Do follower counts change credibility?

Yes. A follower count can change how credible someone appears.

But it is not a credibility score.

In controlled experiments, a larger displayed audience sometimes raised trustworthiness and competence. In another, both the smallest and largest network counts lost to the middle. A more recent Instagram experiment found that micro-influencers looked more credible under both message conditions, although a stronger bundle of post cues narrowed the gap.

The count is one visible status cue. Viewers combine it with the account’s network pattern, message quality and fit with the subject. Change those cues and the same number can mean something different.

Credibility is not the same as popularity

The first trap is linguistic.

Popularity asks whether a person seems widely followed. Likeability asks whether observers respond positively to them. Opinion leadership asks whether they appear influential in a domain. Credibility is usually split into judgments such as trustworthiness and competence.

None of those measures proves that a statement is true.

A large account may be popular but wrong. A small account may be expert but unknown. A creator can look credible while promoting a product that makes no sense for their expertise.

The experiments become useful only when their outcomes stay separate.

The middle count beat both extremes in one Twitter experiment

David Westerman, Patric Spence and Brandon Van Der Heide created six mock Twitter pages. The profiles varied both the number of network connections and the gap between followers and accounts followed.

The result was not bigger is better.

Both very low and very high connection counts produced lower judgments of expertise and trustworthiness than the middle level. The follower-to-following relationship mattered too: a narrower gap raised perceived competence.

Goodwill did not change.

This is the cleanest warning against treating a public count as a linear scale. The interface cue moved some credibility dimensions and left another alone. It also came from the Twitter norms of 2012, not a timeless threshold for every platform.

A high celebrity count raised trust and competence

A second experiment points in the opposite direction.

Seung-A Annie Jin and Joe Phua randomized 160 U.S. undergraduates across mock Twitter pages for a semi-fictitious celebrity. The pages varied product, positive or negative brand tweets, and high or low follower count.

The high-count celebrity scored 4.02 for trustworthiness, compared with 3.63 in the low-count condition. Competence was 4.03 versus 3.45. Both differences were statistically significant on the study’s seven-point scales.

Audience size also interacted with tweet valence on product involvement, buying intention and willingness to pass the message along.

That matters. The count changed the source impression, but the commercial response depended on what the source said.

The study used one student sample, one fictitious celebrity and two low-involvement products. It shows a causal impression effect inside that experiment—not that millions of followers create expertise.

More followers produced popularity more reliably than leadership

Marijke De Veirman, Veroline Cauberghe and Liselot Hudders moved the question to Instagram.

In their first experiment, 117 Instagram users saw a fictitious profile with 2,100 or 21,200 followers. The account also followed either 32 or 32,200 people.

More followers made the influencer look more popular. That popularity helped likeability. The route from popularity through opinion leadership was much smaller.

In plain language: viewers noticed the audience, but they did not automatically convert reach into perceived opinion leadership.

The network ratio changed the picture again. Following many accounts hurt likeability when the influencer had relatively few followers; that conditional effect disappeared in the higher-follower condition.

In a second experiment, a high count also weakened the perceived uniqueness of an unusual product. Popularity helped the person while making an exclusive object feel less exclusive.

Product fit was a larger cue than the count

Lieke Janssen, Alexander Schouten and Emmelyn Croes randomized 432 Dutch Instagram users to one of two adapted health-and-fitness influencer profiles with either 5,039 or 578,000 followers. They also varied the promoted product: a protein shake was intended as a good fit and ice cream as a poor fit.

The larger audience produced a small positive credibility effect: F(1,428) = 5.16, p = .024, eta squared .012.

The intended-fit condition produced a much larger one: F(1,428) = 42.47, p < .001, eta squared .090.

There was no interaction between the two for credibility. The visible count helped a little; a coherent reason for this person to discuss this product helped much more.

But fit and product identity changed together. The experiment cannot tell us whether every difference came from fit rather than other differences between a protein shake and ice cream.

Credibility then predicted ad attitude, product attitude, purchase intention and influencer likeability in the authors’ mediation model. That downstream path should not be rewritten as more followers caused sales. The experiment shows how several perceptions connect under one controlled setup.

Stronger bundled content narrowed the gap

Rumen Pozharliev, Dario Rossi and Matteo De Angelis compared micro- and meso-influencers in an online experiment with 192 participants and a laboratory study with 112 participants.

They also varied what the paper called argument quality in a positive Instagram product post. The stronger condition bundled longer copy, factual details, a 30% discount, hashtags and emojis. It did not isolate reasoning alone.

Micro-influencers were perceived as more credible in both conditions. Under the weaker post condition, their rating was 5.14 versus 3.50 for the meso-influencer. Under the stronger bundle, the ratings were 5.41 versus 4.57. The larger influencer gained ground but did not overtake the smaller one; the remaining difference was statistically significant.

In the separate laboratory sample, the meso-influencer plus stronger bundle produced higher frontal-theta activity, interpreted by the authors as greater cognitive work. That study did not measure credibility, so it cannot show that the extra cognitive work caused the credibility improvement.

The experiment covered one product category and positive reviews. It does not establish a universal advantage for every small account or a penalty for every large one.

The same number can carry different meanings

The table is deliberately asymmetric. A cue can justify a question without settling the answer.

Primary sources

1. Westerman, Spence and Van Der Heide, 2012
2. Jin and Phua, 2014
3. De Veirman, Cauberghe and Hudders, 2017
4. Janssen, Schouten and Croes, 2022
5. Pozharliev, Rossi and De Angelis, 2022