Semantic targeting is a type of target group approach used in online advertising. Its key feature is that ads are displayed in a semantically relevant environment, thus avoiding irrelevant and possibly damaging misplacement.
Terms which have a thematic reference to the overall content of a website are semantically relevant. This includes synonyms and ambiguous terms as well as different linguistic relationships between individual words of a website. The process of the semantic alignment is supposed to determine meanings and contexts of individual words and the entire website content in order to place ads where they would be suitable and to avoid incorrect associations.
The concept of the semantic web can be considered as a template for semantic targeting. On the one hand, various types of data are distinguished and made usable in a platform-independent manner. On the other hand, this data is categorized in a certain way so that machine formats can be read and interpreted. The principle was transferred to the field of display marketing.
The scientific preparation for this procedure comes from an international research group led by linguist Prof. Dr. David Crystal from the University of Wales. His team created the algorithms during a 12-year research effort. Two specific integrated processes are integral to semantic targeting, lexical analysis of a website and their categorization into a thematically appropriate context.
In semantic targeting, the following basically is done: Using semantic (meaning-related) data and metadata, different resources (such as websites) are classified and placed in a larger thematic context. The linking of concepts within the natural language is modeled using text corpora and represented by algorithms.
This data allows the automatic classification of websites and databases into the context of the natural, spoken language. It therefore attempts to make the meaning and context of the terms on a website machine-readable in order to classify the website in a specific context. The context is then used to place relevant ads exactly where they have a thematic reference and are not inappropriate.
Technically speaking, the ad server does a lot of the work. Within milliseconds, it scans the content of a website and assigns it to a category. However, complex algorithms are required for these tasks, which can at least partly understand the natural language or model meanings. The ad server takes ontologies and taxonomies of language and terms which are filed in a database in order to assign the content of a website to the existing categories of ontology and taxonomy.
Each targeting option has advantages and disadvantages. Behavioral targeting requires cookies to be set in order to be able to track user behavior. This may result in possible privacy and data protection issues. Contextual targeting, on the other hand, focuses on identifying terms to advertise using keywords. The result may be that ads appear on websites that are a bad neighborhood for the marketer (see Bad Neighborhood).
A further development is semantic targeting. No cookies are set and the content of a website is included in the process. Marketers hope to get various advantages with semantic targeting:
Semantic advertising is associated with important current topics in e-commerce including reputation management, privacy, data protection, and performance marketing. Semantic targeting of ads can avoid damage to a company’s reputation. For example, you can prevent a last-minute display ad for a certain destination country from being placed on a news portal that is just reporting on an earthquake having taken place in that country.
Moreover, concerns regarding the privacy and data protection of users can be eliminated since no cookies are set or personal data collected during the process. Last but not least, the performance of advertising campaigns on the Internet plays an increasingly important role. This is because semantic data allows the precise control of campaigns, especially in the case of large campaigns with many page impressions.