Marketing in 3 · Issue 43 · Thursday 29 October 2026
The followers who join in are the committed ones. In my survey of 1,356 people who follow brands on social media, a decision tree found that four beliefs about a brand decide how much people read, share and post about it, and commitment to the brand comes first. That produced five kinds of follower, from casual fans who mostly watch to committed fans who create. Feeds are now ranked and partly written by AI, but those four beliefs are still what moves a person to act.
With AI now part of almost every marketing conversation, this felt like the right paper of mine to revisit, so this time the series looks back as well as forward.
The question was simple. Brands want followers to read, comment, share and post about them. Most research at the time looked at people's motives. I wanted to know how what people think of a brand turns into what they do with it on social media, and whether those beliefs sort followers into recognisable types.
Brand pages in Poland shared my survey with their followers. Each person named a brand they follow and answered questions about it: how well they know it, what they associate with it, how good they think it is and how loyal they are. They also said how often they consume, contribute to and create content about it. Instead of averaging effects across everyone, I used a machine learning method, a conditional inference tree, which keeps splitting people on the belief that best separates them and ends with rules a manager can read.
Out of twelve beliefs, the tree kept four. The first split was commitment, a single statement: "I am committed to this brand". People who rated it 5 or above on the 7-point scale went one way, everyone else the other.
For the committed, the next question was familiarity: do they know the brand well? For the less committed, it was recognition, then whether the brand's associations were favourable. Perceived quality, for all the attention brands give it, did not make the cut.
Five end points came out of those splits. Brand passionates (625 people) recognise and like the brand without being committed. Brand fans (558) are committed and familiar. Low engaged followers (108) recognise the brand but think little of what it stands for. Brand curious followers (41) have neither commitment nor strong recognition. Prospect brand fans (24) are committed but only partly familiar with the brand.
| Group | Value (%) |
|---|---|
| Brand passionate | 46.1 |
| Brand fan | 41.2 |
| Low engaged | 8.0 |
| Brand curious | 3.0 |
| Prospect brand fan | 1.8 |
Terminal nodes of the conditional inference tree, share of 1,356 participants (Table 2).
Brand passionates are the quiet majority. Almost nine in ten consumed brand content at a medium level, and more than half created little or nothing. They are the audience that watches, likes now and then, and moves on.
Brand fans did a bit of everything. About one in five contributed at a high level, by commenting and sharing, and almost as many created content of their own.
The surprise was the small groups. A third of the brand curious created a lot, in what looked like learning by doing. And among prospect brand fans, committed but still discovering the range, a third were heavy creators too. Low engaged followers almost never created anything: two in three stayed at the bottom on creation.
| Group | Value (%) |
|---|---|
| Low engaged | 0 |
| Brand passionate | 2.54 |
| Brand fan | 17.96 |
| Brand curious | 33.34 |
| Prospect brand fan | 34 |
Creation of brand-related social media content, high level set at the top decile (Table 2). Brand curious and prospect brand fan groups are small (41 and 24 people).
I collected these data before generative AI was part of every marketing brief. Since then, feeds have been ranked by recommendation systems, brand posts can be drafted by a model in seconds, and people ask chatbots about brands instead of searching. It's fair to ask whether a survey of brand followers from that era still says anything useful.
My honest answer is yes, and for a simple reason. The technology decides what lands in front of someone. It does not decide whether that person feels committed to the brand, recognises it, knows it well or likes what it stands for. Those are human judgements, built slowly, and they are what turns seeing into joining in. When content becomes cheap and endless, the scarce thing is not another post. It is a follower who cares enough to answer.
A second reason makes me fond of this paper now. Much of today's AI in marketing is a black box that scores audiences without saying why. A tree like this one is machine learning you can read: if commitment is high and familiarity is high, expect a fan. I'd still choose a model a manager can explain over one they have to trust.
What I would do differently is test it. The tree describes patterns. It cannot prove that raising commitment causes more posting, and those people all already followed a brand in one country. If I ran it again, I would add AI-made brand content to the mix and see whether committed followers respond to it the same way.
Marketing in 3 steps
Found this useful?
Marketing in 3 is free and ad-free, with new issues every Monday and Thursday. A coffee keeps it that way.
In this study, four beliefs about the brand did: commitment, recognition, familiarity and favourable associations, with commitment the first and strongest split (Schivinski, 2021).
Brand passionates, brand fans, low engaged followers, brand curious followers and prospect brand fans. Passionates and fans together made up more than eight in ten of the people surveyed.
Committed followers. Fans who know the brand well were consistent contributors, and committed followers who were still learning the range had the highest share of heavy creators.
The study predates generative AI, so that part is my view rather than a finding. AI shapes what people see, but whether they act still depends on how they feel about the brand.
Yes. The brand equity and engagement items are listed on the paper page, together with a free copy of the accepted manuscript.
Journal of the Academy of Marketing Science · 2026
Pauwels & Aksehirli
A review that updates a well-known study of free word of mouth by setting it beside paid word of mouth, and online talk beside offline talk. It turns fifteen years of research into seven practical questions, from who and why to metrics and methods, and notes where AI and virtual environments change the picture.
Journal of Research in Interactive Marketing · 2026
Freitas
Comparing 204 matched pairs of AI-run and human-run Instagram accounts, this study finds the same engagement rates and likes, and no gap in comments. AI accounts did gain more followers per post, so synthetic creators help reach, while real conversation stays scarce for both.
Schivinski, B. (2021). Eliciting brand-related social media engagement: A conditional inference tree framework. Journal of Business Research, 130, 594-602. https://doi.org/10.1016/j.jbusres.2019.08.045
This brief is written for practitioners from the published article. Numbers are as reported by the authors; read the article for full methods.
More issues on Social media · UGC · AI
Posts carrying TikTok's AI label collected 7% to 8% fewer likes at the same reach. Not because people distrust AI. Because the label says the creator did not try.
A meta-analysis of 210 experiments finds virtual influencers win on novelty, lose on credibility, and that how human they look decides which one matters.
A detailed "this person does not exist" label cut the sense of being manipulated and raised the sense of artificiality. The two effects cancelled out, except among younger viewers.
Bruno Schivinski is Associate Professor of Marketing at Gdańsk University of Technology. He studies how consumers engage with brands in digital and social media.
