Today when people think of advertising, people usually think of digital advertising. Digital advertising has turned advertising upside down: in the old days, advertising used to be creative, about concepts, visuals and ideas; now, it is about numbers: performance metrics, algorithms, KPIs, or numbers associated to brand campaigns.
If the rise of digital has made budgets more accountable, what are those accountable budgets being spent on? How can we be sure consumers’ behaviour is being influenced by advertising? Often the metrics seem to have replaced the purpose: have we lost the sense of what advertising truly meant? And has the profligacy of metrics and numbers helped us understand what consumers really want?
We think semantics applied to advertising will bolster meaning and help understand intent better. We believe the surge of semantics in the field of advertising will actually come from this numerically-heavy discipline, digital advertising.
What is semantics?
Semantics is an area of linguistics (the study of language) which is devoted to the study of meaning.
Semantics is not about words, it is about the way we use them. One could think search marketers are experts in semantics. But search marketing is programmatic text-based direct advertising. Search marketers do not look for meaning; they want to serve their text ads to the users who are most likely to be in the market. Most advertising data is data obtained from navigations (contextual, behavioural…), but not from the intent leading to the navigation. Search data, if well utilised, can provide meaning.
So, if search data is more powerful at indicating intent than any other form of data, where do you find it?
Where do you find search data on the internet?
Search is not restricted to Google. Very far from it. Most studies show that premium search engines account for around 60 per cent of all searches (for desktop; for mobile, the number is significantly lower). If one aggregates searches on desktops and on tablets and mobiles (where a search engine is not the most commonly used point of entry into navigation), then premium search engines represent just one large source of search. Users doing searches on publishers sites (special-interest, specialist…) are often more engaged and further down the purchase funnel. What is very interesting when one compares premium search engine data with other search data is how different and complementary there are. People search in many different places, but they also search in many different ways.
The difficulty and appeal of unstructured data
Any science of language is about structure. Today the challenge with unlocking the wealth of available search data is the ability to bring structure to the unstructured.
Captify is a search intelligence company. We use search data to recognise intent, then use this intent to apply user-level targeting and serve relevant ads to users who happen to be in the market for the brand. We apply the same techniques to brand campaigns and produce unique and differentiated insights which brand clients cannot find anywhere.
At the beginning, we would capture huge volumes of search data through an extensive, multi-layered and mostly exclusive data network. And we would derive intent from looking at this unstructured data. Inspired in the early stages by search marketing, we would use different match types until we realised that we had to do a trade-off between accuracy (and therefore performance) and scale. By going too restrictive on the match type, we would get great performance but limited scale; by going too broad, we would scale, but performance would go down. And we still had those huge volumes of unstructured data which we were not able to exploit. This is when the penny dropped…
We had to turn the problem upside down. Humans cannot conceive the number of different search journeys that other humans will use to express their intent. Therefore, figuring out the most obvious search paths was bound to be limited and would only let us scratch the surface of the unstructured data gold mine. We had to create a semantic engine, aiming to dynamically understand the meaning of billions of search keywords and phrases in real time whilst identifying billions of connections between those keywords, ie we had to embark on something very ambitious: creating a dynamic ontology.
The benefits of structuring the unstructured
Understanding the intent behind billions of keywords and phrases along with their billions of connections is of tremendous value, as it not only allows to enhance the relevance of advertising, it also transforms the nature of advertising. We are not just running campaigns anymore. With technology based on search data, it is now possible to run brand campaigns whilst at the same time understanding the efficacy and the impact of the campaign. Search data helps identify audiences and search data helps understand the impact on key brand metrics, such as brand awareness, brand recall, brand sentiment, brand attributes, and identify non intuitive correlations between the brand dimensions and other attitudes, interests and behaviours.
By relooking at advertising as an always-on means of engagement and learning about the audiences, search intelligence creates a dynamic feedback loop of ad messages and audience responses. By applying semantics to the field of digital advertising, one has the ability to understand true human behaviours, not only measure them. Semantics and search intelligence (based on semantics applied to display advertising) have just started to transform advertising.
Vincent Potier is chief operating officer at Captify