The ICDS Methodological Series has progressively constructed the analytical foundation for understanding how brand media peaks translate into changes in consumer behavior. Article 1 established the baseline correlation between media peaks and brand search activity. Article 2 demonstrated that expressive intensity — the share of evaluative mentions in total peak volume — is a stronger predictor of consumer response than peak volume alone. Article 3 introduced the Balance of Expression variable as a tool for characterizing the internal composition of evaluative content within a peak. Article 4 integrated these findings into a five-stage Signal Chain model from media event to purchase decision.

The present article adds this theoretical-methodological cycle with the new empirical results. The FMCG corpus was selected as the new testing ground precisely because the FMCG sector spans a wide range of product categories — from impulse snacks to premium cosmetics — and thus provides natural variation on the key moderating variable: perceived purchase risk.

However, before proceeding to hypothesis testing, a preliminary step proved analytically significant in its own right: the empirical verification and partial revision of the assignment of FMCG product segments to the ICDS typology. We begin with this step.

The ICDS typology of purchase behavior was first introduced in Article 2 of the series and has since served as the structural moderating variable across all subsequent research. It organizes consumer categories along a single, analytically powerful dimension: perceived purchase risk. This risk has two components — financial (price and the cost of a wrong choice) and psychosocial (the significance of the purchase for self-perception and social positioning). The higher the combined risk, the more deliberate cognitive processing the consumer applies to brand information before making a decision.

Throughout the ICDS series, this typology has functioned as a moderating variable: the same media stimulus was expected to produce qualitatively different behavioral responses depending on the category type. This prediction found support in Article 2, where Cautious categories showed significant growth in transactional search queries at t+1 under conditions of high expressive intensity, while Spontaneous categories responded immediately without durable aftereffect.

One caveat has been present from the outset: the assignment of specific product segments to category types rests on expert judgment about the modal purchase occasion. Alcoholic beverages, for example, can be both Spontaneous (a beer at a bar) and Cautious (a collector's whisky). The original classification reflected the most typical scenario for the brands in the corpus. The present article applies an empirical test of that assignment for the first time.

To validate the original assignment, we used the behavioral fingerprint of each product category — a set of metrics characterizing the temporal profile of search response to a media peak. The logic is straightforward: if the typology is correct, Cautious categories should exhibit a delayed and durable response (search interest growing from t0 to t+1), while Spontaneous categories should show an immediate spike followed by rapid decay.

Three diagnostic metrics were applied: the difference between the normalized search index at t+1 and t0 (positive value = sustained or growing response); the Spearman correlation between the tonality index and search response (high correlation = sensitivity to the content structure of the media field, characteristic of central-route processing); and the proportion of brands within a category whose search index at t+1 exceeds t0.

Confectionery (Ferrero Rocher, Ben & Jerry's, Lindt): the most significant departure from the original classification. Mean search response continues to grow by 28 index points from t0 to t+1 — the most pronounced delayed-response pattern in the entire corpus. The economic rationale is clear: these are premium and gift-oriented confectionery brands where the purchase carries significant psychosocial risk (a gift is a public signal about taste and status). This segment is reclassified as Cautious.

Food Products: a sharp t0 spike followed by a steep decline at t+1 (differential of -38.6 points) — the behavioral signature of Spontaneous processing. The category is internally heterogeneous: brands involved in crises or viral trends (Kellogg's, Indomie) behave as classic Spontaneous; Korean FMCG brands with longer-term cultural penetration (Nongshim) show more durable responses.

Alcoholic Beverages: stable, near-zero response without sharp spikes or drops — the behavioral fingerprint of a Routine category rather than a Spontaneous one. Most brands in the corpus (Heineken, Budweiser, Brahma) are positioned as social rather than impulse purchases.

Unlike previous articles in the series, the present study uses three independent search indices available in the corpus: the brand index (search demand for the specific brand), the category index (search demand for the product category as a whole), and the competitor index (search demand for the main competitor in the same period). All indices are normalized to the baseline period t-1 = 100.

The three-level framework resolves a key methodological problem: separating the brand-specific effect of a media peak from the general category uplift. The normalized index (brand minus category) reflects how much a brand outperforms or underperforms its market context — this is the primary dependent variable in hypothesis testing.

A methodologically important finding emerged at the validation stage: the brand and category indices are virtually uncorrelated (Spearman rho = +0.10, p = 0.41). A brand's media peak does not lift the entire category — it creates an isolated, brand-specific search impulse. This constitutes an independent empirical result with direct implications for how media effectiveness should be measured.

The Balance of Expression variable measures the composition of evaluative content within a media peak. It is defined as the ratio of positive to negative evaluative mentions within the expressive subset of a peak (excluding neutral mentions). The ICDS classification uses the following scale:

After applying the corrected category assignment and transitioning to the normalized search index, the correlation picture becomes analytically clear.

All metrics in Cautious categories show significant positive correlations — both at the immediate and delayed measurement points. In Spontaneous and Routine categories, no correlation reaches statistical significance. This demarcation is the central result of the study.

Assessment of the durability of search response — through the normalized index difference t+1 minus t0 — reveals a qualitative difference across category types:

Critically, durability in Cautious categories is observed regardless of the level of Balance of Expression. This means that category membership determines the type of behavioral response — delayed-response — while the Balance of Expression determines its intensity. In Spontaneous categories, no balance structure produces a durable search effect.

Analysis of the competitor index distinguishes three market-effect types of media peaks. Category expansion (both brand and competitor indices rise) is observed in 26% of cases.

Zero-sum win (focal brand rises, competitor does not) occurs in 23% of cases. Zero-sum loss (competitor rises, focal brand does not) occurs in 17% of cases. Both declining — in 34%.

In Cautious categories, brands significantly more often win against competitors (28%) than lose (11%), and this tendency is significantly correlated with the Balance of Expression (rho = +0.59, p = 0.010). In Routine categories, peaks are competitively neutral (23%/23%). This means that in Cautious categories, a media peak with a pronounced positive disbalance operates as a share-redistribution instrument, not merely a general category stimulator.

The behavioral differences documented above are theoretically predicted by the Elaboration Likelihood Model (Petty & Cacioppo, 1986). In Cautious categories, consumers are motivated to process brand information carefully: the cost of a wrong choice is high and the decision carries social weight. A media field with a pronounced positive disbalance signals not the absence of objections, but that competing arguments have been considered and weighed — this is structurally analogous to a refutational two-sided message in Allen's (1991) taxonomy.

For Spontaneous categories, this mechanism does not apply: consumers do not seek grounds for a decision — they respond to an emotional signal immediately. The structural composition of that signal (disbalance level) has virtually no effect on response intensity: what matters is the presence of a peak, not its valence architecture.

The independence of brand and category search indices deserves special attention. It confirms that in Cautious Categories a media peak activates specific interest in the focal brand, not diffuse curiosity about the category. This supports the central thesis of the ICDS framework: an expressive media peak initiates purposeful information search, not general category-level engagement.

One category-specific pattern in Spontaneous segments warrants separate attention. Under conditions of high positive disbalance in the media field, a weak tendency toward growth in the category-level search index emerges at t+1 — that is, after the brand-specific search peak has already subsided (Spearman rho = −0.286 between balance and category index growth at t+1, tendency at p = 0.17). In other words, a homogeneously positive media field in Spontaneous categories appears to generate not so much sustained interest in the focal brand as a diffuse category appetite: a consumer saturated with positive signals about a snack or beverage begins searching within the category — but not necessarily for the brand that originated the peak. This observation, while not yet reaching conventional significance thresholds, is substantively consistent with the peripheral-route processing logic: an affective signal activates desire without forming brand-specific preference.

Absence of informational/transactional search split. The current corpus contains only aggregate brand search indices. The series methodology (Article 5) requires separate metrics for informational ("[brand] reviews") and transactional ("[brand] buy") queries. Obtaining these through KeywordTool Pro is the priority next step — particularly for Cautious categories, where the temporal lag between informational and transactional response is a structural variable in the model.

Sample size. 18 observations in the Cautious group are sufficient for initial analysis but inadequate for reliable interaction testing. Corpus expansion is a prerequisite for publication of definitive results.

Observational design. The Balance of Expression in a media peak is not experimentally manipulated. Causal interpretation requires caution: brands with inherently higher audience loyalty may generate peaks with more positive balance because of pre-existing attitudes, not because disbalance causes attitudinal change. The temporal lag method (analyzing t+1 while controlling for t0) partially mitigates this threat but does not eliminate it entirely.

The empirically supported reclassification — moving confectionery to Cautious and food products to Spontaneous — not only improved the model's predictive power but is itself a substantive result: it demonstrates that perceived purchase risk is determined not by the product group as such, but by the specific brand positioning within it.

The aggregate of results supports hypotheses H2 and H3 from Article 5: the moderating role of category type is confirmed, and brand-specific, durable search response in Cautious categories has been documented. Hypothesis H1 — that Very Strong Positive Disbalance is optimal for Spontaneous categories — did not receive support: in these categories, balance structure is not a significant predictor of any search metric.

The broader implication for communication strategy is precise: the assumption that maximizing positive sentiment is universally desirable is empirically correct only for Spontaneous categories, where the peripheral processing route makes valence the primary cue. For Cautious categories, what matters is the magnitude and durability of the positive signal relative to competing voices — and this is exactly what a Strong Positive Disbalance (25/75) provides: clear dominance without the cognitive vacuity of a homogeneous environment.