Micro vs Macro Influencers: What 2,393 Posts Say

Every article ranking for this question runs the same play. Define the tiers by follower count, quote an average engagement rate per tier from someone's industry study, note that micro is higher, conclude that micro influencers offer engagement on a budget while macro influencers offer reach, and close with "the right choice depends on your goals." I have read a dozen of them this week and the striking thing is not that they disagree. It is that they agree completely, and that none of them tests the one assumption the whole framework stands on: that knowing an account's follower band tells you what its posts will deliver. We have an index that can test that. So this post does.

One disclosure before the numbers: I work on UGCSignal, and the data below comes from our tracking index. It appears at the end as the alternative to choosing by tier at all, not as a secret third tier.

The tiers, quickly

The vocabulary is worth thirty seconds because every study you will ever quote uses it. Bands vary slightly by source; these are the most common cuts, and the ones used in every figure below.

The table stakes claims are real enough as far as they go. Bigger accounts do cost more, smaller accounts do reply to their comments more, and the engagement-rate studies behind the "micro sweet spot" line measure something true about likes. What the guides never do is follow the tiers forward into delivered outcomes: views, the thing a brand is actually buying. That is the gap, and it is measurable.

The test population

We took one consumer brand's category feed from our index: every organic TikTok post our category listening detected, back-caught to 2021 and live-scanned daily since late April. After deduplication, and keeping only creators whose follower count we actually hold rather than guessing, that is 2,393 posts by 1,887 creators, 329.8 million combined views, queried at the end of August 2026. Nobody briefed these creators and nobody picked them by tier. They are simply everyone who posted about one category, which makes them a fair jury for a question the campaign case studies cannot answer without selection bias.

What the tiers do predict

First, the steelman, because the data grants the taxonomy more than I expected.

Tier	Posts	Median views	Top 10% start at	Best post
Nano under 10k	1,031	500	6.5k	1.41M
Micro 10k to 100k	790	802	27.9k	4.79M
Macro 100k to 1M	429	22.2k	585k	8.68M
Mega over 1M	143	453k	2.42M	23.6M

The median post climbs three orders of magnitude as you walk up the bands: 500 views for a typical nano post, 802 for micro, 22,000 for macro, 453,000 for mega. If your question is "what will the middle post from this tier do," follower count answers it well. Anyone telling you tiers are meaningless is selling something too.

But look at the other two columns. The top tenth of micro posts start at 27,900 views, which is more than the typical macro post delivers. The best nano post in this feed reached 1.4 million views from an account with fewer than ten thousand followers. The spread inside each tier is wider than the gap between tiers, and that spread is where campaigns are actually won and lost. A tier predicts its median. You do not hire the median. You hire twenty specific accounts, and the tier tells you surprisingly little about those.

Where the top posts actually came from

Here is the same feed cut the way a results-obsessed buyer should cut it: take the 100 most-viewed posts and ask who posted them.

51Mega

35Macro

9Micro

5Nano

Top 100 posts by views. Population: 2,393 posts with known follower counts in one brand's TikTok category feed, queried August 2026.

Mega and macro accounts earned 86 of the top 100, which is the part the reach argument gets right. The other 14 came from accounts under 100k followers, including five from under 10k. Widen the lens to every post that cleared 100,000 views and the picture sharpens: 65 of those 303 posts came from accounts a "macro only" filter would have screened out before the campaign started. And the funnel runs both ways. Only half of macro posts beat their own tier's median by definition, and in this feed just one macro or mega post in every few clears a million views. Most posts from big accounts are unremarkable; some posts from tiny accounts are enormous. On an interest-graph platform like TikTok, distribution follows the content, not the follower list.

This is the same physics we measured from a different angle in the Modash review, on this same category feed at creator grain: the rank correlation between an account's followers and the views it actually delivered was 0.398, and 71 of the 168 creators in the top tenth by delivered views came from outside the top tenth by followers. Follower count is a real signal. It is just a weak one, and every tier framework treats it as the only one.

The engagement gradient does not survive contact with views

The load-bearing claim in the micro-influencer pitch is the engagement gradient: the smaller the account, the larger the share of its audience that responds. That is true of likes. Here is what it looks like when you measure what brands actually buy, the median post's views as a share of the poster's followers.

Nano

under 10k

34.8%

Micro

10k to 100k

3.8%

Macro

100k to 1M

9.2%

Mega

over 1M

18.5%

Median per-post views ÷ followers, per tier. Views, not likes: this is what a brand buys, and it is not the tidy downward slope the tier guides print.

Not a slope. A U. Nano accounts routinely out-deliver their follower counts, reaching a third of their audience size on the median post, because the algorithm shows a good video to people who never followed anyone. Then the curve collapses: the median micro post reaches under 4% of its follower count, the weakest tier in the feed, with macro not much better. And mega accounts, which the engagement studies place dead last, delivered 18.5% per follower, second best. One category, one platform, organic posts only; I would not carve this into stone as a law of nature. But it directly contradicts the gradient that every micro-vs-macro article treats as settled, and the mechanism is not mysterious: likes-per-follower measures a relationship, views-per-follower measures distribution, and TikTok distributes by content quality, not audience loyalty. If you are buying views, "micro influencers have higher engagement rates" is answering a question you did not ask.

How to actually use tiers

So the bands are neither meaningless nor decisive. Here is the honest division of labor.

Budget by tier, because pricing is tiered even though delivery is not. Rate cards follow follower counts. That is precisely the inefficiency you can exploit: the market prices the median, and you are allowed to hire above it.

Never pick individuals by tier. Within any band, the difference between the accounts that deliver and the accounts that do not is invisible in the follower count. The only signal that predicts a creator's next category post is their previous category posts, which means the pick list has to come from observed delivery. That is what detection-based discovery is: instead of filtering a directory by band, it watches your actual category, sees the nano account whose video just did forty times its follower count, and hands you the list ranked by what happened rather than who is big. The mechanics of building that list, with or without software, are in the finding-creators guide.

Judge the mix by outcomes, not by averages from other people's studies. A portfolio of twenty micros against two macros is an empirical question your own data can answer in a quarter, if posts are tied to orders at post grain. Every tier debate inside a brand that has attribution running ends quickly, because someone pulls up the number.

The short version

• Tiers predict medians well: in 2,393 organic category posts, the typical post climbed from 500 views (nano) to 453,000 (mega). The reach argument for big accounts is real.
• Tiers predict individuals badly: 14 of the top 100 posts and 65 of the 303 posts over 100k views came from accounts under 100k followers, and the top decile of micro posts out-delivered the median macro post.
• The engagement gradient inverts at view grain: micro was the weakest tier per follower (3.8% of audience per median post) and mega nearly the strongest (18.5%). Likes measure relationships; views measure distribution; brands buy views.
• Use the bands to budget, since pricing follows them. Build the actual pick list from observed delivery in your own category, which is a detection problem, not a filtering problem.