A diagnostic framework that classifies a brand by the convergence state of the consumer's prediction model, distinguishing four states that call for opposite interventions.
A brand can be perfectly recognisable and still be one whose every move the consumer already anticipates.
Why recognition is not enoughDistinctive-asset and brand-recognition measures, including those used in the Ehrenberg-Bass tradition, establish whether a consumer can identify a brand from a cue and attribute it correctly. That measure is necessary, and it leaves over-convergence invisible. A brand can score at the ceiling on recognition while the consumer's prediction model has fully closed: every move known in advance, forecast with confidence, and no longer generating the prediction error that sustains attention and refreshes memory.
The Brand Harmonic Model reads a second question that recognition cannot answer: can the consumer forecast what the brand will do next, and with what certainty? The joint pattern of recognition and predictability is what surfaces a condition a recognition measure conceals.
The consumer holds a strong model yet cannot fully anticipate the brand. Anticipation and encoding are sustained.
Intervention: hold the current balance.
The model has closed. The next move is known in advance and forecast with confidence; attentional persistence is falling.
Intervention: calibrated deviation at the execution level.
No stable expectation has yet formed. Added surprise reads as noise, not signal.
Intervention: increased consistency.
An execution sustaining optimal prediction error in one population reads as over-familiar or as noise in another whose exposure history differs.
Intervention: adjust execution complexity to each population's baseline.
Two brands showing the same decline on a conventional tracker receive opposite prescriptions.
The instruction to vary execution and introduce surprise is sound for an over-converged brand and damaging for an under-built one, whose model is not yet stable enough to register variation as anything other than noise. A conventional tracker shows both brands the same falling line. The diagnosis separates them, and the cost of confusing them is concrete: the wrong intervention accelerates the decline it was meant to reverse.
The framework's instrument, the Brand Prediction Task, assesses predictability directly, placing a brand on the two axes and returning the state that governs which prescription applies. The procedure is developed for institutional and licensing partners.
Through the 1990s the Burberry check became one of the most recognisable visual assets in fashion, licensed across the product range until it was both universally known and universally anticipated. On a distinctive-asset measure it scored at the ceiling, the textbook ideal that asset-based guidance would advise a brand to exploit further. Yet at that point of maximal recognition the brand's value collapsed: a signal everyone could decode had lost the prediction value on which a luxury identity depends.
The recovery inverted the asset-based prescription. Rather than deploying the famous asset more widely, the company restricted it, re-establishing the trench coat as the structural anchor. The framework reads this as the restoration of prediction error to an over-converged signal, a move recognition metrics could not have motivated, because recognition was never the problem.
Offered as illustration, not as proof. The case motivates the diagnosis; the empirical programme tests it.
The Brand Harmonic Model is introduced in The Prediction Engine (Lebedeva-Gule, 2026). Its empirical validation, via the Brand Prediction Task applied to real brands at different lifecycle stages, is in development with institutional partners.
The Brand Prediction Task is designed to be validated and deployed with research and brand-measurement partners. Enquiries are welcome.
Research enquiry