At the start of 2013, I wrote about the trends that would dominate website conversion optimization discourse and practice for the year. The overarching theme then was context, with digital marketers struggling to create relevance across channels and platforms. While those trends will continue as the laggards attempt to catch up with early adopters, new disruptions—this time in the field of machine learning—present new opportunities and challenges for digital marketers in the coming year.
Rise of the Machines
Regardless of how you feel about it, machine learning is hitting it big. Search engines like Google are beginning to understand not just the strings that make up a word, but the connections between words. You can ask Google if it will rain in San Diego tomorrow, and it will tell you:
You can follow it up with the search “what about in Seattle?” and it will understand that you’re still talking about the weather.
Meanwhile, over at Facebook, they’re starting to answer questions that few else can. If you need to reach the fan base of Honda who are also into robotics because your product skews towards wearable computing, you can do that with Facebook. In no other medium, and no other channel did that used to be possible.
What both of those amount to, along with the continued proliferation of mobile devices, is that machines are starting to understand so much about who we are, how we communicate, what we’re looking for and when we’re looking for it—and that a lot about the old models about websites, testing, and conversion have changed.
Those who have embraced testing on social, those who began to track and understand multi-device behavior, those who understand why structured data helps even ecommerce sites—their lead is widening. If you’re not part of that first wave who embraced the changes, you have some major catching up to do. You need to look at what changed in acquiring visitors, what changed in conversion, think big, and adapt.
The Future of Visitor Acquisition
Machine learning-based disruptions that took place this year really hit the top of the funnel (visitor acquisition) hard. If you were doing a lot of keyword analysis to drive people with a certain type of intent … well, keywords are not provided by Google anymore. If you were driving people from search engines towards a popular definition of something to start the branding effort, more and more, Google will present that information without getting people to leave the search engine:
It’s not all doom and gloom, though. Machines are also reading through things like who the authors are for particular websites, and those can help you get higher click-throughs from search engines.
What does this mean for you?
For one, you need to focus on answers, not keywords. If there’s one thing you need to change about your visitor acquisition efforts, it’s this: Google is getting better at presenting the basics—basic definitions, NFL scores, the weather, directions to places, and the like. What you should be providing, for visibility, is not about basics—it’s not about keywords. It’s about answers. Think about the questions that your visitors want answered, and focus your content efforts towards answering those questions in a way no one else can, and no machine can easily replicate.
Second, incorporate structured information. You should definitely start with the authorship markup, but you shouldn’t stop there. Visit Schema to learn about what information you can provide about your websites so they show immediately on the search engines, enhancing the likelihood of you getting the click-through.
The Future of Conversion
Changes at the top of the funnel naturally impact visitor movement deeper down. So if you only do three things different about the middle and bottom of the funnel activities, make it these three:
Track and improve your on site search technology. When Google took away keywords leading to your website, that essentially left two good places to find information about intent. The first is Voice of Customer tools like iPerception and Opinionlab—you can ask users about their tasks and intentions directly. The second is the search box inside your website. People using that box are telling you what they need, and what they can’t find easily. With keywords gone, it’s one of the best places to find intent—so you better be tracking what the visitors are looking for and improving results. Once you have tracking set up, you should also use the data to improve prioritization on your website.
Social Media Testing. You can use a lot of metrics for social media, and a lot of them will lead nowhere. That’s not a knock on social media. If you are trying to boost your “Likes” by having an iPad Air raffle, and expecting conversions to rise, that’s not Facebook’s problem, that’s yours. To use the channel appropriately, you need to test what’s different between people you engage on social media, and those you don’t.
That’s right. To measure how effective Facebook is, you have to look outside of Facebook. You can upload your email list to Facebook, lead nurture a segment, and then use normal split tests on tools like Google Analytics to measure the conversion lift.
Social is affecting your middle of the funnel engagement right now, whether you like it or not—the choices you have are to ignore it, or to actively track and improve it.
Multichannel Attribution. As more and more people do things on multiple devices, your conversions get messier. Direct visits to your product page and the succeeding conversion in a desktop or laptop presence may be due to the exposure to an ad on mobile. You need to capture that information to know which channels have the best ROI, and to fix the right things, in the right order.
If you’re not investing in personalization at least for the different devices and regions, you’re missing out. If you’re not seeing the connections between how your traffic converts across channels, you’re missing out again.
Are the Machines Getting Too Smart?
As algorithms get so smart that they can converse and provide basic answers, your ability to provide the deeper answers will be what matters. If you are doing purely keyword-based SEO, strictly A/B testing-based conversion optimization, and reviewing only desktop and laptop website-based analytics, you are on a long slog towards becoming obsolete.
If you track how well you’re doing across multiple channels, get smart about what data you have, improve your understanding of social and move towards researching answers and not keywords, it won’t matter how smart the machines get—you’ll be just fine.
Circuit board with head with gears icon image courtesy of Shutterstock