ICDS · Open Research Series · Methodology Track
The Path to a Responsible Index
Media-Driven Brand Impact Index (MDBI) — First Edition · Pilot computation across fifteen brands · July 2026
This paper sets out how the Media-Driven Brand Impact Index measures the effect of media on consumer behaviour. It follows the index through its first stage: the reasoning behind it, the initial computation across fifteen brands, the questions that computation raises, and the direction of the next edition.
It is a methodology paper, not a final ranking.
It is a methodology paper, not a final ranking.
Key Findings
The first computation covers fifteen brands with complete data. It already produces a working diagnosis for each, even before category and country adjustments are finalised.
Psychologically, people trust those who are capable of self-irony and honesty. According to the Journal of Consumer Psychology, acknowledging a minor flaw in advertising increases trust by 32%.
The index resolves each brand into one of four behavioural profiles after a media peak: full-funnel conversion, awareness without conversion, retention, and media presence without behavioural response.
The next edition moves the denominator from volume of conversation to audience reached, introduces a common scale, and extends the computation across the full corpus.
Huawei illustrates the fourth profile: an active, high-intensity media presence that does not translate into search behaviour — a reading no reputation or brand-value index surfaces.
01 / The Question the Index Answers
Most current instruments answer one question: what audiences think of a brand. That view is assembled from surveys, expert assessment, and financial reporting, and it describes a state — reputation, value, awareness. A state, however, is not yet an action.
A brand's value lies not in being thought well of, but in whether its public communication moves people to act.
MDBI measures the action. It asks whether a brand's communication works operationally — whether attention becomes observable consumer response. Not whether a brand is admired, but whether it moves behaviour.
A measure of brands is only as useful as it is verifiable — which is why every limit in this pilot is stated in the section that follows, not left for the reader to discover.
A measure of brands is only as useful as it is verifiable — which is why every limit in this pilot is stated in the section that follows, not left for the reader to discover.
02 / The Reasoning, in Plain Terms
To measure action rather than opinion, the index needs a chain of events that can be observed.
A media event is a stimulus
The index does not count everything said about a brand. It looks for a media peak — the moment a brand becomes distinctly louder than usual: not a minor fluctuation, but a clear surge against a quiet baseline. Such a peak is treated as a stimulus with an onset, a magnitude, and a consequence.
The response has two stages
After the stimulus, behaviour divides. First a person wants to understand — searching for a review, a comparison, an explanation. This is the cognitive stage. Then, if interest matures, the same person wants to buy — searching for a price, a retailer, a way to order. This is the transactional stage.
An adjustment for where and what
An identical response means different things under different conditions. In a market where nearly all search runs through Google, a surge in queries is visible in full; where audiences search through Yandex, Baidu, or Naver, part of the response falls outside the data. The index therefore carries a country adjustment.
Category matters as well. A smartphone is considered over weeks; a lipstick is bought on impulse. The weight given to the cognitive and transactional stages is set not by assumption but by how durable interest is within each category.
Category matters as well. A smartphone is considered over weeks; a lipstick is bought on impulse. The weight given to the cognitive and transactional stages is set not by assumption but by how durable interest is within each category.
The weights are not chosen to fit a theory. They follow from how long interest lives in each category.
Two measures follow directly:
SCS — how far a media peak turns into the wish to understand (cognitive search).
CAS — how far that interest reaches the wish to buy (transactional search).
CAS — how far that interest reaches the wish to buy (transactional search).
A positive figure means the media peak becomes behavioural response; a negative one, that the brand generates volume without response.
03.1 / Initial values by category
03 / The First Computation
The index has been computed for the fifteen brands with complete data. The values are not yet placed on a common scale, so a brand in one category cannot yet be compared directly with a brand in another.
What has been computed
15 brands · 31 brand-peak observations · three categories (Considered, Habitual, Impulse).
The Risky category — finance, pharmaceuticals — sits outside this pilot; the corpus does not yet hold enough of it for a sound calibration.
The Risky category — finance, pharmaceuticals — sits outside this pilot; the corpus does not yet hold enough of it for a sound calibration.
Index (raw value)
Brand
cATEGORY
−2
Miu Miu
+10
Rare Beauty
+13
Rhode
Impulse purchase
+29
Kylie Cosmetics
−30
Medicube
−4
GAP
+1
VT Cosmetics
+6
L'Oréal (UD)
Habitual purchase
+45
American Eagle
−25
Huawei
−3
NINJA
+2
Xiaomi
+77
Dyson
+158
Apple
Considered purchase
based on public data
Values are shown as computed, within category; cross-category comparison is not yet valid at this stage.
03.2 / What the computation produces
A raw figure is not yet a ranking. What the first computation produces is a diagnosis: the pair of signals, understanding and purchase, resolves each brand into one of four behavioural profiles.
Full-Funnel Conversion
Apple, Dyson
The peak moves the whole path: both understanding and purchase. Attention converts to action end to end.
Awareness Without Conversion
Rare Beauty
Interest is high but does not monetise. The brand gathers attention and conversation, not purchase intent.
Retention Brand
Kylie Cosmetics
The loyal buyer skips the understanding stage and moves straight to purchase. The brand mobilises its own, rather than recruiting new buyers.
Media Presence, No Response
Huawei, Medicube
Media volume without behavioural response. Visible to monitoring systems, present in the press, yet inert at the level of consumer action.
03.3 / Huawei
Huawei keeps an active, high-intensity media presence: its peaks are visible, expressive, and widely reported. Those peaks do not become cognitive search — audiences do not go on to find out more. The index turns negative.
Neither a survey nor a brand-value estimate surfaces this; on those measures Huawei reads as strong. MDBI registers the distance between how loudly a brand sounds and whether that sound moves anything. That distance is the reading the index exists to provide.
based on public data
An active media presence that does not convert into behaviour is precisely the reading MDBI provides and no existing index does.
04 / Open Questions from the First Computation
Corpus coverage
The full computation covers the fifteen brands for which the data is complete. This is a pilot, and the corpus grows from here; extending it is the first task ahead of the next edition and will roughly double the sample, giving the calibration a firmer base.
The denominator measures production, not reach
At present the denominator is the volume of conversation — the number of mentions. A post seen by two people and a post seen by two million enter the computation alike. Reach, not volume, is what governs behaviour.
Quiet peaks strain the arithmetic
The index divides response by the magnitude of the peak. Where a peak is barely perceptible, that magnitude approaches zero and the division becomes unstable, producing inflated values. A technical threshold is already in place: a peak counts only when it exceeds its baseline by at least five per cent.
The scale is not yet common
Raw values range from −30 to +158. Apple, in the Considered category, and Kylie, in the Impulse category, therefore live on different scales and cannot be compared directly. A common measure, within and across categories, is required before a cross-category ranking is meaningful. For now the paper reports diagnoses by profile rather than a single table of leaders.
05 / The Next Edition
The questions in the previous section set the direction of the work. The next edition of the index moves to a revised frame.
What changes
Values are placed on a comparable footing, and the first valid cross-category ranking becomes possible.
A common scale
The computation moves from how much was said to how many were reached: each peak weighted by the audience it actually reached.
From volume to reach
When different platforms tell materially different stories about a brand, the signal blurs. The revised frame accounts for this coherence as a separate coefficient.
An adjustment for dispersion
The computation extends across every brand in the corpus, not only the fifteen with data already assembled.
The full corpus
Why the frame is introduced in stages
The reach-weighted frame depends on data and on retrospective validation that the corpus does not yet hold in full. Introducing it before that base is in place would put uncalibrated coefficients into circulation as if they were measurements. The index is therefore built in stages: the present computation stands on its own, and the next extends it.
Returning to the computation on fresh data is not a correction to the first edition but the substance of a longitudinal index. Each recomputation is a new point in a brand’s record, and the movement between points is a finding in its own right.
What follows
The first edition is an open pilot with diagnoses for fifteen brands. The next is the computation in the revised frame, across the extended corpus, on a common scale, with the first cross-category ranking. The path is set out in advance so that the reader follows the work as it develops rather than receiving a finished verdict.
The distance between a brand's value today and its reach-weighted value in the next edition is itself a reading — the first measured point in a brand's trajectory.
MDBI is not a measure of how much a brand is liked. It is a measure of whether a brand works.
An index in which each value traces to a specific consumer behaviour — not a survey, an expert opinion, or a modelling assumption.