David Aaker's Brand Equity Capital Model

• brand awareness;
• brand loyalty;
• perceived quality;
• brand associations, including individuality;
• other brand assets — patents, distribution channels, trademarks.

Later, Aaker expanded the model to ten factors of brand equity: distinctive traits, consumer satisfaction, quality, market leadership, perceived benefit, individuality, associations, awareness, market share, price, and geographic coverage. Separately, he proposed the 'loyalty pyramid' — five levels of consumer relationship to a brand: from full loyalty to consumers without any attachment, ready to switch to any other product without regret.

Aaker versus Keller: two different questions, not competing theories

Comparisons between Aaker's model and Keller's model (1993) are popular in articles, but most often such comparisons reduce to listing differences without answering the main question — which model to choose for a specific task. The difference lies in the very nature of the approach: Aaker describes a brand as a set of company assets, while Keller sees it as the result of cognitive processes in the consumer's mind. His four-level pyramid goes from salience (recognizability) through performance and imagery to judgments and feelings, and ends with resonance — a deep connection with the brand.

In practice, it makes sense to use both models together: Aaker provides accountability of 'what we have,' while Keller offers diagnostics of 'why it works or doesn't work.' Using one model as a substitute for the other is a mistake: they answer different questions.

How to measure each component in practice

The problem with most Russian-language materials is that they name components but do not say how to measure them. Below is a working set of metrics used in tracking studies:

• Awareness — top-of-mind awareness, unaided and aided recall, share of mentions in spontaneous responses;
• Loyalty — Net Promoter Score, repeat purchases, price premium that consumers are willing to pay;
• Perceived Quality — comparative scales of 'worse/better than competitor' for key category attributes;
• Associations — free associations, semantic differential, perception maps;
• Other Assets — number of patents, exclusive distribution channels, legal protection of trademark.

For quantitative measurements, 10-point Likert scales are usually applied with reliability checking via Cronbach's alpha coefficient — an acceptable threshold is considered to be a value no lower than 0,7. For time tracking, the recommended minimum sample size is 200 respondents per measurement wave; otherwise, changes in indicators will appear to be statistical noise rather than real dynamics.

What does not work: criticism and limitations of the model

Academic research confirms what practitioners encounter when attempting to separate the model components into individual questionnaire items. The study by Christodoulides et al. (2013), conducted on samples in Germany, Great Britain, and Greece, revealed a serious discriminant validity issue: brand awareness and brand associations are statistically collinear — respondents do not actually distinguish these constructs in their answers. The same applies to the pair 'associations — perceived quality'.

The problem is exacerbated for service and online brands, where the consumer has no physical product to grasp a clear quality attribute. For a product category such as soap or household appliances, it is easier to separate 'quality' and 'associations' — there are measurable product characteristics. For a bank, insurance company, or marketplace, these boundaries become blurred: service quality is essentially a set of associations.

A separate weakness of the model is the lack of an integral formula that would link the five components into a single index. Aaker deliberately did not propose weighting coefficients, leaving aggregation to the researcher's conscience, which makes the model flexible but poorly comparable across industries.

Ivory and other examples: where the model shows itself honestly

A classic example of Ivory — is not just a historical anecdote. This is a rare case where all five components are documentedly traced at a distance of more than century. Slogans "99 44/100% pure" and "it floats" with a budget of 11 000 dollars in 1882 year created stable awareness; associations with infants and purity held for decades; white color and softness formed perceived quality. And when in 1941 year a competitor Swan entered the market with a direct attack — positioning "first really new floating soap" — consumer loyalty to Ivory did not allow the new product to take a notable share. This is a teaching example precisely because the product category is simple and stable.

The model works less predictable with service and digital brands: with an online platform "quality" is blurred between speed of operation, interface and support service, and "associations" are formed not by advertising but by reviews and recommendation algorithms. This is exactly where discriminant validity collapses fastest, and blind application of the classic Aaker questionnaire gives unreadable data.

Why perceived quality is not a routine component, but a mediator

In most empirical models, perceived quality behaves not as an equal fifth component, but as a mediating variable: awareness and associations influence loyalty mainly through perception of quality, not directly. Consumers rarely formulate loyalty with the phrase "I know this brand" — more often it sounds like "this product is more reliable than others", and this judgment of quality includes both recognizability and accumulated associations. Practical conclusion: if budget for research is limited, quality should be measured first and more thoroughly than others — it gives the greatest explanation of variation in loyalty.