Depending on who you listen to, artificial intelligence (AI) is either going to revolutionize the world in a positive flurry of intelligent solutions, or cause the downfall of society in one of a number of inevitable ways. It’s testament to AI’s potential that predictions around it are so extreme. For most, though, intelligent solutions will have a positive if gradual effect on working life. At least in the short- and medium-term, it’s unlikely that AI will bring about any major revolution.
But the potential power of the technology has been well demonstrated by some high profile wins. Over the tumultuous 2016, AI made correct predictions where the rest of the world made false assumptions. The United Kingdom’s Brexit vote and Donald Trump’s successful election campaign fooled not just the media and the wider general public, but also the polls. Neither polled well, but a number of AI companies called the election long before it became clear that Trump could win. Sieving through countless social media posts, videos, comments, shares, etc, these AI systems pointed to a Trump win by assessing sentiment, rather than surveying the public.
It’s this that makes AI such an exciting prospect for digital marketers. The tech has the potential to take existing data analytics practices and further empower them. Analytics can look at customer behavior and influence decision-making, whereas AI can get at previously unattainable information like sentiment. Trump was clearly popular, and the conversation on social media clearly betrayed this, but the open support at the polls or at rallies didn’t accurately reflect it.
Of course, there is far less reason for consumers to hide their feelings toward a brand in public, but what sentiment analysis gives you is a much fuller understanding of popularity, on a previously impossible scale. To assess conversation around a brand’s campaign manually wouldn’t work, there is simply too much of it, particularly when the campaign in question is on a large scale. It wouldn’t take a large scale AI solution to uncover the disdain for Pepsi’s most recent ad - a misfire starring Kendall Jenner and some particularly smiley policemen - or the popularity of H&M’s Wes Anderson short film from last year.
Reactions to marketing campaigns are generally more nuanced than these, though, and sentiment analysis can help marketers find out exactly what the consumer likes and doesn’t like, and can tailor its output accordingly. For example, Expedia Canada aired an ad which featured violin music that audiences found incredibly irritating after multiple airings. Through sentiment analysis, Expedia was able to determine that over half of the online comments made about it were negative, and so ran a counter ad with the violin in question being smashed. The U-turn showed that the company has a sense of humor and is willing to listen to its customers, when the ad could potentially have gone down as a dud.
Having said that, sentiment analysis is a long way from being a perfect science. Though the technology is currently intelligent enough to pick up general sentiment, it still has difficulty assessing the tone and complexities of some pieces of text. Sarcasm is probably the most obvious example of language that could trip a machine up. But humor, irony, exaggeration, spam, trolling - AI will have to become intelligent enough to distinguish between all of these before it can present a truly comprehensive roundup of public sentiment. There is also the feeling among experts that sentiment analysis must also develop categorization beyond the ‘positive’ and ‘negative’ binary it currently uses. Only with this next level of insight can brands fully rely on AI for sentiment analysis. Gartner estimates that, by 2020, 85% of customer interactions will be managed without a human - companies will have to be sure their machines are sophisticated enough.
And AI in digital marketing will not just be limited to campaign assessment and sentiment analysis. Last year, outdoor clothing retailer The North Face became the first company to enlist IBM’s smart computer Watson for its mobile app. Rather than having the user scroll through its 350 jackets, The North Face asks its customer a series of questions - where the jacket will be worn, what the season will be, and the gender of the wearer - and generates results based on the answers. This may not immediately sound revolutionary, but the technology is intelligent enough to deal with all manner of responses, and draws a number of determining factors from each one. The North Face’s VP of Digital Commerce and Experience, Cal Bouchard, described the tech as being ‘at a second or third grade level,’ but it will grow in sophistication as it learns more responses. From a marketing standpoint, the initial effects of Watson have been positive. Bouchard told VentureBeat that the technology generated a 60% clickthrough rate to try product recommendations, while 75% of users said they’d use it again.
Currently, AI cannot run marketing campaigns alone. Humans are still very much needed for both creativity and validation, but the growth of the technology will give marketers more and more insight into the efficacy of their campaigns as it develops. As for customer interaction, the explosion of AI-driven customer experience solutions will give marketers something to sell, with clickthrough rates and customer satisfaction levels likely to rise so long as the solutions are effective. AI is as unlikely to immediately take over the world of digital marketing as it is to destroy it. What it will do, though, is make the lives of digital marketers easier, and ensure that campaigns hit the mark more often than not.