eCommerce brands around the world are stepping into the new age of customer experience, using marketing’s hottest technology – artificial intelligence – to go from reactive to proactive, boost conversions, and predict revenue.
We’ve assembled a detailed list of 25 of the most impactful use cases for AI in eCommerce to help you get a grip on where to focus now and in the future.
Intel & Analysis
Artificial intelligence is changing how marketers can collect customer data and drive business intelligence. This is evident in the way data is being used to inform marketing outputs in ways that marketers hadn’t been able to access (or scale) before.
1. Descriptive analytics
Descriptive analytics is the foundation of all analytics — it’s concerned with “what happened” and the basic analytics infrastructure including Google Analytics. Descriptive analytics help ensure your house is in order before tackling more advanced kinds of analysis.
2. Diagnostic analytics
Diagnostic analytics helps explain “why” things happen. Let’s say that you know your website was up 40% yesterday but you want to know why. This subset of data analysis can be helpful in explaining to the CMO/CFO/CEO why numbers are up or down without having to do significant backend work.
3. Predictive analytics
Predictive analytics leverages data mining, data modeling, and statistical models to make predictions about future outcomes. Historical and behavioral data sets plus rules allow algorithms to determine likely user responses before they actually happen.
Coupled with AI, predictive analytics identifies trends and launches campaigns likely to inspire action. Proactive marketing depends on predictive analytics – and predictive analytics has arguably the most utility and potential among any AI-related technologies today. We’ll dig into predictive marketing further down this list.
4. Driver analysis
Driver analysis is a subset of predictive analytics. It’s concerned with what makes a result happen. If you were to take all the analytics you have access to and you have an outcome you want – form fills, leads, shopping carts, checkouts, revenue, repeat customers, etc. – machine learning can help you understand what combination of variables lead to that desired outcome.
Driver analysis is the epitome of data-driven marketing because you can understand which methods ultimately lead to conversions, sales, customers retained, and the like.
Software exists today that helps us with these questions.
5. Time series forecast
Time-series forecasting can help determine what’s likely to happen in the weeks, months, or years ahead. With this method, the machine takes any data series which can be organized by time and forecasts it forward.
6. Last-touch attribution
Last-touch attribution helps identify the last link a customer clicked on and attributes previous actions to an ultimate sale. The Google Marketing Platform – formerly known as Google DoubleClick – is one example.
But, attribution models are changing, and even though last-touch attribution is a good use case today, it will eventually become obsolete as we’ll more effectively be able to attribute actions in more granular ways. For instance, position-based attribution allows us to take all touchpoints into account as opposed to just the first or last.
7. Channel selection
AI is helping marketers become much smarter in terms of channels. We’re able to find out what channel people want and don’t want to communicate on. Then, we can use channel suppression, channel selection, and channel promotion which are being automated on a 1-to-1 level.
This means, for example, you can suppress contacts from email if they don’t want to receive messages from your brand there. Self-learning machines are learning how to handle all of this automatically.
Voice, VR/AR, & Customer Service
By 2020, more than 3/4 of customer interactions with brands may be handled by AI. Let’s look at a few ways that next-gen tech is impacting retail.
8. Virtual assistants
The possibilities and potential for VAs in marketing are virtually – pun intended – limitless.
As Terry Tateossian of Forbes writes, VAs will “automate and streamline tasks [and] add a kind of appendage to our abilities to help usher us into the next phase of digital evolution.”
VAs could add a completely new dimension to online shopping by enabling consumers to speak aloud their desire and then be brought to that page, category, or product immediately.
Google Assistant, Amazon (Alexa), Microsoft (Cortana) and Apple (Siri) have paved the way, but why can’t e-commerce retailers follow suit? By 2020, 7 billion devices will have voice-powered assistants. And they’re getting better.
Okay, so chatbots are sort of a subset of VAs. But they’re making a huge impact already on the CX. Retailers and consumer brands like Domino’s Pizza are automating repeatable processes (like ordering products) by leveraging chatbots.
Chatbots, VR/AR, voice assistants, and other
kinds of technologies are improving by the day (the more experience they get,
the better they perform). In fact, 2020 could be the year that bots can outwit
humans (they’re already able to more accurately predict outcomes)!
10. Voice search
Similarly, voice-enabled search is booming in popularity among today’s consumers. By 2020, 30% of web browsing and searches will be done without a screen (Gartner).
Auditory technologies like Amazon Echo, Siri, and others can recognize spoken language and syntax, derive meaning, and not only deliver, but personalize, results. According to Gartner, the voice-first revolution will gain prominence in no time.
AI-enabled voice recognition technology can help retail brands serve the needs of customers on a 1-to-1 level, providing custom product information, delivery/shipping options, and even payment instructions.
What’s the difference between VR and AR for retail?
Augmented reality (AR) does just that – augments the existing reality with added digital components within some kind of live view (often using tech like a camera, smartphone or in-store device). Snapchat lenses and Pokemon Go are good examples of AR. Virtual reality (VR) essentially replaces and re-situates a user’s “awareness” in a 360-degree, multi-sensory, immersive experience.
11. Virtual reality
There are basically two types of emerging
VR-related use cases for retail: headset and non-headset. Both are on the
Image Source: eMarketer
It’s fair to say VR is relatively untapped (especially compared to its sister technology, AR). Brands need to understand how, where, when, and why they’d use VR – that is, a completely immersive experience – in lieu of (or in conjunction with) the in-store experience. We see this coming to fruition with mini-theaters or other types of multi-sensory or gamification experiences.
12. Augmented reality
AR has been dubbed the future of retail. In the next 10 years, retail will undergo more change than ever before. The mix of AI, VR and AR are already revolutionizing retail – smart mirrors, for example, use gesture recognition technology to superimpose certain styles on your person which may eliminate the need for dressing rooms.
Use cases for AR are almost unlimited… it will be interesting to see how this technology comes into play in the short and long term.
Content, Incentives, & Targeting
One of the biggest areas where AI will impact eCommerce is, undoubtedly, content.
13. Target new audiences for the acquisition of social
When it comes to social media marketing, it’s all about identifying and connecting with target audiences. AI makes this easier by showing you exactly where to aim and pointing out new or hidden targets that you didn’t (or couldn’t) see. That’s because AI can spot nuanced, advanced patterns in social media usage and purchasing behavior that us mere mortal marketers may not readily identify, and then AI creates new target audiences from those patterns.
This leads to an increase in the amount and relevancy of individuals you can confidently engage with your marketing. With larger bullseyes (and more of them) you’ll see greater conversion from your paid ads and targeted outreach.
14. Topic suggestions for content writing
There are a number of available AI tools that can help marketing organizations identify hot terms and topics that will attract visitors. A few of these include:
MarketMuse → AI-driven assistant for content strategists
Persado → applies “mathematical certainty” to words
Some keyword research tools are also approaching using (or use now) AI to generate hundreds of topics for marketers to consider.
No longer will marketers need to guess which topics to write about. AI adds a new level of intelligence to power more dynamic content creation.
15. Email subject lines
Companies like Phrasee are pioneering language generation for marketing copy. Subject lines, specifically, are one great opportunity to automate content creation. Subject lines are the single most important aspect of emails.
Email subject line optimization (and testing) can be handled by machines. At the least, it can supplement or augment the busy to-do lists of marketing teams who may prefer to focus on strategy, cadence, and creative aspects with email campaigns.
16. AI-assisted content creation
Dawn Papandrea of NewsCred Insights estimates revenue from content marketing will grow more than 14% through 2021. To get in on that revenue, you’ll need great content. And lots of it. Despite what your technophobic copywriters may argue, brand-worthy, loyalty-building content could soon be aided by AI. Current AI-powered software solutions like natural language generation (NLG) use machine learning to quickly turn large amounts of data into meaningful, quality copy virtually indistinguishable from what a human writer would generate.
And according to Manish Dudharejia, companies like FOX and The Washington Post are “already using [NLG] to write weather reports and sports stories.” So is it time to replace your in-house writers with AI? Probably not. But for now, handing over certain aspects of your content creation to AI will allow you to scale your marketing more easily.
17. Email send time
Send-time optimization (STO) staggers email campaign sends for each recipient to learn about preferences and sends each email at a time where the recipient is most likely to open it.
Machine learning can analyze your contacts’ behavior and identify the times when they are most responsive — independent of time zone, language, or region.
Image Source: Emarsys
STO maximizes the chances that your emails will engage your customers. Treating each of them as individuals (catering to their preferences on when to receive emails) will strengthen their loyalty and improve their experience.
18. Live content at open (OTC)
Open Time Content is an AI technology that lets marketers include the most engaging content in emails, populated at the exact time it’s opened… and offer updated content every time that email is subsequently opened.
Website, social, and product data are combined with customer behavior history and then contextualized as the email is accessed.
Open Time Content elevates automated personalization to take place in real-time as the email is opened, ensuring your content is up-to-the-moment with maximum engagement potential.
19. Content personalization and product recommendations
AI in eCommerce can handle content and product recommendations like a pro. AI goes beyond a simple rule-based system and uses business intelligence to predict what an individual customer is most likely to buy.
Image Source: Mytheresa
The machine does this based on all sorts of data including behavioral, transactional, and contextual. So, any and all recommendations (content, product, cross- and up-sell) can be personalized to a tee.
20. Personalized incentive recommendations and usage predictions
AI systems can analyze your campaign launch lists and customer database to identify which contacts are more likely to purchase when offered a certain incentive, and then assign the most efficient incentive to each one.
While some customers require higher incentives, others will buy without any. Algorithms can determine who needs what, then send the most appropriate offer.
Incentive recommendations are a simple but effective strategy that can add value and increase ROI for marketers. Marketing teams simply upload a
series of incentives (in the form of images or code snippets) to their automation platform, assigning a value to each one. Through machine learning
and artificial intelligence, AI platforms can also predict a customer’s most likely response.
Customer Lifecycle, Prediction, & Automation
AI predicts when customers are about to churn, become inactive or intend to make a purchase… and it uses embedded knowledge to automate the delivery of the best content at the best time. This is really the bread and butter of AI in eCommerce.
21. Likelihood to purchase
With a strikingly high level of accuracy, AI can predict who is likely to buy or convert. Based on past purchases and other behavioral data, self-learning systems can “sense” (to make things a little more human) who will buy.
Similarly, AI also understands which sets of customers are likely to remain inactive or defect, and can anticipate which defective contacts are most likely to return.
22. First- to second-time buyers
Many marketers are challenged to get customers buying again and again. Too often, people make one purchase from a brand and then never return. Losing customers to the abyss after they make one purchase is not an ideal lifecycle!
The solution, which AI can handle, is to identify first-time buyers who are likely to convert and encourage the second purchase with an offer. AI can also identify active buyers who are likely to convert, then provide an offer most likely to secure the purchase and increase the cart value.
In their First- to Second-Time Buyer campaign, BrandAlley saw an immediate increase in open rates, average basket value (10% increase), and revenue.
23. Likely to churn customers
Traditionally, marketers have missed out on ripe opportunities with churning segments. For example, marketers might see that a customer has gone inactive or has discontinued their subscription, and then decide they need to send an email to re-engage them. But by that time, it’s already too late.
AI flips that on its head by identifying – before the fact – who is likely to churn, and then sends them the right message(s) to prevent it from happening in the first place.
24. Next cart value
AI predicts, at a one-to-one level, what an individual’s next cart value is going to be. With AI, marketing teams can actually say:
“Customer A will likely spend $60
on her next purchase.”
“Customer B will buy every 60 days
whereas Customer C will buy every 3 weeks.”
“Customer D, who used to be a
high-value customer, is going to churn in the next 30 days unless he/she
receives offer X.”
25. Predict customer lifetime value
AI takes all data points and variables into account to determine an individual customer’s lifelong value to the business.
My favorite athletic apparel brand, for instance, with whom I do quite a lot of business would be able to take all of my data – contact info, preferences, behavior in-store, in real-time in their app and website, catalog views and buys, and all my purchases – to paint a complete picture of my anticipated profitability.
Customer lifetime value is arguably the most crucial long-term metric to get right for eCommerce marketers. Why? It helps brands understand which customers are worth more to them… so they can prioritize communications, incentives, and VIP-like treatment to these segments.
For brands on the leading edge, AI in eCommerce is helping unlock new dimensions of their marketing, optimize resources, anticipate how customer behavior will impact the business, and make decisions on what to do about it in advance.
As many forward-thinking early adopters have stated, it isn’t a question of “if” AI will augment marketing, but of “how” and to what degree. This sentiment is slowly soaking into the collective marketing sphere, too. 85% of marketers believe AI will have a “significant impact on the marketing industry” in the next five years.
As we approach the dawn of a new decade, the time is now for eCommerce and retail companies to take action and begin piloting, adopting, and integrating smart systems.