Two hundred small, or twenty large? · Part 3 of 3
Same budget, both strategies. Then what to actually do.
We spent $2,000 four ways on paper, at prices creators have agreed with us. On the typical post, seventeen big creators beat a hundred small ones by more than two to one. Pick the small accounts on their record and count their rare big posts, and the hundred win.
September 2026 · 8 min read
Engagements from 100 picked small creators
For $1,500, scored on the average post so the rare big ones count.
Engagements from 17 large creators
For $1,989 on the typical post, every large creator on our books.
For a post that gets 100 engagements
From a picked small creator at $15 a post. Bought blind, a large creator costs $351, against $117 picked.
Cost per thousand views in the campaign everyone quotes
The trade calls this CPM. The document promoting the campaign says $30 to $80 is good value.
01 · The sums
Two thousand dollars, spent four ways
Someone proposed a plan: hire around 200 small creators instead of 10 to 20 large ones. Small creators, called micro, cost $10 to $50 a post. Large creators, called macro, cost $500 to $1,000. We tested it at the prices creators have agreed with us, scored on the engagement their posts earned: likes, comments and shares added together.
A tier is a band of follower counts, such as 5k to 20k. On the typical post the plan loses. Picked on record and scored on the average post, it wins.
Four words. The typical post is the middle one when a creator’s posts are lined up by engagement. The average post is total engagement divided by posts, so it counts the rare big ones. Picked means the typical post already gets 30 or more and the account has posted since 2025. Blind means you choose a tier and pay without looking.
100 small creators
typical post · whole tier
for $1,500
17 macro creators
typical post · whole tier
for $1,989
100 picked small creators
average post · rare big ones counted
for $1,500
17 picked macro creators
average post · rare big ones counted
for $1,989
On the typical post, macro wins by more than two to one: two fifths of the result for three quarters of the money. Count the rare big posts and buy only picked accounts, and the hundred small creators win, and cost less.
The whole roster, each creator posting once, says the same on the typical post.
per engagement, macro
Price paid divided by engagements earned, for twenty-five macro creators at the prices they accepted.
per engagement, micro
One hundred and nine micro creators, same test. Fewer engagements, more paid for each.
times apart, inside one tier
Move up a tier and the typical post gets about 3 times more. Inside one tier, the best creator gets 22 to 994 times more than the worst.
One cost these sums leave out is your time. Time ten deals with small creators, and add it to their bill.
02 · The tail
Buy small creators for the big post, and pick them first
Nobody buys a small creator for their typical post. You buy one for the tail: the rare post that does far better than usual. Part 2 counted how often a post clears 100, 500 or 1,000 engagements across four million posts, and priced one post over each bar at what we already pay: $15 for a small creator, $117 for a macro creator.
One post that clears
100 engagements
One post that clears
500 engagements
One post that clears
1,000 engagements
What $2,000 buys over the 100 bar: about 44 posts from picked small creators, 17 from picked macro, and about six from blind macro.
Bought blind, a macro creator costs $351, $1,053 and $2,808 for the same three posts, and a blind small creator costs even more above 100. Never buy the tail blind.
Of those big posts from small accounts, 39% were about a new job, a life event or company news, or reused something made elsewhere. A brand cannot order one of those.
03 · How the feed works
LinkedIn is built to help small accounts
LinkedIn's own paper on how the feed picks posts describes pulling 2,000 candidates from hundreds of millions by what the reader cares about, with the biggest gains for newer members with few connections. Neither it nor LinkedIn's engineering blog names follower count as a reason a post gets shown.
A second paper describes a lift for creators who get little feedback, switched on for some members and not others.
More creators kept posting each week
Against members without the lift. LinkedIn's own live test.
More shares in total
A share puts a post in front of a whole new audience.
Those numbers measure whether creators keep posting, which is what LinkedIn wants, not what a buyer gets. For that, one outside dataset divides engagement by impressions, an impression being one showing of a post.
Typical engagement per view, by follower band
AuthoredUp · 372,812 posts from personal profiles · Sep 2025 to Feb 2026 · engagement ÷ impressions
The rate is flat across the whole small range, at 2.56%, 2.68% and 2.53%, and only falls above it. Whether it drops as followers rise depends on what you divide by. The source sells a LinkedIn tool and reports on its own customers.
So LinkedIn does not hold small accounts back. But a buyer pays for engagements per dollar, not a rate, and section 01 shows what that buys.
There is no public price list for LinkedIn creators. The sites that price every other platform leave LinkedIn out. Our agreed deals are the only measured prices we have seen. Sell that.
04 · The trade press
Twenty-six claims from the trade press, one survivor
We put two paid-for reports on what LinkedIn creators cost and earn through three tests: can the number be traced, does the method hold up, does it change the decision.
One report says micro wins by five to ten times on cost per engagement. Its table is built on guesses: each micro creator has around 5,000 followers, a post reaches 6% of them, about 300 people, and earns roughly 18 engagements. Its own notes say the figures are examples from other platforms. The big finding is that guess, multiplied out. Every price in the reports failed too.
Then the campaign every report quotes. Traced to its source: $12,000, seven creators with around 13,000 followers each, one post a month for three months. Back came 66,449 unpaid views, 45 people asking to see the product and $1.1M in pipeline, which is possible future sales the company declared itself, not money in the bank. Divide spend by views and you get the number nobody quotes.
Cost per thousand views, in dollars
CPM · $12,000 divided by 66,449 views, against the range the same document calls good value
Six times the bottom of the range the same document calls good value.
One idea in it is worth keeping: a “how did you hear about us” box on the sign-up form. It costs almost nothing and tells you whether a post led to a sale.
Ask what a tier claim divides by before you believe it.
The same posts, divided by different things, point to opposite answers. Both charts can be honest.
05 · What to do
Four simple rules
- 1
Run a hundred small creators picked on their record, not two hundred bought blind.
Price them on the average post and buy them for the tail at $15. Use the rest to get a message out early and test what lands.
- 2
Book creators again, not tiers.
Rebook the ten with the lowest measured cost per engagement, whatever their tier. And stop paying the going rate for 5k to 20k followers, the worst typical cost on our books at $1.30 per engagement. Bring it down towards $20.
- 3
Put a tracked link on every paid post.
Of 476 recorded clicks, 29% are flagged as not from a real person, and ten paid creators have no real click at all. Three creators bring in 73% of the real clicks. Find out who is behind them.
- 4
Spend about $1,400 on one test.
A second and a third post from about fifteen creators across all three tiers. One post can be luck; three show skill. If cost per engagement holds, buy creators, not tiers. Agree in advance which result leads to which decision.
That is the series. Tier tells you little. The creator tells you a lot. The plan that started all this works in one form: a hundred small accounts picked on their record, priced on the average post since 2025 and bought for the hundreds of engagements they deliver.
Sources · show
- our production database: 11,729 posts by 189 creators, 54 payout lines across 33 creators, 476 recorded clicks. Two internal test accounts excluded from price figures. Queried 2 to 3 September 2026.
- AuthoredUp, “How to calculate engagement rate on LinkedIn”: 372,812 posts from personal profiles, September 2025 to February 2026, impressions denominator. Vendor-published.
- LinkedIn, out-of-network content retrieval for the LinkedIn Feed, arXiv 2510.14223, October 2025. 2,000 candidates from hundreds of millions by topical interest.
- LinkedIn Engineering, “Engineering the next generation of LinkedIn's Feed”, 12 March 2026. Author follower count is not named as a distribution factor.
- LinkedIn, creator-side ranking term, ego-cluster randomised production experiment, arXiv 2106.11312. Retained weekly creators +5.26%, total reshares +12.04%. No member counts, dates or durations disclosed.
- Vector LinkedIn creator campaign, as reported by its own content lead: 7 creators, 21 posts over three months, $1.1M self-declared pipeline credited through a “how did you hear about us” box. Not audited. CPM of $181 is $12,000 ÷ 66,449 impressions; the $30 to $80 efficient range and the $80 renegotiate line are from the same document.
- Verification runs, 2 to 3 September 2026: 26 claims from two research documents, 1 survived, 17 judged untraceable by two or more of three independent verifiers, meaning the source could not be found, not that the claim was proven false. The two reports reach opposite conclusions and only one of them measured anything.
- our corpus tail pass, September 2026: 4,125,900 posts by 11,115 cleaned accounts, no per-author cap; share of posts passing 100, 500 and 1,000 engagements by band and by account cell. Mean engagement per post 109.48 for picked 5,000 to 10,000 follower accounts and 464.59 for picked roster macro creators; costs at the deal-book medians of $15 micro and $117 macro. Picked on the 30+ typical post alone, without the 2025-on check, small-account costs are $60 and $900 at the 100 and 500 bars. Buyable shares for picked 5,000 to 10,000 follower accounts: 51%, 37% and 47%; 39% of tail posts were a new job, a life event, company news or reused material. The macro tail has not yet been read post by post, so no buyable share is applied to it. The 2,904 row is 100 times the roster micro figure of 3,165 over 109 creators. Method in Part 2.
- Full report: Two hundred small, or twenty large?, linklabs.wiki, report 02, September 2026.
- Photographs: Pexels, free licence. No posts are shown in this piece.