The Influencer Offer That Exposed a Coordinated Campaign

The useful clue was an awkward question: who is paying?

In May 2021, a German YouTuber received an offer from a representative of Fazze to share material attacking the Pfizer vaccine. The recipient publicized the approach and asked who financed the campaign. The representative treated that information as confidential. Journalists from netzpolitik.org and ARD’s Kontraste subsequently reported examining a restricted section of Fazze’s website containing the campaign brief. It instructed influencers to avoid sponsorship labels and present the material as their own independent perspective. Original investigation, May 26, 2021.

The proposed deception concerned a relationship the audience would be denied the chance to evaluate. An established creator could remain recognizably themselves while delivering a message shaped by someone else. No computer-generated face or automated posting schedule was necessary to that arrangement.

That makes the episode a useful test of the question people often ask when a coordinated campaign surfaces: were these bots, paid influencers or real people? Those categories overlap. A real person can accept payment; an authentic account can publish supplied material; an invented persona can be operated by a human. Each possibility calls for a different kind of evidence.

What the platform could connect

On August 10, 2021, Facebook announced that it had removed 65 Facebook accounts and 243 Instagram accounts in a network it linked to Fazze. Facebook described it as a subsidiary of a marketing business registered in the United Kingdom, with operations conducted mainly from Russia. Facebook’s announcement.

Its detailed report described fake accounts promoting misleading material and attempts to recruit influencers with existing audiences. The removal also covered authentic accounts associated with the operators. Facebook’s July 2021 report, pages 7–8.

The account totals therefore need careful handling. They describe the properties Facebook removed under its enforcement policy. Calling the combined total “308 bots” would add a claim about automation that the figures do not supply. Calling it “308 paid influencers” would make a different unsupported leap.

The recruitment evidence and the platform evidence also do different work. A brief can reveal what an intermediary wanted a creator to say and conceal. A platform investigation can connect accounts beyond the recipient’s inbox. Neither, on its own, tells us that every person encountering the content understood its origin or had agreed to help distribute it.

When a pattern is only a pattern

Often there is no exposed brief. Investigators begin with repeated behavior: accounts sharing the same material, using similar language or appearing together around particular stories. The challenge is to explain why a pattern deserves attention without assigning its participants a motive prematurely.

A 2021 study by Diogo Pacheco and colleagues illustrates that boundary. In a case examining images shared around Hong Kong protest discussions, the researchers identified three suspicious clusters comprising 315 accounts. The clusters included pro-protest and anti-protest material. Their method detected accounts circulating overlapping sets of images unusually often; it did not establish whether the coordination was malicious, benign, automated or human-operated. Pacheco and colleagues, Case Study 2.

The same paper warns that ordinary website sharing buttons can generate identical messages, and that people can deliberately coordinate for legitimate causes. The authors recommend inspecting identified clusters rather than treating the output as a complete explanation. Methods and Discussion.

Consider a hypothetical neighborhood group campaigning to save a library. Members agree to share the same announcement that evening. A chart of their activity might correctly show coordination. The relevant questions would concern what they represented to their audience: did they openly organize as residents, or fabricate additional residents and conceal who controlled them?

A genuine pattern can therefore be paired with a false accusation. Detecting that people acted together is one finding. Demonstrating that they deceived an audience about that relationship is another.

The client who remained unidentified

The Fazze investigation reached a consequential boundary. Facebook expressly left open who had commissioned the campaign. It also said the provenance of an allegedly hacked document used by the operation remained unresolved. July 2021 report, page 8.

That qualification belongs beside the attribution. Identifying an intermediary leaves another question about who hired it. A location, a corporate connection or a message favorable to a particular government cannot fill the missing space in a contract.

Readers should look closely at the verbs in an investigation. “Shared” describes distribution. “Operated” concerns control. “Paid” concerns a financial relationship. “Directed” concerns instructions or decision-making authority. Replacing one with another can turn a limited finding into a much larger allegation.

The same discipline applies to recruited creators. An offer documents an approach. Acceptance, payment, publication and knowledge of the ultimate sponsor each require support. A person who exposes a recruitment attempt occupies a very different position from someone who knowingly carries it out. A careful account should preserve those distinctions even when all the names appear in the same investigation.

What attention would prove

Facebook reported that nearly all the campaign’s Instagram posts received no likes, while paid influencers’ posts attracted some limited attention. July 2021 report, page 8.

Those observations help describe visible engagement. They cannot establish how many viewers believed a claim, changed their plans or discussed it elsewhere. A like may express approval, but silence has several possible explanations. Someone may never have seen a post, may have ignored it, or may have read it without responding.

A claim about persuasion needs evidence connecting exposure to a relevant change, with attention to other plausible explanations. The existence of a campaign establishes an attempt. Evaluating its effects is a further research task, and the measures should match the claimed outcome.

Reading the evidence in order

For a reader confronting the next dramatic network diagram, a useful evidence ladder starts with the observation. What, precisely, did the accounts do? Is the pattern repeated and unusual in that setting? Could a public event, a common source or an openly organized effort explain it?

The next step concerns the relationship. What connects the accounts beyond similar posts? Then comes the alleged deception: which identity, affiliation or sponsorship was misrepresented, and what supports that conclusion? Evidence about payment or instructions should be tied to the particular participant and action it concerns.

Attribution and impact need their own supporting records. Strong evidence on an earlier question does not automatically settle either of them. It is possible to establish a deceptive campaign convincingly while leaving its ultimate client, the role of automation or its persuasive effect unresolved.

The Fazze episode rewards that patient reading. The offered arrangement exposed a concrete problem: an audience would be shown apparently independent judgment while a consequential relationship was concealed. Keeping the remaining questions open makes that finding more useful. It tells readers exactly what they have learned, and what a future investigation would still need to establish.