By Artiom Enkov, Head of Insights & Analytics, Nano Interactive
At Nano Interactive, we sit on a vast intent dataset. Our team’s speciality is to delve deep into it and provide bespoke insights to our clients. Most often, this then translates to bespoke targeting to drive success for their campaigns.
We see over 1.5 billion live intent signals per day – to make sense of that huge mass of data, we create a so-called intent map for each vertical, which we call Sector Modelling.
What is Sector Modelling?
At high level, Nano’s Sector Modelling process consists of three steps:
Firstly, we mine all intents within the landscape for a specific sector. In this case, let’s take the example of the automotive sector.
Then, we deep dive into this broad data set and use natural language processing (NLP) techniques to understand which entities are most important within it. This could be brand, person, place, events, product or any number of other variables and sub-intents. A major benefit of this is that we can also understand intent at various levels – so for automotive it could be by manufacturer, or by model group, by engine type or even car colour and parts.
Finally, we move into the machine learning aspect, where we train our algorithm to better understand each entity and category within that vertical. For instance, for the Volkswagen manufacturer this means we can then target all VW-related content, live in the moment, without reliance on keywords. And what’s more, whenever a new VW car model is eventually released in 6-12months time, we can also target that content. We don’t need to update our keyword list as the vector-based understanding is in place to understand that the new model closely relates to the VW manufacturer.
From an advertiser perspective, Sector Modelling gives so much flexibility and so many targeting customisation options that it is pretty much unrecognisable when compared to traditional keyword-based contextual targeting.
How does Sentiment and Emotion play a part?
Sentiment and emotion targeting are still somewhat under-explored areas in the media landscape. For our part, we have been testing sentiment and emotion metrics for quite some time now. In simple terms, our NLP technology understands whether an article happens to be negative, neutral or positive on a topic in nature – which we can then target according to.
What is interesting is that for some competitor conquesting campaigns we have found that negative content drives a better clickthrough rate. In fact, this would seem to contradict some widely practised approaches across the industry right now.
We are also excited about the possibility of being able to analyse campaign performance by emotion, so we can help brands better align with content displaying a particular emotional state.
If I could give any advice to advertisers, in no doubt very over-simplified terms, I would suggest reaching out to our Insights Consultancy team with their next challenging brief to learn how Nano can help with a bespoke Sector-Modelling-powered activation strategy.
