research · analytics

AI & Brand Loyalty

Does AI personalization actually build brand loyalty? I ran the numbers.

BusinessAIResearch

MBA research project

A quantitative research study testing whether AI-driven personalization shapes brand preference and loyalty. Built on a 178-person survey, it runs correlation, multiple regression, and ANOVA to separate what actually moves loyalty from what just sounds good in a pitch.

Read the full study (PDF)

n = 178 survey respondents

R² = 0.50 regression fit

r = 0.84 preference → loyalty

5 hypotheses tested

What the data showed

the question

Does AI personalization build loyalty — or just preference?

178-person survey, 5-point Likert
Five hypotheses
Correlation, regression and ANOVA

finding 1

Personalization drives preference

r = 0.61AI personalization → brand preference

finding 2

Preference is what drives loyalty

r = 0.84brand preference → loyalty (the strongest link)

the model

Half the variance, explained

R² = 0.50multiple regression · F = 33.8 · p < 0.001

Trust was not a moderator in a young sample

The question

Everyone says AI personalization builds loyalty. I wanted to test it properly: does personalization actually move brand loyalty, or does it just nudge short-term preference? And is trust the thing that converts one into the other?

The method

A 178-person survey on five-point Likert scales, structured around five hypotheses, with a full statistical workup — descriptive statistics, a Pearson correlation matrix, multiple regression, and one-way ANOVA. Grounded in the marketing-AI literature (Kotler; Huang & Rust), so the constructs weren’t invented on the spot.

What the data said

A clear chain emerged. AI personalization lifts brand preference (r = 0.61), and preference is by far the strongest driver of loyalty (r = 0.84) — stronger than personalization’s direct effect. The model held up: R² ≈ 0.50, F = 33.8, p < 0.001, explaining about half the variance in loyalty.

The part that surprised me

I expected trust to be the moderator — the thing that turns personalization into loyalty. In this young, digitally-native sample it wasn’t significant. For this cohort, relevance seems to matter more than reassurance.

What I took from it

Personalization works — but indirectly, by building preference, not by buying loyalty outright. The real lesson was method: treat a marketing claim like a data problem — state a hypothesis, gather evidence, run the test, and let the numbers settle the argument instead of the loudest voice.