If you’ve opened your Gmail inbox lately, you may have seen a new “Top Picks” feature that surfaces emails from the Promotions tab into the Primary tab, based on what Google ‘thinks’ the recipient will open.
What we ‘know,’ however, is that Google is taking advantage of the massive amount of data it collects daily and harnessing it to, potentially, bring value to users of its wildly popular email service, Gmail. To learn more, let’s turn to Jeff Cheal, director of product strategy for Personalization, Campaign and Analytics at Episerver, who provides us insights into Gmail’s recent updates and what they mean for artificial intelligence (AI) fueled engagement and users’ propensity or aversion toward getting by with a little help from machine friends.
Jeff Cheal, Episerver: Google’s ability to find trends, patterns and behaviors and translate them into suggested actions could, in theory, make for a more intuitive experience. The problem, however, is the perception of intrusion. These new tools are quite helpful but could be perceived as “pulling back the curtain” to show Google’s power over a tool many would have never considered “broken” to start.
Cheal: AI has the burden of being beneficial. Proper applications of AI bring value to a person’s life by making things faster and easier such as product recommendations. Google’s use of recommendations here to suggest “Top Picks” for me to see essentially takes the “star” or “priority” away from a user and automatically moves up emails in my queue that it thinks are important for me based on my behavior. Since Google can analyze past behavior and determine patterns to predict current and future behavior, these recommendations should be valuable the minute the tool is turned on. Likewise, if Google has collected enough behavioral data (likely the case), will consumers be creeped out? Despite this, Top Picks will likely be the most successful new Gmail feature (more on others below), but it is saddled with the general expectations of a common user’s opinion of AI, that while it tries to get things right, it rarely does.
Cheal: The amount of emails that a person receives every day is overwhelming. Users could love a feature that helps them easily respond faster to the volume of content they get. The problem is the trust in the machine. Email can be such a communicative and emotional experience – inbound and outbound communication between friends, family and coworkers versus inbound, stale communication from brands – that doesn't justify auto responses in either case. Google only has a small chance to impress people; if a user sees that the response sounds fake or inaccurate, they will probably abandon this feature and fast. Consumers typically don’t have amazing patience for machine learning to get things right. It is possible, however, that Gmail “super users” will love this feature in the long run especially since companies are adopting Google’s suite of business tools.
(Image via Google)
Cheal: Have you ever complained to your loved ones about the frustration you get when someone you email doesn’t respond back? Or have you ever forgotten to email back someone because you lost track of time? If this feature can get reminders like this right through machine learning (which emails people are most likely to care about based on previous engagement rates), it could bring value to even the newest of users.
Cheal: Like businesses use retargeting to remind customers they left something in their shopping cart, the challenge for Google will be to get their timing right of how quickly or often to send those nudges. If I use Gmail for my business, a one-day lag time in responding could be huge. If I am emailing my parents who check email once a week, this could be much less important. The minute a user deems this intrusive, annoying or overbearing, they will turn it off. This is the problem with new added features I didn’t choose to add: If I never had it before, I won’t miss it if it goes away (especially if Google turns on the creepiness factor too much like, “Did you forget to respond to your mom?”).
(Image via Google)
Cheal: Brands that already have higher engagements through personalization and other forms of relevant, high-value email marketing will continue to perform higher than those that don’t. It is likely that Google will surface Top Picks from the Promotions tab to the Primary tab for brands that recipients already have a good relationship with as concluded by previous open rates and click throughs. The brands not surfacing as Top Picks could further suffer, as they sit below the fold. The Promotions tab is even less desirable with Top Picks being moved to prime real estate.
Cheal: People want email to be easy, to filter it fast, process information quickly and scan when needed. Google has accomplished most of this with a nice user interface (UI) update to their web platform that many will enjoy. The rest of these features? I’m not ready to pat them on the back yet. It’s an example of showing what tech can do that many users never asked for. While consumers are willing to provide more information to companies for a relevant experience, email may be too intimate for Google to expose just how much data and AI they can use to process what people are saying, to whom and when.
Cheal: Good AI experiences must be seamless and beneficial; the second Google suggests an email to me that they think is important to me and it isn’t, I will be searching for a way to make them stop. This is not the same problem that retail has with product recommendations. I am looking for products, thus I am more willing to be told about another product I might be interested in. Google is attempting to tell me what is most important to me out of context, and that is a huge hill to climb.
Like all new technology, end-users adopt at varying speeds. When the likes of Google start to make machine learning/artificial intelligence part of people's everyday lives, however, it slowly chips away at fear of the unknown and sets expectations that other products or services individuals interact with will learn and react to their preferences too.