Can machine learning solve the problems of marketing? - Yieldify | Customer Journey Tools

Search Yieldify

Can machine learning solve the problems of marketing?

Machine learning is very exciting-sounding and there’s probably a TED talk out there somewhere telling us it’s the future. But what bearing does it have for marketers and e-commerce optimisation?

 

First of all, what is machine learning anyway?

Simply put, it’s the ability for computers to do stuff without being specifically programmed to do so, instead taking their cues from data. Even if you’ve only scratched the surface of coding HTML, you can already see the potentially huge impact this could have.

Right now, your marketing technology is getting exposed to billions of datapoints and you’re seeing an accumulation of them on Google Analytics or Pardot or whatever tools you’re using. But it relies on you – yes, you the human – to make the decision of what to do next based on that data. Whether you decide to change course and how is down to you.

Generally speaking, machine learning takes the ‘you’ out of the equation. It basically means that instead of you having to tell your computer what to do with all that data you’ve gathered, it’ll take its own cue.

You don’t need to look further than Facebook to see this in action. As you scroll down your newsfeed, you might pause on a certain friend’s updates and comment on another’s; you then end up seeing more of those friends. But when you scroll straight past the updates from the guy that you met once in a hostel in Dubrovnik in 2007, you’re unlikely to see much of him again. That’s machine learning at work.

The key thing to remember is that as you supply machine learning software with more data, it keeps on learning and adapting – like a really intelligent person, only with exponentially more capacity and less need for sleep!

Yes, it’s widely considered (though not universally) to be a type of Artificial Intelligence (AI) but no, that doesn’t necessarily mean it’s scary. Remember that there was once a time before marketing automation (I know, try not to think about it); in many ways you can see the use of machine learning as the next stage in taking the heavy lifting out of marketing.

Machine learning visualisation

 

Why is machine learning being talked about now?

The answer is data. As marketers, the last couple of years have seen an avalanche of big data and – let’s be real – the vast majority of us don’t know what to do with these spoils. To our credit, machine learning exists because our human capacity to process data is inherently limited (there’s only so much Red Bull one person can drink).

The level of data that we have now could – in theory – allow us to make decisions on an individual visitor level to build an experience around them rather than have them fit into our idea of what the experience. With 89% of marketers saying that their customer experience is going to become their key differentiator this year, we see it as the marketer’s best strategy to win.

As marketers, if we’re going to be able to take real advantage of this data to convert visitors and increase sales, we need some help. And that’s where machine learning comes in.

 

Machine learning and your conversion rate

Right now, you likely have more than one IFTTT workflow set up somewhere. If your customer subscribes to your email, you might have a series of drip emails welcoming them – maybe with a new offer. This is a great way of making sure you’re always deploying the most effective marketing action, and you should do as much of it as you can.

But we all know that interacting on a personal level will always be better than a catch-all, and machine learning comes in to let you get really granular; it’s where the computer will drive the decision, not the rule. Instead of a workflow, software will know from data showing the most effective actions driving your goals for a certain type of customer what to execute in order to increase your conversion rate.

So for example, machine learning could take the legwork out of A/B testing for you; instead of having to keep reviewing your results and make the switch yourself, your software can pick the winner for you and set it to 100%.

Artificial intelligence

 

What might future solutions look like?

One of the key themes we’re seeing now in marketing technology is the proliferation of channels that a marketer will have at their disposal to reach customers, and the ways they’ll need to adapt content and messages to use them effectively. For example, voice solutions such as Amazon Alexa are forging a new kind of interaction that marketers will need to learn to use. To take real advantage of these and make the customer experience consistent and coherent, you’ll need to make thousands of decisions – and therefore probably get some machine help to do so.

For example, seeing that your customer has booked a holiday to Spain and offering her swimsuits in her size, with an option to deliver it to her office the day before she departs. There’s a huge selection of possibilities that ended up with that swimsuit being offered, purchased and delivered – and software can help you get that granular. These kinds of experiences could one day even take the form of digital personal assistants that combine services and products around a customer’s individual profile.

 

What’s the risk of machine learning?

When it comes to using machine learning in marketing, it’s important not to try to solve all your problems with it; if machines optimised everything about your business then everything would look like Amazon.

Your customers buy from you not just because of when and how you interact with them, but because of the quality of those interactions – your creatives, your brand heritage and the emotional connection you make all help add up to your conversion rate. When the success of your output hinges on human response, your input requires human judgement.

In this respect, machine learning can be its own worst enemy. We know how off-putting many consumers find it when they are ‘stalked’ by retargeting ads, so if we allow for over-optimisation marketing starts to veer into ‘creepy’ territory.

That’s why this is good news for the creative marketer; machine learning is coming to marketing technology to take the heavy lifting off your hands, but it still needs human creativity and common sense as we step into the next era of digital marketing. So, no – the robots are not coming for your job.

 

Want to learn more about machine learning? Good for you!

This cheat sheet from Gartner is a great primer on machine learning terminology

See Econsultancy’s round-up of the wider uses of AI in marketing.