As in-store technology evolves and customer use of smartphone apps increases, we're seeing a breakdown of the traditional barriers between traditional in-store retail and e-commerce - digital engagements are now transcending online, mobile and in-store. This Unified Commerce is a single approach to your sales, marketing and customer experience, wherever the customer happens to be. As retailers move towards this unified commerce strategy, marketers are learning a lot of lessons about the curse of data silos; the practice of keeping data from different sources separated from each other. Silos may or may not be under the control of one department within a company, but the defining feature is that they don't play nice with each other. If the marketing department can't easily combine data from in-store point of sale, social media and e-commerce sites, then unified commerce's promise of a seamless customer experience isn't going to be fulfilled.
One of the reasons platforms like Plexure are being used more and more is that they make the convergence of previously siloed data a streamlined process. In a nutshell here's how these platforms lead more cohesive data for more successful personalization...
Collect data from diverse sources
Plexure gets a massive amount of raw data – around 10k activities a second – from a wide range of sources. We've previously explained how this massive data is managed and stored, but even without knowing the technicalities of data lakes and blob storage it's obvious that dealing with this amount of data would be a stretch for smaller operators. We collect data from devices, from in-store tech, via integrations with other platforms and connected systems – and not a silo in sight.
Correct any problems
Anyone who's ever exported a spreadsheet from a marketing automation system knows why data cleansing is so important. From relatively minor annoyances like incorrectly formatted inputs to mis-mapped or completely missing data, there will always be something that needs fixing to fit your standardized format. Doing this manually when you're dealing with even a fraction of the data we see on a daily basis would be a nightmare of a job (ie, physically impossible to actually keep up with), so having data cleansed automatically is obviously preferable.
Organize and structure the data
Normalizing data is the process of organizing it to make data management and analysis more efficient, minimizing possible confusion or complications. The idea is to create a robust and flexible data structure, so it's easier both to manage the data and to query it. Imagine you're dealing with a constant stream of millions of data points that may or may not need reformatting, recombining, merging or otherwise rationalizing to meet your requirements (which may or may not change over time). Relatively speaking, it's easier to organize away the complexities, repeats and redundancies at the start than it is to deal with a mess of tangled data down the track.
Plexure collects and analyzes these millions of (well organized) customer data points and makes intelligent decisions at every stage of the customer journey using machine learning, business intelligence and cognitive analytics.
- Transaction analytics analyze data from in-store Point of Sale, mobile payments and e-commerce checkouts to understand, predict and influence consumer buying behavior.
- Digital Campaign ROI analytics measure campaign performance; collecting data on in-app activity, redemptions, tray and basket value, loyalty status, device interactions and more.
- Segmentation analytics are contextual, cognitive analytics providing real-time insights into customer segments: expected lifetime value, loyalty, buying behaviors, responsiveness, connectedness, location, media preferences and other data that's useful for our ongoing one-to-one segmentation.
Get actionable information
We use Microsoft Power BI to provide our customers with quick visual insights into our analysis of this massively complex data. There's a wealth of information tied up in these millions of data points, and it's really beyond the realm of spreadsheets and Excel charts to interpret and display. Business insight platforms make it much easier to get actionable insights from big data – something that just isn't possible with siloed data.
Use data insights to guide valuable behavior
Analysis for the sake of generating numbers and charts will never be the end goal. Retailers don't want to just know the numbers, they want to find ways to connect with their customers that result in some form of value – more visits, more purchases, higher basket value, more referrals. While most analysis can return descriptive data (icecream sells well in summer), it takes a powerful analytic platform to interpret that more basic sales information in terms of the complex behavioral and contextual data that enables far more personalizedmarketing (it's hot right now and X is 2 blocks away, send her an offer to redeem when she stops in for gelato). Understanding context is key to influencing positive (read: valuable) customer behaviors – and this real-time capability requires some real firepower.
Converging individual data vs converging segments
It's important to remember the entire point of data convergence is to bring together the distinct and disparate data we hold about individuals into a single 360 view of each customer. This isn't old-school segment convergence, where we try to find as many common data points as we can in order to form (mostly) homogenous customer segments. We've already discussed the importance of being as granular as possible for more effective personalization. Our data convergence process means we can help our customers target their segments-of-one more precisely – rather than relying on just knowing their demographic data or purchase history.
Preservation of data for future-proofing
This is one critical point when working with large volumes of data that escapes many companies and can leave you in a spot of difficulty in the future. And it's particularly important when working in a multi-channel environment. While maintaining a database that is clean and optimized is important for the health and effectiveness of your system right now, you need to keep future requirements in mind and ensure you're not optimizing in a way that throws out or disregards any data or context that is going to be useful in the future.
Many companies are now finding that insights from their current platforms are limited because historic data records either aren't complete or lack the detail to connect them to current customers or activities. This is why we store data in several different layers - with regularly accessed data in a more organized, easily accessible format and then more complete records held separately in a less structured format to allow for maximum flexibility in the future.