In October 2016, digital marketing agencies across the world were forced to consider the horrible possibility of not being needed by any brand. Cosabella, the leading lingerie retailer, had decided to replace its incumbent agency with the innocuously named Albert – an Artificial Intelligence platform. If traditional agencies were of the opinion that only digital agencies were in trouble, they couldn’t be further from the truth.
For decades, marketing was about reaching the largest possible audience at the lowest possible price. Today, value is based on unique and personalised experience of each individual consumer. Generic messages are not good enough for the consumer anymore and marketers need to create specific messages for each consumer based on their behaviour.
Here are a few examples of the same:
Content creation: Gartner predicts that by 2018, 20% of all business content will be authored by machine learning algorithms. A Japanese novel written by an application built using the open source library developed primarily by Google, TensorFlow, had come close to winning a literary award before the revelation that it was written by a machine. Content creation via chatbots at companies like Uber is already emerging as an effective way of managing customer expectations and is being built into their websites by many a brand today.
Semantic analysis: Content curation and recommendation engines powered by semantic analysis have been acting without any human input for many years now. For example, Netflix estimates that their recommendation engine helps generate revenue of as much as $1 billion a year and 80% of the content discovery happens through recommendations. These engines play a similarly strong role at Amazon, Facebook and Twitter as well. Timelines on social media platforms are built by machine learning algorithms that consider millions of factors to decide what would be the most interesting content for a user.
Intent analysis: AI plays a role in customer service optimization as well. Intent analysis is revolutionising this space. No more does a consumer need to reach out to the marketer through a predefined phone number/ email id. One can use social media channels or any online forum to post their review/ concern.
Using AI, marketers can identify underlying intent even when the text is “unstructured” and classify the response among various categories. Then trigger the correct resource to provide a relevant response to the consumer.
Intelligent pay-per-click campaigns: Tools like Albert mentioned above and Frank are taking the guesswork out of pay-per-click campaigns. These tools manage the campaign with minimal human intervention and increase marketing effectiveness manifold. AI is also utilised to create personalized websites and push notification based on available consumer metadata like location, demographics and device used.
In each phase of marketing, AI and machine learning can play multiple roles. Enumerated here are the most mature applications to give a glimpse into these roles. With marketing becoming increasingly data-driven and machines stepping into the creative space as well – marketers no more have a choice but to accommodate some of these tools to harness the market forces.
Concerns: Since behavioural targeting is based on collecting data about individuals, many consumer advocacy groups have raised privacy concerns around such application. As a result, marketers insist that none of this data is seen by human eyes before it has been stripped of identifying information. The captured data is processed by a machine that has a specific goal programmed and does not use the data for any other purpose without an express permission from the consumer. Better returns from permission-based marketing compared to broadcast/ intrusive marketing has only strengthened the culture of respecting the consumers’ privacy.
Behavioural targeting: Marketing is no more a one-way street. Marketers now collect a plethora of information about the consumer, analyse it to build individual profiles and take personalized action based on these profiles. Marketers, like all cognitive agents, follow a collect-reason-act cycle. Artificial Intelligence can play a role in all the three phases and beyond.
With the advent of information systems, a continuous stream of data is available on each consumer. This is not just a history of the online activities, but also data from the offline world – like POS data, travel trails, etc. Collect phase of the marketing cycle pertains to simply capturing this data in a format that an intelligent system could process.
Improve Data: Some of the captured data is not always accurate. For example, the GPS location of a consumer’s device may become highly unreliable at times. At such times AI improves the reliability based on other factors like Wi-Fi connections available and locations of historically nearby devices. Further, AI is being used to derive additional information based on raw data. For example, based on rate of change of location data, AI can accurately guess the mode of transport.
This phase of the marketing cycle works to generate actionable insights based on the collected data. This is the phase where the use of artificial intelligence and machine learning is most mature at the moment. A joint study published by Wu Youyou, David Stillwell from Department of Psychology, University of Cambridge, and Michal Kosinski, Department of Computer Science, Stanford University found that a machine learning algorithm that analyses just 300 likes per individual can better predict a consumer’s actions than their spouse. Imaging how specific these predictions get when the algorithm has access to the complete digital footprint of the consumer.
Emotion & sentiment analysis: Under the reason phase, AI is being leveraged for sentiment analysis based on social media conversations. The increasing role of emotion detection in marketing is well-documented. This not only helps in getting a quick feedback on mass media marketing message and A/B testing message formats, but also allows the marketer to target (and decide not to target thereof) consumers based on their emotional state.
Insights are no good if no action is taken based on them. Here, too, artificial intelligence plays a role today. Zeroing down on the best time of the day for marketing message, writing a message most likely to trigger desired action, designing the right incentive for the target consumer – machine learning algorithms can do each of this better than human counterparts, once enough training data is available.
(Disclaimer: The opinions expressed in this article are those of the author. The facts and opinions appearing in the article do not reflect the views of BestMediaInfo.com and we do not assume any responsibility or liability for the same.)