{{title}}
{{textarea}}
{{task}}
{{select}}
{{team}}
{{input}}
{{solution}}
{{result}}
ICDS Research Brief Consumer Behaviour
How Purchase Risk Perception Determines the Influence of Social Media on Consumer Behaviour
Across 66 brands in 10 FMCG categories, the balance of tonality within a media peak predicts consumer behaviour — but only where the consumer perceives the purchase as risky. Here is how it works, and why it changes the logic of communication strategy.
VM
Vadim Matyushkin Behavioral Scientist & Sociologist · ICDS
Imagine two marketing managers. One works with a snack brand, the other with a smartphone brand. Both have just received identical media monitoring data: a media peak, 70% positive mentions, 30% negative. Both are asking the same question: is this good or bad for sales?
The correct answer is different for each of them. And the difference is not in the quality of the campaign. The difference lies in how consumers process brand information depending on how risky they perceive the purchase to be. This is the central conclusion of a two-year ICDS research programme across a corpus of 66 FMCG brands in 26 countries.
66FMCG brands26 countries · 10 categories
2 yrsOf observationYouScan + Google Trends
ρ = +0.75BoE → search (Cautious)p < 0.001 · durable · at t+1
0Effect in SpontaneousNo prediction in snacks, soft drinks, basic food
01
Why Consumers Think Differently in Different Categories
When a person buys a bag of crisps, they are not thinking — or rather, they are thinking differently: quickly, on the basis of emotion, availability, and habit. Psychologists call this the peripheral route to information processing. You like the brand — you buy it. The packaging is bright — you buy it. Someone wrote something positive on Instagram — you noticed, and you might try it sometime.
When the same person buys a smartphone, they do think. They read reviews, compare options, weigh pros and cons. Psychologists call this the central route. The consumer actively seeks arguments, evaluates their credibility, and pays attention to the balance of positive and negative signals.
This distinction is not new to persuasion psychology. The Elaboration Likelihood Model (Petty & Cacioppo, 1986) described it four decades ago. What is new is that it measurably manifests in social media and search behaviour data, and that it can be predicted through the structure of a media peak.
| Type | Example categories | Perceived risk | How brand information is processed |
|---|---|---|---|
| Spontaneous | Snacks, soft drinks, chewing gum, basic food products | Low | Peripheral route: emotion and availability. Detailed sentiment analysis is not activated. |
| Routine | Household cleaning products, basic cosmetics, pet food | Low–medium | Habit plus periodic review. A media peak may disrupt the habit but does not sustain attention for long. |
| Cautious | Smartphones, premium cosmetics, confectionery*, vitamins, financial products | High | Central route: the buyer actively gathers and weighs information. Sentiment is read analytically. |
* Sweets and chocolate were initially classified as Spontaneous. Empirical analysis of the corpus showed that consumer behaviour in the premium confectionery segment matches the pattern of Cautious categories. The classification was revised on the basis of data, not subjective judgement.
02
What Balance of Expression Is — and Why It Is Not Simply Sentiment
Media monitoring conventionally tracks tonality: what percentage of mentions are positive, what percentage negative. This is useful, but it is not what predicts consumer behaviour.
The key variable is Balance of Expression: the structure of evaluative content within a media peak. Not how many mentions there are in total, but what the ratio of positive to negative assessments is among those mentions that actually evaluate anything — the expressive subset.
An important distinction
96.5% of all brand mentions in social media are neutral. They evaluate nothing — they inform, quote, share. Balance of Expression is calculated only within the remaining 3.5% — the evaluative mentions. That is where the signal that the cautious buyer responds to is located.
Expressive mentions divide into four balance levels — from near-uniformly positive (>90%) to roughly equal distribution (≈50/50). And it is precisely this balance — not total volume, not average sentiment — that proved to be a predictor of behaviour, but only in Cautious categories.
| Balance level | Pos./neg. ratio | Spontaneous categories | Cautious categories |
|---|---|---|---|
| Very strong positive disbalance | >90% positive | Neutral or slightly worse | Weak effect: market perceives it as ‘advertising noise’ |
| Strong positive disbalance | 75–90% positive | Neutral | Optimal: ρ = +0.75*** · t+1 effect is durable |
| Moderate positive disbalance | 60–75% positive | Neutral | Good: central route activated, search grows |
| Balanced (≈50/50) | 40–60% positive | No effect | Mixed results: high engagement but no clear signal |
03
Three Key Findings
1
Balance of Expression predicts search behaviour exclusively in Cautious categories.
In Cautious categories (N=18 brands), the correlation between Balance of Expression and brand search growth was ρ = +0.75 (p < 0.001). The effect manifests at t+1 — the month after the peak, not at the moment of the peak itself. This is the behavioural signature of the central route: the consumer first gathers information (growth in informational search at t0), then moves to purchase intent (growth in transactional search at t+1). In Spontaneous and Routine categories, no analogous correlation was found at any time horizon.
2
The effect is brand-specific — not a reflection of general category dynamics.
The critical question when interpreting search data: is the specific brand’s search growing, or is the whole market rising? Analysis across three indices — brand, category, competitor — showed that brand search growth in Cautious categories after a peak is not explained by general category growth and does not coincide with competitor movements. This is not a market tide. It is a specific audience response to the media peak of a particular brand.
3
The effect is durable over time: search grows after the peak, not only at the moment of it.
In 72% of brands from Cautious categories, search interest continued to grow after the mention peak had already passed. This fundamentally differs from the Spontaneous category pattern, where the interest spike coincides with the peak and decays rapidly. The durability of the effect means that a brand has a time window after the media event — four to eight weeks — during which the consumer is actively gathering information and making a decision.
04
Why 90% Positive Is Not Always the Best Strategy
The intuitive takeaway from everything above: maximise the share of positive mentions. More positive means better. The data say otherwise. In Cautious categories, a near-uniformly positive environment (>90% positive) works worse than an environment with clear positive dominance but the presence of criticism (75–90%). Why?
Because the cautious buyer — the one who reads reviews analytically — interprets the absence of negative mentions as a sign of a dishonest information field. An overly smooth media backdrop activates scepticism. The presence of counterarguments — in the language of classical persuasion theory (Allen, 1991) — signals credibility and encourages deeper engagement.
The homogeneity paradox
For a consumer in a Cautious category, 25% negative mentions alongside 75% positive is a signal: ‘people are genuinely discussing this brand, including its weaknesses.’ This activates the central processing route: the buyer starts searching, reading, comparing.
But 95% positive is a signal: ‘this is advertising.’ The central route is not activated. Search behaviour does not change.
05
What This Means for Communication Strategy
The conclusion is practical and precise. Not universal — category-dependent.
1
First, determine which type your category belongs to.
If your product is a spontaneous purchase (snacks, basic food products, budget cosmetics): maximise positive reach, minimise criticism, create simple emotional associations. Balance of Expression is not a priority metric for you.
2
If your product is a cautious purchase: manage the balance, not the level.
The goal is not the maximum share of positive content, but clear positive dominance with the presence of a genuine critical voice. The data-driven target: 75–90% of positive evaluative mentions. Below this threshold — no clear signal. Above it — the media environment is perceived as inauthentic.
3
Use the post-peak window — it is real and measurable.
In Cautious categories, search activity continues to grow for four to eight weeks after the media peak. This is not decay — it is the purchasing process in progress. Content to address informational search (reviews, comparisons, detailed feature descriptions) should be prepared before the campaign launches, not created in response to an unexpected surge in queries.
Methodology note
How the study was built
The research is based on Articles 5 and 6 of the ICDS Methodological Series. Corpus: N=66 FMCG brands, 26 countries, 10 product categories; data from YouScan + Google Trends. Balance of Expression is computed as the ratio of positive to negative mentions within the expressive subset (all evaluative mentions). Search indices — brand, category, competitor — are normalised for cross-brand comparison. Statistics: Spearman rank correlation, normalised metrics. N=18 in the Cautious group is sufficient for initial analysis but not for robust testing of interactions; expansion of the corpus is planned.
Limitation. Balance of Expression is not experimentally manipulated. Causal interpretation requires caution — brands with high audience loyalty may generate more positive peaks due to pre-existing attitudes rather than the reverse. The temporal-lag method partially mitigates this threat.
Full research articles available at icds.institute.
VM
About the author
Vadim Matyushkin
Behavioral Scientist & Sociologist · ICDS
Vadim Matyushkin is a psychologist and sociologist with close to twenty years of research into digital behavior, trust, and information dynamics, and a researcher behind the Institute of Communication and Data Science (ICDS). His work has supported organizations including Coca-Cola, PepsiCo, Mars, Danone, Nestlé, and Bayer in moving from self-reported survey data toward direct behavioral evidence.
ICDS — Institute of Communication and Data Science is an independent research institute focused on understanding how trust, behavior, and reputation are formed in digital environments. ICDS operates as an intellectual institution — not an agency, not a platform, and not an educational provider.