Can AI restore trust in marketing?

AI-based-marketing-automation-platform_B2B-automation

Most of my friends aren’t marketers and, like many people, they don’t think I’ve chosen a very trustworthy profession (marketing and advertising tend to rank the lowest in people’s integrity list). My friends are suspicious of marketing and fear that companies are increasingly using their personal information to persuade them to buy things they don’t need, spend more time in that app, or even manipulate their belief systems. A recent research shows that only 7% of the respondents felt positive about digital advertising. I guess after years of spam emails, unsolicited ads and most recently, the Facebook data breach controversy, the fear is justifiable.

But there is hope. Technological advances have created many new opportunities for marketing in the last decades and we are now at the forefront of yet another digital revolution. The emergence of artificial intelligent (AI) systems is set to give marketers a comprehensive view of consumers like they never had before. This creates the opportunity for us to deliver a truly relevant and personalised customer/user experience through segments of one. The era of irrelevant advertising and content could soon be over.

While there has been much fear mongering in the market about AI, there is also plenty of research suggesting that consumers want personalised advertising and are willing to provide personal information in return for a more individualised digital experience. Used wisely and responsibly, AI will bring immense benefits to brands as well as consumers.

Artificial intelligence can make marketing more intelligent

The opportunity for marketers lies in how AI can help us realise the power of data in decision-making to create personalised experiences. This means using AI algorithms or machine learning engines to analyse thousands of data points about customers and detect patterns in user behaviour in order to predict which individual messages, channels and products are most likely to be relevant to individual users. Customers will no longer waste their time on marketing messages that add no value to them. Content will be targeted to their tastes, concerns and stages in life and will only focus on things consumers really want to hear about.

At present, a common way of trying to deliver relevance is by segmenting and ranking customers and prospects according to a set of variables such as demographics, response to previous campaigns, previous purchases, type of device, etc. The problem with this process is that it’s labour intensive and involves a lot of guesswork and gut feeling trying to find the right combination of variables and the scores that will maximise conversions. Another issue is that large part of the data in this equation relies on direct input from users. They may have completed a form to sign up to your product newsletter and the static data they provided is unlikely to be updated and will become obsolete.

More importantly, algorithms reflect the biases and limitations of their human creators and cannot analyse all possible variables or recognise all patterns and trends that could signal a customer’s intent. Marketers may have the best intentions but by using this process, they can’t ensure that they are presenting the right message at the right time to the right customers.

This is where Artificial Intelligence comes in. Machine learning, a subset of AI, enables computers to analyse data the user left every time he/she interacted with the brand, uncover patterns of behaviours and draw new conclusions that weren’t previously programmed. It works by creating models that learn from raw data sets and each new piece of data received allows the machine to learn even more. Hence, machine learning continues to optimise itself; the more data it analyses, the smarter it gets.

What does this intelligence look like in practical terms? Netflix and Amazon are the classic examples of the use of machine learning algorithms to match consumers with relevant content. When you visit their websites, you can see that the movies or merchandising they are recommending have been tailored to your taste, interests or needs. You don’t have to tell Netflix you like movies with Julia Roberts. They know your browser history, the programs you have watched or abandoned and will factor that in on your recommendation list. Your profile is created dynamically and is being written as you interact with the service. And every time Netflix misses the movie recommendation, or Amazon doesn’t get a sale, the model learns from this new data and adapts to provide better product or movie suggestions next time.

Old ways of segmenting could soon be over. The gender or age you provided when signing up to the service does not have the same weight as your online behaviour. Netflix has learned that demographic data is almost irrelevant in predicting the right content for its users. Behavioural data proved to be much more precise in segmenting their audiences. Today, consumers are more complex and no longer follow certain expectations and pre-determined paths and their purchasing journeys have become much more diverse and unique. A rules-based approach or traditional segmentation misses out on insight that can now be gained through machine learning algorithms.

AI begins with a good data foundation

AI may sound like magic but really it’s data crunching in more sophisticated ways. For machine learning to work well, it needs data from various customer touch points in order to power the algorithms and deliver significant results. The more data sources and data points, the more capable the algorithms are of anticipating your customer’s needs.

While large amounts of data is paramount, you don’t need to have access to the same amount of data as Netflix. Database marketing has been around for decades so if your company has been collecting most of your customers online and offline footprint, you should have enough data to use machine learning. The number of data sets you need will vary according to the type of algorithm that will be created but for something like lead scoring, experts say that you need roughly 500 “positive signals” or “won deals” in our CRM in order to create a successful predictive model.

If you aren’t collecting insight on all relevant customer touch points, you need to build a plan for future collection which must include the integration of all marketing databases so you can analyse data along any dimension that is important to your marketing decision-making.

With great knowledge comes great responsibility

By obtaining and analysing such wealth of data you are getting to know your customer in an almost clairvoyant way. The question is, will you be using this information to help or hinder customer trust? Better knowledge about customers allow you to become more persuasive in your marketing messages and this raises questions about a marketer’s and a company’s ethical obligations. If you have identified that a customer follows a pattern of shopping addiction, would you persuade her/him to buy yet another pair of shoes? Would you promote soft drinks to customers if you learnt they have diabetes?

AI will place a spotlight on ethics in marketing. Professionals that use AI ethically to assist the customer will gain trust and reputation. Those who don’t will be relegated to the category of trolls and scammers. Just look at the world of abuse that Facebook is copping from customers and investors right now.

Marketers should also consider issues around transparency. Do your customers know the value of the data exchange, are you allowing them to make an informed choice as to whether they are willing to participate in your marketing program?

Ultimately, marketers are responsible for creating strategies and campaigns based on the insights they pull from AI systems so they need to consider codes of ethics and governance to protect brands from making unethical choices. By doing so, they will create the opportunity to turn customers into partners rather than accidental targets of their marketing. My friends, and I’m sure yours, will welcome that.

 

Article originally published on LinkedIn by Rosana Wayand.

 



Get In Touch
Get B2B Research