What a link in your post costs you
A post that sends the reader off X gets 22% fewer views than the same kind of post without a link. Put the link in your first reply instead.
for a post that carries a link out, against the same kind of post without one
The answer, before the working
Everything here compares posts of the same kind: a post with nothing attached against another post with nothing attached. And it compares a founder with themselves, never with somebody who has a different-sized following. The unit is views, because views are the thing you actually get.
Put the link in a reply to your own post, not in the post. A post carrying a link out gets 22% fewer views than the same kind of post without one. Compare each founder only with themselves and the figure is 34% fewer, with 24 of 27 founders who do both coming off worse. A reply carries no such cost, because it is not competing for the same distribution.
Comparing you against your own posts rather than against everybody, a post that normally reaches a thousand people reaches about 660 with a link in it — that is the 34% figure, not the 22% one.
First, a rule we published and are withdrawing
We published that the single worst thing you can post is a bare link. On this data it is not merely unsupported — it points the other way. Posts stored as nothing but a short link pull 6,995 views against 3,259 for the same founders' other posts. That is 115% more, not less.
The reason is not that bare links work. It is that these are not bare links. They are posts whose text and media our collector failed to record — 204 of them in six weeks, disproportionately from large accounts. The biggest is a 23-character row from a 986,000-follower account with 11.7 million views. Nobody gets 11.7 million views from a naked link; that is a post whose contents we lost.
A published number that turns out to be a fault in our own collection is the most important kind of correction to make loudly, because it is the kind nobody outside can catch for us.
The trap that hides the real number
Ask the obvious question — do posts with a link do worse? — and the answer comes back as no measurable difference. That answer is wrong.
X appends its own short link to the text of any post carrying a photo or a video. So every photo post and 99.8% of video posts technically have a link in them. A flag that reads like "did you send people away" is really "did you attach something" — and attaching a video is one of the best things you can do to a post. The two effects cancel and you measure nothing.
The fix is to hold the format constant: posts with nothing attached and a link, against posts with nothing attached and no link. Same kind of post, same authors, one difference. Every number on this page is that comparison.
X adds its own short link to any post with a photo or video attached, so the plain "has a link" flag is really a "has something attached" flag.
6 July to 16 August 2026 · founders board
Swipe the chart sideways to see all of it →
What a link actually costs you
Both bars hold the format constant, so the only difference between them is the link.
A link costs 22% of your views and 52% of your reactions, on 441 posts with a link against 5,374 without.
Posts with nothing attached on both sides, so the format is not doing the work.
441 linked and 5,374 plain posts · 6 July to 16 August 2026
Swipe the chart sideways to see all of it →
| Posts with nothing attached | Typical views | Typical reactions | Posts |
|---|---|---|---|
| No link in the post | 2,444 | 31 | 5,374 |
| A link in the post | 1,916 | 15 | 441 |
| What the link costs you | 22% of your views | 52% of your reactions | — |
Same kind of post either way, so the only difference is the link.
And it is not a handful of founders dragging the average
The figures above pool everybody together. The stronger test is to compare each founder only with themselves — their own posts with a link against their own posts without. That depends on how many posts of each kind somebody needs before they count, so here are three answers rather than the one that flatters us most.
The stricter the bar, the larger the cost and the more lopsided the head count. The honest reading is that a link costs somewhere between a fifth and two fifths of your views, and that it costs almost everybody something.
Three bars, three answers to "how many posts before somebody counts". None of them is flattering.
2+ linked: 18% on 51 founders · 3+ linked: 34% on 27 founders · 4+ linked: 41% on 24 founders
Swipe the chart sideways to see all of it →
| Founders who count | How many | Views lost to the link | How many came off worse |
|---|---|---|---|
| 2+ linked and 3+ plain posts | 51 | 18% fewer | 39 of 51 (76.5%) |
| 3+ linked and 5+ plain posts | 27 | 34% fewer | 24 of 27 (88.9%) |
| 4+ linked and 6+ plain posts | 24 | 41% fewer | 23 of 24 (95.8%) |
The middle row is the one quoted at the top of this page.
Does it depend on how big you are? We cannot say.
An earlier version of this finding said the cost grows the bigger your account gets, and gave three neat numbers to prove it. Measured in views those numbers do not line up that way, so that claim is withdrawn.
The cost is there at all three sizes. The ordering is not what a "bigger accounts pay more" story predicts, and it does not stay put: set the bar at two link posts before somebody counts and the largest accounts lose most, set it at three and the mid-sized ones do, set it at four and it moves again. Each band ends up resting on between 4 and about twenty founders, which is nowhere near enough to rank them.
A link costs you reach at every size we can measure. Whether it costs a large account more than a small one, we do not know. If you are very small the amounts are small in absolute terms, so the traffic may well be worth more to you than the reach. That is a judgement, not a finding.
The cost appears in all three bands. The ordering between them does not hold up, which is why no size rule is published here.
under 5K 25% fewer · 5K–50K 8% fewer · 50K+ 50% fewer
Swipe the chart sideways to see all of it →
| Account size | Typical views, no link | With a link | Views lost | Posts compared |
|---|---|---|---|---|
| under 5,000 followers | 208 | 155 | 25% fewer | 92 with, 969 without |
| 5,000 to 50,000 | 1,695 | 1,563 | 8% fewer | 199 with, 2,463 without |
| over 50,000 | 11,142 | 5,537 | 50% fewer | 150 with, 1,942 without |
The cost is there at all three sizes. The ordering is not.
It shows up again in a separate block of posts
The strongest thing we can do with a finding is ask it again of posts that had nothing to do with producing it. On January and February 2025 — a completely different set, eighteen months earlier — posts with nothing attached and a link got 3,503 views against 5,084 without. That is 31% fewer, on 746 link posts from 141 founders.
One wrinkle in the older figure: those posts carry no record of what was attached to them, so "nothing attached" has to be worked out from whether the post had an image or a video rather than from the label. On the 2026 posts the two ways of asking pick out exactly the same posts and give exactly the same answer, which is why the older figure is quotable at all.
| Block of posts | Typical views with a link | Without | Views lost |
|---|---|---|---|
| 6 July to 16 August 2026 | 1,916 | 2,444 | 22% fewer |
| January and February 2025 | 3,503 | 5,084 | 31% fewer |
Two separate sets of posts, eighteen months apart, same answer.
What to do with this
Put the link in the first reply, not in the post. The post costs you 22% of its views for carrying it; a reply costs nothing, and the reader gets there in one extra tap.
Attaching a photo does not buy the link back. Every comparison here already holds the format constant, so this is the cost of the link itself, not of the format.
Do not read a size rule into it. It costs you reach whether you are small or large. Anyone giving you an exact multiplier for your follower count, including the earlier version of this page, is over-reading the data.
Ignore any bare-link rule, including ours. It cannot be measured on data like this, because a post recorded as nothing but a link is almost always a post we failed to record.
How this was measured
Every original post from the founders board on X between 6 July to 16 August 2026 that had nothing attached: 5,815 posts from 99 founders. Replies and quotes are excluded — they compete for different distribution — and so is anything carrying a photo or a video, for the reason in the section above.
A post "carries a link out" when its text contains a link that is not the one X adds for attached media. Posts with a photo or a video attached are excluded from both sides, so the comparison is between posts of the same kind.
The pooled figure compares the typical post on each side. The founder-by-founder figure compares each person's own linked posts with their own plain posts and then takes the typical result across people, so somebody with a large following cannot move it.
What this cannot tell you
We cannot see where the link went. A link to your own site and a link to somebody else's article are the same thing here, and they are plainly not the same decision.
We measure views on X. A link that costs you readers and earns you signups is a trade this data cannot price for you.
Restricting the comparison to posts with nothing attached is what makes it honest and also what makes it small: 441 link posts across 99 founders.
The founder-by-founder figures rest on between 24 and 51 people depending on where the bar is set. That is the price of holding everything else constant, and it is why this page gives a range rather than one confident number.
More from this board
Every number on this page comes from tracKeting, founders board, 99 founders on X, 6 July to 16 August 2026. Questions are answered from this page's own data, and where the data cannot answer one, it says so.
Want to ask questions about this? Paste it into Claude or ChatGPT once you have connected tracKeting, and it can read the whole thing and answer from it.
tracKeting tracks what founders, AI brands and Geneva estate agencies publish, and what it earns them. See what a board costs.