At first glance, evaluating a brand may seem straightforward. However, a deeper and more accurate analysis requires examining multiple attributes and processing vast amounts of data. If this task were carried out solely by humans, without the aid of algorithms, it could take years to gain a clear picture of what consumers truly think about a brand.
Even then, the results would likely be imprecise, since it would be impossible to analyze millions of surveys collected from consumers familiar with the brand.
Artificial Intelligence (AI) represents a real breakthrough in this field. With AI, it is possible not only to evaluate how consumers perceive a brand in far less time but also to measure a range of attributes that help brands understand where to invest to achieve stronger market results. AI conducts Big Data analysis, examining elements such as comments, web texts, and images.
But which attributes matter most in this evaluation?
What Consumers Like and Dislike
One of the most important aspects to identify is what consumers appreciate and what they dislike about a brand’s products or services.
For example, in the case of computers, it could be valuable to determine whether customers value design over innovation, or vice versa. This insight enables brands to make targeted investments that better meet customer expectations and demands.
At MRC Omnichannel, our professionals also measure Net Sentiment—an aggregated metric of customer satisfaction. A negative Net Sentiment signals that a product or service should likely be withdrawn from the market, as it is not well received. To succeed, every product or service a brand offers should achieve a positive Net Sentiment.
Our role is to pinpoint the most and least appreciated attributes, both in absolute terms and in relative terms (for instance, being “the least bad” option). Human interpretation remains essential to give meaning to the insights extracted by the algorithm.
Measuring Customer Loyalty
Another critical attribute analyzed by MRC Omnichannel experts is consumer loyalty. Loyalty measures how likely a customer who purchased a product would buy it again.
It is therefore important to analyze post-purchase consumer behavior. Using algorithms, we can assess whether customers expressed themselves positively or negatively about the product or service after purchase.
Evaluating the Reliability of an Indicator
To ensure an indicator is reliable, several parameters must be considered:
-
Volume of texts analyzed: If the dataset consists of only a few hundred items, the indicator cannot be considered reliable. A reliable analysis typically requires several hundred thousand data points at minimum.
-
Stability of the indicator: If the results fluctuate wildly from one day to the next—for example, showing a brand as highly appreciated one day and despised the next—the indicator is unreliable.
Determining whether an algorithm is effective requires time and pilot studies, often over several months. Some fluctuations are natural, and analysts must be able to understand the reasons behind them.
For instance, sentiment may spike sharply upward or downward due to unforeseen external events. If the CEO of a major company—seen as the sole capable leader—were suddenly killed, consumer sentiment toward the company could plummet overnight. Similarly, scandals such as labor exploitation could negatively impact Net Sentiment. In these cases, the algorithm would still be functioning correctly, as it reflects real shifts in public perception.
What’s Needed for Effective Attribute Analysis
The most accurate results come from combining AI with human evaluation. Human expertise ensures that the algorithm’s output is interpreted correctly and placed in the right context.
By working this way, MRC Omnichannel professionals can determine whether a promotional campaign is successful. If an indicator shows a strong upward spike, it suggests the campaign is positively received by consumers.