A deep dive into predicting the behavior of users in general and your site’s visitors
This January, Sitecore held the Benelux Winter Camp 2015 with a series of twelve webinars about digital marketing. A kick-start for the new year with a lot of interesting sessions about online customer experiences and online trends to help you improve your digital marketing strategy. At Colours we took part and hosted one of the webinars, called ‘Predicting visitor behavior’.
Lars van Tulden, our User Experience Consultant, and I spoke about the following topics:
- Why would you want to predict behavior?
- What’s in it for the visitor of your website?
- What are the trends in the market and what can we expect in 2015.
- What are the possibilities regarding your current Sitecore solution.
- Sitecore 8 and its aim towards customer experience and automation.
The technical evolution is speeding up and our visitors are expecting and adopting new technologies faster than ever. This requires marketeers to follow trends quicker. Research shows that 93% of all marketers think online personalisation is essential for their business and over 90% believes that succesful brands are using customer data to base their marketing decisions on. But according to research by Adobe, only 20% of the content of about 65% of the websites is specifically targeted at an audience. Why? A diversity of reasons. Lack of tooling, not knowing where to start, not enough data or maybe too much. Fact is, the customer wants it. Expects it.
Predicting online behavior means you are helping your visitor by showing the information that is relevant to him at this moment, even before he has discovered his own need for this information.
And then there’s privacy. People expect relevancy but are worried sharing their personal data. There’s a fine line between empathy, getting the right context and on the other hand influencing the customer’s experience or behavior to a point where it’s unexpected or uncalled for. In this webinar we will elaborate on this.
So what it is we’re after? We need to sense needs, circumstances. Or people. Slowly more large brands are evolving regarding this matter, like Google Now. Or YouTube, keeping your experience fluent across devices. But a lot of services still can and should be improved.
We will give you some tips and tricks on where to start. But the most crucial of all is to start small and to start right now. Build up profiles, collect user data and feedback. By learning as you go you’ll get to know your visitors, hence being able to serve them better. In short, the maturity model would be: get your basics to work (a Sitecore website built fully customizable using the Page Editor), gather knowledge by collecting analytics data, enhance your content strategy, personalize and eventually.. predict!
Don’t let your visitors lose their way…
A good example of predicting visitors’ needs is shown by Google Maps. If you have visited a certain restaurant (frequently), Google knows you did via Google Now. When looking at Maps later on, Google will show you the location of this restaurant at a zoom level that reveals no other information of this type of data (restaurants). If Google would have shown you all restaurants, you wouldn’t be able to tell where your favorite one is located; the map would be completely filled, with useless data as far as you’re concerned.
Sitecore XP 8
Going back to Sitecore 6.1 with OMS, online personalisation isn’t new. But Sitecore’s premium marketing instrument came a long way since and its capabilities are becoming endless. The Experience Database and Experience Profile features really live up to their name. In this webinar, being targeted at marketers – or a non-technical audience – I will briefly describe how personalisation can be applied and which tools are at your disposal.
But the most exciting news is that Sitecore will evolve from enabling marketers to personalize their content based on their own experience or offline research, to a platform that learns from personalisation rules by itself! Sitecore already uses machine learning techniques by utilizing AzureML features and their roadmap is highly focussed on this new area of personalisation. Predicting visitor behavior will become the new trend in online marketing!
Weather forecasters have been using machine predictions for years; they know that when the temperature drops over the Atlantic and it gets foggy in the Amazon – then it’s 99% certain to rain in Manchester.
– Mark Simpson, Mando Group
I like this quote a lot, because it clearly explains how machine learning works. By collecting enough data, which is done by the Sitecore Experience Database, the AzureML platform can be thought what success looks like. By doing so you will enable Sitecore to predict future trends, segments and customer behaviour. Where a multivariate test in Sitecore 7.x would only report back the variant with the highest value, Sitecore XP 8 is able to suggest future tests and segments based on this test.
Segmented multivariate testing
The last new feature of Sitecore XP 8 I want to share with you is segmented multivariate testing. This is a lot less complex than the machine learning concept, but not less useful! When testing two options in a multivariate test, it’s obvious that although one option will be the most popular and effective one, there still is a segment that likes the alternative better. So what if you can discover a specific target audience in this, like ‘weekend visitors’? Sitecore now lets you personalize your multivariate tests and while getting your conversion rates up while shooting at the majority of your visitors’ needs, you can also learn how to serve that other group as well.
This webinar shows the principals on and motivation for using machine learning techniques. Hopefully I can get back to you later this year on this topic, showing the advantages of this technique in a more detailed and technical fashion!