Last week, I checked into my favorite New York hotel and was greeted by the amazing front desk staff by my name. The clerk asked how my trip from Boston went and whether I needed a cab in the morning, like I usually do. In my room, there was a nice welcome note from the GM that proactively offered a wakeup call.
Just to clarify, I am not a big spender or fancy celebrity. I just stay there often enough that the hotel employees know my habits and therefore extend excellent customer service and hospitality. The staff has predicted my needs based on my prior behavior. However, that is a tiny slice of the intelligence that Netflix—which I visit often—has compiled about me.
Netflix has the ability to learn a lot about what I do when I visit the website: what I view or where I click through for more detail, how long I look at each item, my past orders, ratings, items in my queue, and other factors (including social connections) that affect my purchasing behavior. It turns all that data into a highly personalized experience for me that reflects my habits on the Netflix website and creates trust in its brand.
Image courtesy of Netflix
Websites can collect a ton of data about their users’ and their interactions with the site. Taking a cue from my favorite hotel, why aren’t we experiencing more digital hospitality as website visitors?
I think the answer is very closely related to my last article about Geo-Location, Geo-fencing and the Creep Factor; websites and devices are already collecting unfathomable amounts of data, but marketers are afraid of using it in more consumer-connected ways because people might feel like their privacy has been violated.
I agree that most consumers have no idea why or how they are served digital ads that eerily appear on their devices or browsers based on their location or searches. However, in an age of tough market competition, everyone benefits from strong customer service, which is what brands can deliver online through predictive personalization.
Image courtesy of JWT Intelligence
Predictive Personalization—the Digital Concierge
Wikipedia defines predictive personalization as “…the ability to predict customer behavior, needs or wants – and tailor offers and communications very precisely.”
Predictive personalization in the digital space uses implicit behavior (what the customer’s site activity implies about intent) to drive explicit behavior. It combines the time-tested marketing techniques of getting to know your customer, as my hotel has done, with the power of all the data that website analytics provide. JWT Intelligence named predictive personalization one of the top 10 trends for 2013.
Implicit Behavior: Is the behavior/path/actions a user takes when navigating across your site. It’s the navigation path, view duration on individual pages, search refinement and back-and-forth navigation, items in the cart, and comparing products.
Explicit Behavior: Are measurable activities such as the completion of a form, checkout, registration, and process completion.
Predictive personalization uses your visitors’ implicit behavior to evaluate:
- Where in the buying process they are, then,
- The relevant messaging needed to move them further down the sales funnel and create an explicit behavior.
When brands and agencies employ this strategy, they create online hospitality that is in direct response to their customers’ needs or wants based on those customers’ website activity. They connect in meaningful ways with their consumers in real time with highly tailored offers or content that is relevant to their real-time activities.
Why Predictive Personalization Trumps Rules-Based Models
However, many digital channels are still stuck in rules-based personalization, which is an outdated and static way to communicate with visitors.
With rules-based website personalization, marketers establish the rules that determine what customers will see (the personalization). These rules may be a search term, time of day, location, segment groups (based on customer profile attributes), content groups (such as products), and more. The rules can be applied to content display as well as other forms of internet marketing—such as who receives promotional emails about sales or discounts. But it is very basic.
For example, you visit a brick-and-mortar financial institution’s website and the site shows you offices nearby (based on location rule) and if it’s tax season, it also displays tax-related reminders and information (seasonal rule). But the site does not remember what the consumer searched for, nor does it communicate meaningfully with the consumer.
Leveraging Consumer Insights to Personalize the Experience
A consumer’s digital journey today spans across many channels and touch points, from websites to apps to social networks. In order to employ predictive personalization, we need to collect and combine as many consumer insights and implicit signals as possible. With today’s amazing computing power and database technologies, we are able to do so very affordably.
When you are thinking about your site’s next redesign, consider areas where predictive personalization could be applied, such as on the homepage sliders (areas with specials). If the consumer previously searched your website for stores in Massachusetts, you should show specials relevant to Massachusetts. If the user spent the majority of the time on the page looking for kitchen appliances, do not show ads or specials for linens.
Here are five prime examples of how brands can leverage all this data and use predictive personalization to deliver great online hospitality:
1. Create sales reminders. Evaluate the elapsed time between my shampoo purchases and as the previous purchase time frame approaches, message shampoo content to me.
2. Monitor page views to determine content display. If 85 out of 100 pages I viewed on your automotive website are related to convertibles, do not show me any content related to trucks. Also, display dealer information relevant to my location.
3. Use referral information smartly to customize content. When I come to your financial site from a 401k review site, do not push any messaging or content at me regarding home loans—show me content related to retirement plans.
4. Use search activity for cross-selling opportunities. If I previously conducted automotive research but now have stopped, this implies I am no longer in buying mode. Therefore, offer me content that’s relevant after a vehicle purchase (floor mats, auto insurance, etc.). Conversely, don’t try to sell me something that does not fit my search criteria. Once I have selected my shoe size, stop showing me shoes that are not available in my size.
5. Connect it with individualized A/B testing. If I frequently come to a site and check on a specific product but then leave the site, play with the price. Reduce it by 5 percent for me only, and see if that changes my actions on the next visit. If it does, consider evaluating your pricing.
All of this consumer behavior data can be used to group people into custom audiences to expand into broader testing. You can serve each audience profile different prices or messaging to see what works for specific audiences. Based on the results, you can then apply them to the entire audience.
Just as my hotel does so well, brands can become more helpful to their audiences through the insights provided by consumer behavior on their websites. As more and more businesses move into the direct-to-consumer space, they will need to improve their digital hospitality by predicting what consumers want based on what they already do online.
For better or worse, Google My Business (GMB) and Knowledge Graph (KG) are transforming mobile local search. It pays to watch the areas of innovation, such as hotels, restaurants and movies as these signal Google’s intentions.
Click-through rates for a business website fall with its position in organic search results. But what is the effect when organic results are pushed further and further off screen by paid ads, Google My Business listings and Knowledge Graph?
When you’re just starting out as a business owner it’s easy to become wrapped up in the seemingly endless number of metrics ... read more