A recent Blog post described the difficulties that businesses have with achieving a Single Customer View. Whilst this can mean many things, depending on who you talk to, the overarching goal is one of interconnectedness. While getting an entire organization’s data into one place has it’s challenges, a vast amount of data can often be delivered to the front-end of a website in a Data Layer, and be used to supercharge personalization efforts. This post will address using these for two purposes – Pre-Test Segmentation, and Test Data Exploration.
Why do it?
In many walks of life, we consider data to be as valuable as gold. This time last year, Facebook’s targeted ads delivered them a revenue of $6.24 billion in Q2, at an increase of 63% from the previous year. Google is set to exceed that significantly, with projected earnings of $72.69 billion for their targeted ads for 2017. Meanwhile, globally used Content Management Systems (CMS) are evolving to better utilise user-centric data, and tools such as Segment are being adopted to help distribute collected data amongst utilized tools/vendors. Our challenge, then, is to do a great job of both collecting useful data, and ensuring that as much as possible is ready for consumption.
For the uninitiated, Segmentation, Targeting and Personalization are amongst various names that we’ve come up with to describe finding an audience for your Project. As marketers, we should always be thinking about who we want to see our content, and where this information is currently kept.
For e-commerce websites, the “what” is often quite easy. We want to know who did what, and when. Did someone see a Product Listings Page (PLP) for some jeans in the last week, but not buy? And have they purchased jeans from you, over 3 months ago? If so, now could be a great time to present some jeans to them! Or, did someone look for a holiday in the last 3 days, but not purchase? Odds are that they’re doing their research and preparing to purchase soon – another reason to present a more relevant experience should they return.
Being able to lookup a User’s history, both online and offline if possible/appropriate, and present it back to them allows any business to provide a comprehensively personal experience to a user. This works alongside their wishes, with many Users “happy to give up a certain amount of privacy so that their experience can be enhanced and made seamless.”.
Our list therefore starts to look similar to:
- Browsing History
- Purchase History
- Loyalty Card Membership
With data collected such as:
- Pages browsed
- Categories browsed
- Selections/preferences expressed
- Products added to basket
- Products purchased
- Purchase and Exit cues detected
Once we’re able to both collect this information, and resurface it in a Data Layer, we’re ready for some pretty sophisticated Segmentation to occur.
It’s very important to note that personalization doesn’t need to happen in the moment that Users see your initial project. Sure, there are some projects that shouldn’t be shown to everyone, but there’s a strong argument for serving the same experience to everyone possible/appropriate, and then performing extensive data exploration/mining after the fact. The goal of this is to find well/poorly-performing sub-segments. This, again, is best handled where all information available is in one place. Can you push this information into your Optimization tool? Or, can the tool get it’s data out, and into a system that’s better integrated with your historical and user-centric information?
Once your data is in one place, you’re able to gain detail that absolutely insane segmentation would, but without the messy setup. For example, “people from Manchester, who recently viewed jeans, love our homepage takeover during the weekend, but hate it during the week”. Advanced Querying, or Machine Learning, plays a vital role in being able to drill down on masses of data. This would be done with the intent of adding dimensions to your data, applying the required math, and ending up with a more detailed knowledge of who does/doesn’t like what content on your website.
So, how do I get started?
The practicalities of achieving such Rich Data Layers vary based on the tools available. Working at Webtrends, I’ve seen it all – from many businesses who have a completely cold approach to every engagement, through to businesses who are able to present a very rich history of customer engagements at any given moment. Familiarising yourself with these capabilities is definitely Step 1.
Everything you do from that point onwards should have the goal of feeding rich user-centric information into your Testing tool, either beforehand or when analysing data. Once you do so, you’ll end up with a seriously supercharged personalization programme, that will contend with the most experienced of organizations out there.