Why apply Machine Learning to Marketing Automation

Interview with Rinaldo Zambello, CEO&CSO @NUR Internet Marketing

Rinaldo Zambello
CEO & CSO @ NUR Internet Marketing

Modern marketing has very high requirements and is very fast-paced.

The multiplication of communication channels and the frequent interactions with users have made it harder to intercept the interests of each individual. As it is well known, without personalized experiences it is difficult to attract the attention of users.

If you mean to focus on users, you need data. In this context, a helping hand comes from Content Intelligence, a term indicating those strategies that apply Artificial Intelligence to content, in order to customize the visitor online experience. 

Content Intelligence works on two layers: the first one, through Speech-to-text, Image recognition, semantic analysis, automatically classifies content with tags. As a second step, Machine Learning algorithms track the content distributed on the front-end channels to collect data on users who use it. To sum up: tags describing viewed topics = tags describing user interests.

The data set thus obtained constantly fuels your CRM, so you will have a tank of information on the user navigation path, updated in real time, that will make your Marketing Automation actions increasingly targeted.

In addition to this, what are the advantages of applying Machine Learning to Marketing Automation platforms? Let’s discuss it with the expert Rinaldo Zambello of NUR Internet Marketing, a company that carries out web activities aimed at bringing business results.

 

Q: How does Machine Learning allow us to go beyond merely automation to enter the field of predictive analysis?

A: Actually, we are already using this technology extensively in Marketing Automation (MA) applications. Let's first clarify how MA systems work: they are programmed by people who teach them, through elementary logical mechanisms, to react to a given data input with a corresponding output or action (e.g. sending an email whenever a cart is left abandoned), and this will influence the performance of the whole platform.

There’s where Machine Learning comes in: it is no longer people who set the rule establishing which product can be related to a certain type of person, but the platform itself, collecting as much data as possible from the users themselves, even through social media; ML can analyze it and define in self-learning the degree of affinity of a person with our e-commerce products, and adjust accordingly. When an abandoned cart occurs, the machine will decide on its own the most effective lead generation and nurturing strategy to use in order to keep the customer.

Thanks to data aggregation and self-learning, these new MA features can predict user behavior and quickly react with customized responses. What you get is predictive data: you can judge the percentage of affinity of a person with your products (for example 80%), and even with products found outside your sales platform. Tying MA to social activities and Machine Learning as we do means being able to intercept people desires in real time, and to establish the degree of interest of a cluster of people in a product or service.

 

Q: How can Machine Learning make a Marketing Automation process like recommendation more efficient?

A: The combination of social media, data and Machine Learning can be beneficial for e-commerce. We tried it with Facebook and Twitter for the perception of sentiment towards the brand, its products and the audience "wishes", and once all this data from social communications is processed by Machine Learning algorithms, you get recommendations that allow you to really personalize your marketing strategies.

 

Q: Why do you think it's so important, in order to really be "customer centric", to use Artificial Intelligence tools that can extract real time data about user preferences?

A: These needs have been emerging for years. We all want to feel treated in a unique way, so brands must gather differentiated information to ensure more effective communication. This is the case with advertising: if I receive offers I’m not interested in, they make me uncomfortable as they demonstrate that company does not know me, but if I get something relevant for me, I will pay more attention to it.

Having an automatic tool that recognizes affinity with the products enables a more targeted communication, in the same logic of inbound marketing. There are no alternatives: to provide a customized final action, data must be accompanied by a training of values. We brought this experience to both B2C and B2B. Take the case of a pharmaceutical company we are working with: when a customer is about to leave, by using the data collected the system can make a prediction on the best action to suggest (this customer is leaving for this reason... so you can intervene in this way...)

Machine learning constantly learns from data, including historical data, so you can improve its intervention more and more. It is no longer a static analysis, but something more advanced. The results are very convenient: greater loyalty, increased engagement and conversion rates.