A Q&A with Lars Petersen, Vice President, Business Optimization, Sitecore
When our Mind in the Machine blog series last checked in with Lars Petersen, he explained how machine learning would give marketers a 360-degree view of customers. Today, he shares details of exactly how that happen, and what marketers can do to get the best results.
Q: What recent changes in machine learning have made it easier for marketers and front-end developers?
Lars Petersen: It's more accessible, connected, affordable, and easier to apply without needing a team of data scientists. That has changed a lot. And it is much more embedded in marketing platforms. Instead of living in isolation, it's becoming a part of the user interface that the marketers are using, so it’s easier to use machine learning. Of course, we are seeing more and more organizations using it—and they’re getting results. That’s putting pressure on those who aren't using it, so they need to get on the bandwagon.
Q: From your perspective, are larger companies leading that transition?
Lars Petersen: Some are—the large companies that have a good data foundation, as well as a vision of being on the leading edge, as it typically takes more from a large organization to get something launched. But it's also mid-size companies because they are more agile in how they can apply it—machine learning can make them much more competitive. It's an edge if you get in there fast, and you use it to get insights about your customers and their behavior. You can use that to optimize experiences. You will sell more because you will provide better and more-connected experiences.
Q: Will machine learning produce new insights on its own about segments and personas?
Lars Petersen: Yes. If you have structured content, you can derive more insights. So it's a matter of starting with the end in mind, and then put in the details of what you want ML to look at. That way, you give the machine a lot of detail to work with.
Typically, you’ll have two sets of data. You have usage data, which is all about your different customers and their behavior. From that you can see what leads to success for different segments—or by individuals. And then you have your catalog, which is your content, products, etc. In that catalog, you can add additional categorization to it. When you have enough usage data, then the machine will try to find insights and patterns that help you succeed. A lot of traffic and many conversions speeds up the process—and accuracy of ML.
Q: Are some companies already using artificial intelligence without human intervention?
Lars Petersen: Nobody is doing that yet because a machine can't make up empathetic experiences—yet. That's requires human intervention. We need the creativity of humans. We do see a big demand coming from organizations to us AI in order to automate personalization. What ML enables is speed, scale, and automation. Used correctly, it will enhance operational efficiency and let marketers be marketers.
Q: Are there some downsides that marketers should consider as they head into machine learning?
Lars Petersen: There's no real downside to machine learning itself, because it will help you achieve more conversions to support your business objectives. But many people think it's a silver bullet, that it's a red button that you just push, and then everything will be automated and they don't need to do anything. There’s also a need for organizational readiness to understand what machine learning is, how they can work with it, and how effectively it can help them increase business objectives and what the dependencies are for seeing the best results.
Q: What’s the best way for marketers to get started with machine learning?
Lars Petersen: The key to getting started well would be to have good data–think structure and taxonomy from the get-go. It’s not about just pushing in a lot of data and big data, but structured data. That will help you be much faster. It’s also about starting with specific, well-defined machine learning use cases, because you can do a lot with machine learning. You can have a website and there might be 40 different uses of machine learning in different contexts.
So start with scoping what you're actually starting with, and then use that to get some learnings. Learn from that and expand. But get started now. This is not optional, this is coming, and this will help you be much more efficient and bring the creative side back to marketing—and it will increase your revenue.
Lars Petersen is Vice President, Business Operations, at Sitecore. Follow him @LarsBirkholm.