We’re well past 2001 and we haven’t quite delivered to the vision in A Space Odyssey yet – but it’s getting close. There’s certainly a lot of discussion around artificial intelligence (AI) at the moment. In the digital world we live in, AI is really just an evolution of the smart marketing that we have been delivering since ’09. What’s clear, though, is that things are getting easier.
At Y+S, we’ve always had the philosophy that anything’s possible. Come up with an idea and we should be able to make it happen. Now with better connectivity between systems to allow smart automation, coupled with new AI services like natural language processing or image recognition, you can really start to do some interesting things. So, what is possible?
One example I came across recently was for Under Armour. They’re looking to combine user data from their app with third-party data and research on fitness and nutrition to deliver individually tailored training and lifestyle advice based on this aggregated/personalised overview. This delivery of complex recommendations isn’t new – it’s just getting easier to combine data sources and deliver at scale.
We’ve had assistants like Siri with us for a while. Now we have Google voice search and they recently launched Google Assistant. These services are getting smarter at recognising natural language and intention. As they refine their understanding and become more useable we could see a whole generation move away from typing to talking. This will have implications on how we search and how questions form the basis for searchable content. Predictive customer service – building on the idea of questions we think you could start to predict customers behaviour based on what questions they are asking or searching for or the FAQs they are looking at. From this automating a sequence of marketing messages personalised around that specific customers situation would be interesting. It could drastically reduce churn.
This is another interesting area. The ability to ‘visit’ different places (see our WitherHills VR campaign) or through augmented reality see what a new house or kitchen might look like offers clear value. Other simple ways image recognition will impact advertising is applications that can help automatically read images to crop them effectively. This will be significant help on responsive websites where automatically cropped images can end up less than perfect.
This is a core media job and one that’s become increasingly effective over the last few years. Adwords itself was the first ‘programmatic’ self learning system that would optimise campaigns to the most effective keywords and creative options. These foundations have been extended in new programmatic trading platforms that use publicly available demographic data and combine third party data and behavioural signals to target the right creative to the right consumer to drive more effective campaigns.
There are other impacts on how we drive our marketing – these are just a taste of what you can do now. So to do this, what are the core skills you need so you’re not left behind?
Data – is the foundation.
To start taking advantage of new AI platforms you need to have a core capability in data management. You should be collecting everything. With the cost of compute at an all time low in the cloud it makes sense to be gleaning as much information off of a user session as possible and learning how to spot opportunities in datasets to start building predictions. Learning the best way to clean, normalise and collate datasets from some or all of your sources is one of the core fundamental pieces to Machine Learning and training models which will work.
APIs – are the connectors.
A good understanding of how to connect systems so that you can take advantage of data and third party data sources to then create individualised experiences relies on connecting systems to make things happen. APIs (application programming interface) are the instructions that allow people to directly control the features of other applications. Driving these requires higher order development capability. Many of the new AI services are cloud based and accessible by APIs.
Development – is a core competence.
To drive a smart digital team you need development thinking. This starts right at project definition. Knowing what is possible is key. Inspiring the team to think bigger to open up possibilities is important. Your development team will also need to master some of the core fundamentals of Machine Learning; regression, variance, standard deviation etc, although at this point you could be looking at finding someone with skills in data science.
Data Systems – are the heart (or maybe the brains).
Marketing Automation and DMPs (data management platform) are where your core customer data and relationships are held. They contain the segments that are being formed naturally, the triggers that start marketing and the sequences that then happen. Together these systems manage and coordinate the load of content and advertising that we are serving to customer segments.
So what do you need to do now?
Firstly you need to make sure you’re moving into this space. Your competitors are already using AI, whether those companies are local or international. You don’t need to panic but start thinking about how you could be smarter, how you could inspire or add more value to your customers. Look at systems you use and the people you are working with. Do you have the right systems and people internally and are you working with people externally that are thinking about this space and actively investing in it?
With these skills in place it opens up the possibilities to start using some of the Machine Learning libraries that are out there and supplied by Amazon, Microsoft (Azure ML) and Google (Tensorflow). If you’ve got clean data and are curious the tools are out there to start.
Personally I think this space is exciting. It’s going to allow us to make better stuff that has a stronger ability to connect with people and make a difference. Bring it on.