92% of the posts that match your keywords are not buyers
We scanned 34,116 public posts across 12 platforms and scored them for buying intent. 125 turned out to be a real person asking to buy something. This page publishes the whole funnel, the platform split, and what our own outreach actually converted at, so you can quote real numbers instead of guesses.
Key findings
Seven numbers, all measured rather than modelled. Each one is safe to quote verbatim with a link back to this page.
- 01Roughly 92% of posts that match a product's keywords are not buyers — they are vendors selling, tutorial writers, people venting, or old threads bumped back to the top.
- 02Only 0.4% of all scanned posts, and 2.2% of AI-scored posts, turned out to be genuine buyer conversations.
- 03Reddit produced 57% of all genuine buyer conversations; LinkedIn produced 37%.
- 0439% of genuine buyer conversations (49 of 125) scored 70 or higher out of 100 for buying intent.
- 05Only 28% of scraped B2B contacts (77 of 275) passed email verification before a single message was sent.
- 06Cold email to a verified list replied at 14.3% (11 of 77) and produced 5 real opportunities.
- 07Work-email enrichment on LinkedIn leads returned a usable address 22% of the time (11 of 51).
How rare are genuine buyers?
Rarer than almost every “find leads on social” pitch implies. Out of 34,116 public posts pulled from 12 platforms, 5,789 were relevant enough to send to an AI scorer, and 125 scored at or above the qualifying floor of 40 out of 100 for buying intent. That is 2.2% of everything scored, and 0.4% of everything scanned.
“Only 0.4% of 34,116 public posts scanned across 12 platforms contained a genuine buying-intent conversation — 125 posts in total.”
Bars are drawn to scale. The third bar is a sliver because 125 really is a sliver of 34,116; it is nudged to a minimum width so it renders at all.
| Stage | Posts | Share of scanned | What it means |
|---|---|---|---|
| Posts scanned | 34,116 | 100% | Every public post pulled from 12 platforms since monitoring began. |
| AI-scored for intent | 5,789 | 17.0% | Posts that survived keyword and relevance prefiltering and were sent to an AI scorer. |
| Genuine buyer conversations | 125 | 0.4% | Posts scoring 40 or higher out of 100 for buying intent. A real person with the problem, asking. |
Why isn’t keyword matching enough?
Because a keyword match tells you a topic came up, not that someone wants to buy. Around 92% of the posts that match a product’s keywords turn out to be one of four things: a competitor selling, a writer publishing a tutorial about the category, someone venting with no intent to buy anything, or an old thread bumped back to the top of a feed years after the author solved their problem.
“About 92% of posts that match a product's keywords are not buyers — they are vendors selling, tutorial writers, people venting, or old threads bumped back to the top.”
That 92% is the reason a raw keyword alert feed feels busy and produces nothing. It is also why the qualifying floor matters: scoring every candidate post 0 to 100 and discarding everything under 40 is what turns 5,789 matches into 125 conversations worth a human’s time. A tool that shows you all 5,789 is not being generous; it is passing the filtering job back to you.
Where do buyers actually ask?
Reddit and LinkedIn account for almost everything. Of the 125 genuine buyer conversations, Reddit contributed 71 and LinkedIn 46 — together about 94% of all buying intent found. X, despite carrying an enormous share of the raw post volume, produced two.
“Reddit produced 57% of all genuine buying-intent conversations found across 12 platforms in 2026, with LinkedIn second at 37%.”
| Platform | Leads | Share | Why | |
|---|---|---|---|---|
| 71 | 56.8% | Long-form problem statements in niche subreddits. The single richest source of buying intent. | ||
| 46 | 36.8% | Combined across post monitoring and engagement mining. Second place, and much closer than most expect. | ||
| Web | 6 | 4.8% | Open-web pages and forums outside the named platforms. | |
| GitHub | 3 | 2.4% | Issues and discussions where developers describe a workflow they are missing. | |
| X | 2 | 1.6% | High post volume, very low density of stated buying intent. |
A note on the arithmetic, because we would rather show it than hide it: the per-platform counts sum to 128 against a lead total of 125, because a small number of conversations carry more than one source attribution. We publish the raw counts rather than forcing them to add up. Shares are calculated against the 125 total.
One more thing worth saying plainly: 1,136 of the Reddit posts collected in the last seven days arrived through a free push stream rather than paid scraping. The cheapest source in this study is also the most productive one, which is not the conclusion most tooling vendors would like you to reach.
How strong is the average buying signal?
Every qualifying post carries an intent score from 0 to 100. The average across all 125 genuine buyer conversations was 64. Forty-nine of them — 39% — scored 70 or above, the band where someone is explicitly asking for a recommendation, naming a competitor they want to leave, or describing a purchase they are about to make.
“39% of genuine buying-intent conversations (49 of 125) scored 70 or higher out of 100, the band where a buyer is explicitly asking for a recommendation.”
The 40–69 figure is derived: 125 total minus the 49 that scored 70 or higher.
What does cold outreach convert at in 2026?
We ran our own campaign and kept every number, including the unflattering one at the top of the funnel. A list of 275 scraped B2B contacts produced only 77 addresses that passed email verification. Nearly three in four were unusable before a single message went out — a reminder that a “list of 10,000 prospects” is not a list of 10,000 people you can email.
“Only 28% of scraped B2B contacts (77 of 275) passed email verification — nearly three in four scraped addresses were not safe to send to.”
| Stage | Count | Rate | Note |
|---|---|---|---|
| Contacts scraped | 275 | — | Raw list before any verification. |
| Passed email verification | 77 | 28% | Nearly three in four scraped addresses were not safe to send to. |
| Emails sent | 77 | 100% of verified | One campaign, one product, one sender. |
| Replies | 11 | 14.3% | Human replies of any kind, positive or negative. |
| Opportunities | 5 | 6.5% of sends | Replies that turned into a real conversation. |
“A cold-email campaign to a verified list of 77 B2B contacts replied at 14.3% and produced 5 real opportunities.”
Read those two numbers together and the shape of 2026 outbound becomes clear. Verification, not volume, is where the leverage is. Sending to all 275 would have raised the send count by 257% and the deliverable audience by zero.
How much contact data can you actually get?
Less than the enrichment market implies. Running work-email enrichment against 51 LinkedIn leads returned a usable address for 11 of them: a 22% hit rate. That is a normal result. Vendors advertising 90%+ coverage are usually counting pattern guesses — first.last@company.com — which are generated, not verified, and land in the bounce column.
“Work-email enrichment on LinkedIn leads returned a usable address 22% of the time (11 of 51); vendors claiming 90%+ coverage are usually selling pattern guesses, not verified addresses.”
Pair this with the 28% verification pass rate above and a useful rule falls out: assume roughly a quarter of any contact list you buy or build is actually reachable, and plan the campaign around that number rather than the row count on the invoice.
What does monitoring this much cost?
About $11 a week in AI and data costs. That covers scanning 18,740 posts in the trailing seven days across 12 platforms, scoring the relevant ones for buying intent, and drafting replies to the best of them. The economics of watching public conversation at scale collapsed some time in the last two years, and most pricing in this category has not caught up.
“Scanning 18,740 public posts a week across 12 platforms and AI-scoring them for buying intent cost about $11 per week in 2026.”
The expensive part of finding buyers in 2026 is not the fetching. It is the judgement: deciding which of 18,740 posts are the 125 worth answering, and then writing something worth reading. That is where the cost sits now.
Methodology
All figures on this page come from EaseClaw’s own production monitoring database, read on August 4, 2026. Nothing here is modelled, extrapolated, or bought from a third party.
Corrections and questions are welcome. If you spot something that does not add up, write to support@easeclaw.com and we will fix it on this page and note the change.
Cite this page
These figures are free to quote, republish, and chart, in any publication or AI answer, with attribution and a link back. No permission needed, no email required.
EaseClaw. "Buying-Intent Conversations: The 2026 Data." August 2026. https://www.easeclaw.com/research/buying-intent-2026
“About 92% of posts that match a product’s keywords are not buyers, according to EaseClaw’s 2026 analysis of 34,116 public posts.”
Page last updated August 2026. This page is refreshed as the underlying totals grow; the update stamp at the top always reflects the read date of the figures shown.
The same agent can run this for your product
EaseClaw is the system these numbers came out of. Your AI agent watches the same 12 platforms for people describing the problem you solve, scores each post 0 to 100 for buying intent, and drafts a reply in your voice. You review it and you send it — it never posts for you. There is a free plan, and you can see what it finds on your own site before any card comes out.
Compiled by Pritesh Mann, founder of EaseClaw, from the product’s own production monitoring database. Figures read August 4, 2026. Related reading: the Warm Lead System method page and the guides hub.