written by
Sara Sihelnik
Sara Sihelnik

How Can AI Be Used for Data-Driven Marketing?

How Can AI Be Used for Data-Driven Marketing?

Artificial intelligence (AI) is one of the most promising technologies of the future and is already being used intensively in numerous industries - including marketing. But how can AI be used for marketing purposes and what challenges and risks await marketers? We talk to Sara Sihelnik, Country Director DACH at Quantcast, a technology company that specialises in AI-driven real-time advertising.

What are the most significant applications of AI in marketing?

AI is an umbrella term for a branch of computer science that covers various subfields, for example general problem-solving approaches, language processing and robotics, and the most interesting discipline for marketing: machine learning (ML). ML can help marketing to base decision-making on Big Data rather than assumptions, subjective perceptions and selective studies. Much of the data available to marketing today is not in a form that can be readily analysed and translated into decisions by humans. In order to be able to extract relevant and specific information from this data, mechanisms are needed that systematically and objectively analyse patterns, form models and derive decisions from them.

What are the benefits of AI in this regard?

This technology is important in many aspects of marketing, for example the personalised purchase recommendations we see in e-commerce. Based on the interest profile of individual customers and with reference to historic customer behaviour data, the one product among many is recommended that is most likely to be relevant to the customer. It goes even further online: the browsing behaviour of an internet user can be used to determine with relatively high precision which content or products are of interest to this user and this benefits online targeting. Furthermore, ML helps to identify trends, carry out target group analyses and much more. Decision-making in marketing, now a billion-dollar industry, is therefore increasingly based on a solid data foundation and requires less guess work. Many everyday administrative tasks can also be automated, something which reduces both susceptibility to errors and creates efficiencies, as marketers can thus concentrate on strategy and creativity.

With AI there is always the risk of data errors. What do marketing decision-makers need to be aware of?

Our CTO likes to say that even the most advanced technology should never replace common sense. There always needs to be a human corrective element that checks for discrepancies. Ultimately there will always be an interface between the technology and the people who use it - for this reason, translation errors can naturally creep in between intention and actual implementation.

An example of this: when it comes to AI, any data-based decision can only be as good as the data set on which it is based. Most importantly, the data set must be large and diverse enough to accurately represent the various possible circumstances to which the resulting decision is to be applied. A product recommendation mechanism in online shops, for example, that is based on too small a data set is prone to error and cannot take enough factors into account. This could lead to a 19-year-old being recommended a rheumatism blanket because the underlying data set does not accurately reflect the different age groups.

The quality of the data is also important. In online advertising, for example, data that is either already a few weeks old or has been captured incorrectly from the beginning can damage the success of the campaign in the long term. For example, you might see advertising which reminds a customer of an outdated offer, which they have already taken up, because of outdated data records, or an incorrect data record attributes the wrong gender to this customer and as a result they see advertising for completely unsuitable products. Such errors not only mean that the investment goes to waste, in the worst case the experience is annoying for the potential customer, which can result in negative associations.

What are your best cases?

In programmatic advertising there are generally concrete, sales-oriented objectives which can be achieved with technology and our treasure trove of data. However, much more is possible with successful collaboration. A deeper dive into the data allows marketing not only to answer concrete questions on implementation, such as who should see an ad, but also supports strategic considerations as to whether pure performance marketing is the right approach here, whether a brand-boosting element is missing, where there is potential to tap into new target groups and which messaging is best received by them.

What trends do you foresee for AI in marketing?

As the pressure to justify investment in marketing is particularly high in times of financial uncertainty, we expect advertisers' interest in data-driven marketing and automation to continue to grow. A game changer in programmatic advertising will be the question of the future of the ad-financed internet, as browser-based cookies are expected to all but disappear by mid-year. This poses new challenges for advertisers, but also for marketers. Advertising technology offers solutions here which are already being tested in other European markets - it only remains for us to hope that Germany catches up quickly so that neither advertisers, internet users nor website operators suffer.