How AI Will Impact Marketing and Growth in 2020 – Hey everybody! In this article we are going to look at the first steps to applying AI to marketing and growth. We are going to cover the must-haves, the should-haves and the nice-to-haves of different applications. We built this graph for the first time in 2017, but a lot has changed since then. So, it’s time for an update!
Two years ago at Growth Tribe we started to train professionals in marketing and growth on how to use AI in their businesses. It became clear that having an understanding of the potential applications of AI in business would definitely bring a return on investment for a career path or for a company’s capabilities. Not only the more traditional uses like lead scoring and churn prediction were becoming more accessible.
Text analysis and computer vision have now a much wider adoption in copywriting and campaign optimisation. This has actually been supported by the rise of collaborative platforms. These tools allow professionals who are not native to the area of analytics to evaluate, prototype and communicate the results to their teams.
Then we face the problem: Where should they start? If there was a list of skills that a self-reliant growth marketeer definitely had to learn in 2017, there were other less mature applications that should just be understood. In this way, they could know what was possible and evaluate future opportunities.
So we decided to put this in a diagram, ranking the maturity of the AI applications to marketing and growth. Now, two years and many early adopters later, the priorities list has changed. All the applications became more mature and the landscape became flatter, showing a smaller maturity gap between the must-have and the nice-to-have knowledge The most mature applications, those that a professional has to master, are still the same.
Using predictive modelling to anticipate the steps of the marketing funnel and clustering techniques to create data informed personas are still the fundamental skills. These two forms of machine learning, when an algorithm learns from examples without being explicitly programmed, became even cheaper and easier to understand. The book Prediction Machines from 2018 is an excellent publication that explains the economic impact of this cheaper and more accurate technology.
One of the main changes is that we downgraded recommender systems from I should have to a nice-to-have. Despite being one of the most influential applications in our daily lives, by suggesting which songs, videos, products and news we should consume, The necessity of having a recommender system is reduced to a small group of companies These are businesses, typically subscriptions or e-commerce that have a very dimensional number of products and services in which users would have a hard time to find their niche preferences.
This is the so-called longtail problem. Learning how recommenders work is important for us as modern citizens but not so applicable in most of the business models we work on. Now, the analysis of human language stayed as a should-have. Natural language processing has definitely consolidated its applications. It’s cool to understand how the voice assistants offered by the big tech companies work. but it’s even more interesting to learn how to improve copywriting by analysing the sentiment and the implicit personality of your customers. A simple text mining algorithm can help a company to identify topics in customer reviews that would be impossible to be identified by a human.
Moving now to psychographics. Remember in early 2018? We had the data privacy scandal involving Facebook. It revealed that Cambridge Analytica had harvested millions of people’s profiles without their consent and used this data for political advertising purposes. They used online advertising to target voters in the US and in the UK with specific messaging in conformity with their personality and previous beliefs, which is defined as the use of psychographic data.
Since then, public ethical concerns and data privacy regulations like the GDPR made marketeers to be more diligent and try not to look creepy with their content. The risk return here is still unclear so we decided to keep psychographics as a nice-to-have. Finally, good things come for those who wait. In 2017 we first saw image recognition as a hard-to-acquire skill. Teaching computer vision sounded challenging, even with the use of no-coding tools. To our surprise, image recognition became one of the most relevant, easy-to-acquire applications.
Image scoring for ads can make a junior professional produce visual content like an experienced one. Also, the company saves resources by not having to test so many variations to check what will actually work to generate traction and engagement. Use cases of image similarity showed good value as well. Basically, these look-alike algorithms find products that resemble each other and that customers might like them too.
One tool that really represents the time we live in is DataSine. It uses AI to provide text and image analysis to create ads, landing pages and emails that are then recommended to your specific personas. We tested it one year ago and we have been using it in our training since then. So, from traditional applications to the rising ones, the next coming years are promising. Marketeers will have even wider access to tools that use visual analysis instead of coding.
They will continuously exploit tools that improve and scale up personalisation in a timely and relevant manner without sounding invasive. We hope this article will help you to take steps to apply AI in your business and career. Let us know your thoughts in the comments below.
reference – How AI Will Impact Marketing and Growth in 2020
Web enthusiast. Thinker. Evil coffeeaholic. Food specialist. Reader. Twitter fanatic. Music maven. AI and Machine Learning!