November 29, 2017
Predictive Intelligence in Ecommerce, Part III
November 29, 2017
Predictive Intelligence in Ecommerce, Part III
The first step for integrating PI is to make sure your website and the underlying digital data are “clean”. By this we mean that the user experience, user interface, content, and site responsiveness adhere to e-commerce site best practices. Your product catalogue should be search engine optimized (SEO) and the catalogue structure should be user friendly. On the back end, the most time-consuming process is making sure the product, order, and customer data is clean (e.g. no missing values) and well structured so that you can easily identify the online versus the in-store shopper, for instance. Also, in terms of online shopping, its critically important to tag your website with proper “events” in order to track the conversion funnel stages, sources of traffic, advertising campaigns efficiency, types of devices used to access the site, etc. For simpler scenarios, such operational and web-driven data could then be fed into a single database and be compiled into a multi-dimensional data cube which, in turn, could then be used for analysis.
However, for more involved analytical scenarios, building personalization engines brings us to one of most common problems of growing businesses: how can their existing data be organized and processed more efficiently? The problem usually appears along with the growing data volume and increasing number of data sources. All components of a retailer data solutions ecosystem should process that data extremely fast. By “fast” we mean that data storage, enrichment, evaluation and scoring, and its delivery using different channels according to complex scenarios is processed seamlessly and it does not impact user experience negatively. Luckily, there’s almost always someone around us who has seen the problem earlier and worked on developing an appropriate solution, named a “DMP”.
A Data Management Platform (DMP) allows you to collect, organize, and activate your data from any source to gain a holistic view of your consumers in order to power more efficient and effective marketing campaigns and deliver more relative content. Numerous DMP solutions are available, for instance IBM DM Services, Microsoft Azure, Lotame, Salesforce DMP and others. As of today, virtually all advanced and efficient ecommerce PI solutions interact with DMP’s in some form.
There is no “one size fits all” PI solution. With regard to PI software applications, you have a plethora to choose from: IBM Watson, SFCC Einstein, Dynamic Yield, Certona and many others. Choosing the application that provides the best fit for your organization requires an investment of time. Your existing database, enterprise applications, ESB connectivity layers and integrations play a major factor in which PI application may be most suited to your current data processing infrastructure. Part of the decision process is testing a select group of PI software solutions using test data sets and pilot environments.
Partnering with another enterprise who is experienced in matching PI solutions to your unique systems and processes is highly recommended. Though not intentional, group think tends to creep into an organization. External partners provide a fresh perspective and expert level knowledge within the ever-evolving software solutions landscape.
Big Data brings big risks. Given the numerous data breaches that have occurred over the past several years, consumers are becoming increasingly nervous about businesses and other organizations tracking their behaviours. For those of you who conduct business transactions with citizens of the European Union (EU), the EU issued a new policy known as the General Data Privacy Regulation. Not adhering to these regulatory requirements could result in fines or other consequences as delineated by the EU. At times, retailers analyze customer data in their global processing centers without stripping it of personally identifiable information; this is an example of a very clear violation of relevant EU regulations.
In terms of U.S. based consumers, transparency is a key consideration when venturing into the intense data collection and analytic methods that PI requires. Following the trail blazed by Facebook, Twitter, and Google, clearly describing how you will use consumer data is a required step for transparency purposes.
Retailers also have to be increasingly careful regarding the ways in which analytical data insights may trigger serious ethical or privacy issues with their customers. In one such very famous case in 2012, Target ran numerous tests to register useful patterns in ways to understand their buyers’ behaviours. As the company’s computers crawled through the data, analysts were able to identify about 25 products that, when analyzed together, allowed to assign each shopper a “pregnancy prediction” score. More importantly, they could also estimate her due date to within a small window, so Target could send coupons timed to very specific stages of her pregnancy. The company started to mail out coupons targeting identified women with baby items according to their pregnancy scores. One of such women turned out to be a high school student whose parents complained to Target management regarding the “offensive mailing practices” before finding out that their daughter was indeed pregnant. To mitigate possible negative effects of such mailing campaigns, Target then began to mix the specific campaign coupons with other less attractive offers.6
Additionally, you need to ensure that the data you collect is secured through robust cybersecurity methods. This takes careful planning through leveraging the advice from experts within the cybersecurity community.
Visitors who click through recommendations are 20% to 90% more likely to “add more items to their cart”[i] and over 50% of customers state they buy recommended products.[ii] Hence, the answer to “is it really important” is yes. But click through rates and boosting ROI aren’t the only important factors for deciding whether to integrate PI.
Consumers no longer simply expect that your e-commerce site is user friendly and provides accurate, personalized recommendations, they demand it. As many already are aware, the retail space is a highly competitive industry. With so many choices available, if you’re site is poorly constructed or your personalization metrics are consistently offering leather products to a Vegan consumer, it’s easy for them to find another retailer. But, it doesn’t end there. With the pervasiveness of social media, a few keystrokes from consumers voicing their disdain can greatly impact the perception of your brand (e.g. the United Airlines fiasco where activists called for the resignation of their CEO).
Suppose you decide not to go beyond the rule-based product recommendations and limit your shoppers experience only to the product offers that are manually picked and listed by your merchandisers. Will it really matter in driving shoppers to your websites? The answer is YES! A year-old Bloomberg survey7 found that when people searched for products, over 55% of them started their search on Amazon rather than Google or individual retailer websites; people frequently get frustrated with the product selection and quality of recommendations ending up going to the “Earth’s Biggest Store” for product information.
Predictive Intelligence and personalized recommendations have a larger importance outside of simply making sure that you get the right product to the right consumer at the right time. Without PI, you’ll be groping in the dark when it comes to engaging your customers and retaining their business, thus placing you behind the innovators in the retail industry who’ve adopted PI as their go to method for meeting the consumer demand for personalization that shows no sign of slowing down.
If you are ready to make a next step in enabling PI and streamlining data processing and analytics, let us start the conversation!
[i] Tastehit.com. (2016). How personalized recommendations help increase conversion. Available at: https://secure2.sfdcstatic.com/assets/pdf/datasheets/mc-nucleus-salesforce-marketing-cloud-roi-case-study-room-and-board.pdfhttps://www.tastehit.com/blog/how-personalized-recommendations-help-increase-conversion/
[ii] Tamturk, V. (2017). The ROI of recommendation engines. Available at: http://www.cms-connected.com/News-Archive/January-2017/The-ROI-of-Recommendation-Engines
6 Kashmir Hill (2012). How Target Figured Out A Teen Girl Was Pregnant Before Her Father Did. Available at: https://www.forbes.com/sites/kashmirhill/2012/02/16/how-target-figured-out-a-teen-girl-was-pregnant-before-her-father-did/
7 Spenser Soper (2016). More than 50% of Shoppers Turn First to Amazon in Product Search. Available at: https://www.bloomberg.com/news/articles/2016-09-27/more-than-50-of-shoppers-turn-first-to-amazon-in-product-search