% acceptance figure is the study's most transferable number. It quantifies the difference between arriving in someone's invite list as a stranger and arriving as the person whose post they just commented on. Warmth is usually a metaphor; here it is 29.18 points of acceptance rate (derived from the two figures above).
It also explains why the lead-magnet format wins so hard in the companion study. The format's comments are not vanity engagement. Each one is a person raising a hand, and this study is what the raised hands turned into.
In practice
A founder doing this manually can copy the mechanics without the platform: offer a real asset, ask for a keyword comment, deliver by DM, and send invites to commenters you are not connected to. The numbers here say those invites will land at roughly double a cold invite's acceptance, and that about a third of the leads a keyword post creates will end up in conversation.
A team evaluating content ROI should instrument the middle of the funnel before the bottom. Bookings leak into channels no tool can see; replies and connections do not. Leads to conversations is the rate this study found most robustly measurable, and at this team's scale it ran at 32.05%.
Anyone benchmarking should note the volumes, not just the rates. It took 172 posts and 14,752 comments to produce these outcomes. The funnel works; it does not work overnight.
Limitations
The 56.29% vs 27.11% comparison is not a controlled experiment. The warm side is one team's 10 accounts inviting people who engaged with them; the cold side is the whole platform's outreach across many workspaces, targeting strategies and time windows. Population differences beyond warmth are baked into the gap. We publish it because the mechanism is plausible and the gap is large, not because the populations match.
Bookings are undercounted by design. Only meetings booked through tracked flows appear in the funnel table. Meetings arranged in the DM thread, over email, or on a rescheduled link are absent, and our own calendar recorded 250+ meetings across all sources in the same period. Do not compute cost-per-meeting from the 30, and do not compute conversion rates from the 250+ either: the calendar count mixes content and outreach sources and is our operational number, not a platform measurement.
9,221 comment events lacked a resolvable connection state at detection. The stage counts for already-connected and invited commenters describe the resolved subset, not all 14,752 events.
Replies are counted as any inbound DM reply on the lead, without reading sentiment. A "no thanks" counts. The 32.05% measures conversations started, not conversations won.
One team, one format, one niche. The posts behind this funnel offer B2B lead-generation assets to an audience that wants them. A weaker asset or a mismatched audience changes every rate in the table.