Brand sentiment analysis: Why positivity alone is no longer enough

03/10 By IAS team

Context and sentiment

 

This article is written by our EMEA MD, Nick Morley and was originally published in Marketing Tech News.

Have you been using tailored, brand-specific optimisation that detects the true context and sentiment of a page to precisely and accurately classify content? You’re not alone.

Last year saw an industry evolution from brand safety to brand suitability. Born from the need of more nuanced brand stewardship, this move was accelerated by a massive uptick in digital engagement from consumers and a shifting news agenda that made it even more important for each brand to direct spend towards the most suitable environments. The impact this made on brand risk levels were clear – with the latest Integral Ad Science (IAS) Media Quality Report showing a global fall in ads appearing beside inappropriate content between H1 2019 and H1 2020.

But alignment with page-level context and sentiment is only stage one of the brand suitability journey. To reach the next phase, brands must embrace emotional classification.

Why emotion?

Emotion is nothing new in advertising. Brands have long aimed to go beyond simply educating and persuading consumers by tapping emotional triggers that influence their decisions. From ads openly promising happiness in the late 1800s to subtler modern campaigns that seek to create it – see Cadbury’s drumming gorilla – and the now institutional John Lewis Christmas ad, the advertising industry has a rich history of harnessing feelings to forge a connection.

Context is recognised as central to creating the desired emotional response. When research shows more than two-thirds of UK consumers say how they perceive ads is impacted by the surrounding content, it’s widely understood that choosing digital placements with care is essential to spark the right kind of reaction and outcomes.

Striving for relevancy, however, is a broad goal; and one that’s not necessarily refined enough to optimise ad resonance or effectively safeguard brands. To secure maximum outcomes and suitability, brands should enhance their ability to understand the mood associated with any content.

Dialling emotive advertising up a level means taking emotional classification further. As well as assessing the suitability, relevance, and sentiment of content in line with their unique offering and values, brands need to dig deeper into the emotions it conveys. In doing so, they will not only enable better ad targeting – thereby increasing the likelihood of strong audience connection – but also secure maximum context control and protection.

Taking emotive advertising to the next level

It’s one thing to say a piece of digital content is positive but determining whether it’s focused on love or humour is something else. The same is true at the opposite end of the scale; with fear and anger both considered negative, yet still very different and sometimes effective emotions – think charity ads placed alongside shocking or tragic content to push a call-to-action. Once again, the key factor here is context. Being able to steer towards or away from content based on its positive or negative nature is valuable for guiding decisions about the type of messaging brands should use. But when it comes to selecting which specific environments to target, more detail is required.

Vital to achieving this is a granular understanding of the emotion behind each word; and for those who thought semantic technology was only about context and wider sentiment, this is where the good news comes in. When powered by vast knowledge graphs, the advanced contextual barometers brands have come to rely on can map words to potential meanings in all the given context, giving a precise emotional reading.

How does the tech work?

Human language is extremely complicated, but tech innovations are making it easier to navigate. There are now smart tools that have developed huge semantic networks covering all 500,000 concepts of language and the relationships between words, in addition to diverse emotional contexts linked to more than 60 grades of human emotion.

Using contextual catalogues, the text from any page can be used to produce a bespoke knowledge graph; simultaneously detecting the real meaning of every word it contains and identifying associated emotion. Another integral part of this evaluation is a combination of machine learning and sophisticated technologies – natural language understanding and semantic analysis – that allows for human-like comprehension: such as the capacity to immediately spot the difference made by small linguistic variations, from changes in tense right down to the use of a full stop.

What all of this delivers for brands is comprehensive real-time intelligence that enables precise ad matching for multiple contextual factors including emotion, at scale. For example, instead of setting the parameter of “positive sentiment” for a Valentine’s Day campaign, brands would be able to apply in-depth analysis that assesses different emotions tied to the romantic event, ensuring ads appear alongside content that will reach consumers likely to be interested in gift options, while avoiding potentially harmful placements.

Establishing brand suitability as a new gold standard is a major step forward for the industry. By stepping away from binary brand safety measures, brands have gained access to numerous high-quality environments that are not only safe, but also suitable. Yet, many brands are still framing ad campaigns around emotional classifications that aren’t granular enough to keep their reputations or consumer relationships secure. It’s time to start exploring the wider capabilities of semantic technology and introduce deeper emotions into suitability strategies.

 

Explore the IAS Power of Context research here to understand how contextual targeting can be an effective targeting tool in the cookieless world.