Maintenance
Part of How to tell whether sharing or recommendation carried a viral spike
How to Read a Viral Trend's Attention Curve
Not a roundup of viral trends updates 2027: read attention curves, spot manufactured activity, and understand disclosure duties when you pay for reach.
Some of what looks like spread was bought. Not all, probably not most, but enough that any read of a curve must consider it.
This is not a roundup of viral trends updates 2027. It explains how to distinguish manufactured attention from organic attention by reading a trend's curve, and covers the obligations that come when you pay for reach.
What to take away
- Manufactured attention has a different shape from the organic kind, and the shape is easier to read than any individual account is.
- You will never be certain about a single account, and trying to be is a waste of a week. Judge the curve, not the participant.
- If you pay for reach or a mention, the disclosure obligation is yours and does not transfer to whoever you paid.
Three things that get bought
They are different and get discussed as one, which muddles every conversation about them.
Coordinated posting from accounts that hide the coordination is astroturfing: manufactured activity presented as spontaneous public sentiment. The deception is about the source rather than the content, which can be perfectly accurate.
Multiple identities operated by one person is a sock puppet arrangement. This is the retail version, common in disputes and reviews, and usually cruder.
Automated activity at volume is what an internet bot does, and it's the least interesting of the three: easiest to detect, least persuasive. Most automation on any surface is mundane and disclosed.
How much of each of the three exists is not a number anyone has, which is most numbers in this field cannot be had in its usual form.
Reading the curve rather than the accounts
The same basic curve-reading approach can be applied to public trend activity on TikTok, X, Instagram, and YouTube. Each platform surfaces different signals, so compare like with like rather than treating their curves as interchangeable.
Organic pattern
- Onset
- Ragged, with false starts
- Timing
- Follows waking hours of a real audience
- Wording
- Varied, with misreadings and jokes
- Second order
- Replies to replies, arguments, tangents
- Decay
- Long tail with occasional revivals
- Composition
- Mixed account ages and histories
Manufactured pattern
- Onset
- Clean, often starting from a standstill
- Timing
- Even across hours, or bunched in a narrow window
- Wording
- Repetitive, or varied in a mechanical way
- Second order
- Little conversation beneath the top layer
- Decay
- Stops abruptly when the spend ends
- Composition
- Clustered creation dates, thin histories
The two rows that carry the most weight are the second-order activity and the decay. Real attention argues with itself and produces tangents; purchased attention rarely funds the second layer, because the second layer is the expensive part.
Organic decay is gradual, because people keep finding the item, while a paid campaign stops on the day the budget does. Which of the two you are watching is clearest in the four phases of a spike.
None of these rows is decisive alone. Several pointing the same way is worth acting on, and one is worth nothing: the ordinary discipline of not reading one indicator as a conclusion.
Why you will not be certain
Every detection heuristic has a benign explanation. A new account with no history is what everyone has on their first day. A narrow posting window is what a real audience in one time zone looks like. Repetitive wording is what happens when people copy a phrase they liked.
Accusing a specific account on this evidence is a costly mistake. Base rates worsen it, since manufactured accounts are a small fraction of any large population.
Even a good test produces mostly false accusations when applied one account at a time. That is the base rate problem in its standard form, so the sensible unit of judgment is the episode rather than the participant.
The practical rule is to reach conclusions about curves, keep them provisional, and never publish an accusation about a named account on pattern evidence alone.
When you are the one paying
The other half of the subject is your own conduct, where the rules are clearer than most people expect.
If you pay for a mention, provide something of value for a mention, or have a relationship that a reader would want to know about, that connection has to be disclosed, clearly and where the reader will see it.
The Federal Trade Commission's Endorsement Guides: What People Are Asking works through common situations, and its Disclosures 101 for Social Media Influencers covers placement and wording. The Guides Concerning the Use of Endorsements and Testimonials in Advertising appear at 16 C.F.R. Part 255; Section 5 of the FTC Act prohibits unfair or deceptive acts or practices. Read current versions rather than a summary, since this is one area where the details do the work.
Two things follow, worth stating in your own words. The obligation attaches to the person publishing, so handing it to an agency does not discharge it. And a disclosure that a reader has to look for is not a disclosure, whatever it says.
Bought attention posing as spontaneous fails on its own terms before any rule applies. It builds a mismatched audience by construction, and the costs of a mismatched audience are paid in full whether that attention was real or not.
Common questions
How much of a typical spike is manufactured?
Nobody can tell you. The figures in circulation are estimates from methods that cannot be validated, since the ground truth is known only to the operator. Treat any specific percentage as unsupported.
Should I report accounts I believe are coordinated?
Reporting through the surface's own channel is reasonable and low cost, because the operator can see things you cannot. Publishing the accusation is a different act with different consequences, and the evidence you have does not support it.
Does paid promotion count as manufactured?
Not if it is disclosed and the activity it produces is real. The distinction is concealment, not payment. Paying to be seen is ordinary; paying to look as though nobody paid is the problem.
Does this change how I read my own numbers?
It should make you slower to celebrate a clean onset. A perfectly smooth start is more often a distribution decision by the system than a burst of enthusiasm, and telling those apart is the mechanism question the arithmetic sets up. Your own decay will tell you more than the peak did.
