Customer Experience Comes First
Findings from the Adobe Target 2018 AI in Personalization Survey.
Image source: Adobe Stock / kras99.
Earlier this year, we on the Adobe Target team set out to capture current uses of artificial intelligence (AI) in personalization, beliefs and concerns around using it, future plans and investment in it, and more. You talked, we listened, and we heard you loud and clear: AI is the future of experience optimization!
In this post, I’ll highlight a few of the directional trends we observed from the survey results, along with some tips and information related to them. In some cases, we also describe how Adobe Target, the personalization engine of Adobe Experience Cloud, can help you more easily start using and benefit from using AI in personalization.
Where organizations are currently personalizing (using AI or not)
We started off by establishing a baseline of what companies were currently doing with personalization in general—with or without AI. From there, we wanted to determine how much automation they were applying to those efforts.
Figure 1. Areas of current personalization, with or without AI.
Looking at Figure 1, at 66 percent, the top area for personalization is website experiences. That’s probably no big surprise. Following that, exactly half say they are personalizing content, and just over a third (36 percent) are personalizing offers and mobile experiences each.
Looking a little deeper, you find two key areas that appear to be experiencing lower levels of personalization.
Last year more people visited brands from mobile devices than desktop devices, so it’s remarkable that companies are relatively far behind in personalizing mobile sites. With mobile apps, which have been around for years, it still feels like the Wild West when it comes to personalization—just 12 percent are personalizing experiences for apps.
Mobile, and specifically mobile apps, may be the most personal connection a customer has with your brand. Customers typically use your app to accomplish a specific task, and they’ve shown you who they are because they’re authenticated through the app. Mobile presents a huge opportunity. At a minimum, consider using device type and geolocation data to deliver personalized experiences on mobile.
PRO TIP: Mobile app development too hard? Adobe Target makes it easier and more intuitive than ever for non-technical marketers to test and personalize in native mobile apps without leaning on developers for support or being subject to app release cycles. Take a look at the visual experience composer for mobile apps, available in beta right now, and launching with general availability this summer.
Although our findings jibe with those of other surveys, it’s still striking that only 17 percent of respondents are personalizing recommendations for products or content. Of that number, 31 percent are in the retail/e-commerce sector, and 15 percent are from financial services. Take note, though—almost half of those that are personalizing their recommendations for products or content are making over $1B annually.
Recommendations have a long history of boosting engagement and revenue. We’ve seen that backed up by plenty of Adobe Target customers. In fact, German travel company Der Touristik gave a fantastic presentation about the company’s use of recommendations at Adobe Summit.
When asked what approaches to personalization they currently use, 42 percent say they manually curate content. That’s a lot of heavy lifting! When you dig a little deeper, 27 percent of those who manually curate content are personalizing recommendations for products and content—we might assume manually.
PRO TIP: Don’t have products to recommend, or not in retail? No problem. Use Adobe Target Recommendations to use automation to suggest articles, videos, and other media—even technical support and help articles. If you’re in B2B, recommend white papers, blog posts, videos, and other content based on what the visitor has viewed or downloaded.
Limitations to current approach to personalization
When asked about limitations to their current approach to personalization, over half cite the inability to:
- Personalize on a one-to-one deep level
- Use the data they have about their customers
- Personalize as much as they’d like and in as many locations as they’d like
However, Adobe Target has had the ability to personalize on a one-to-one level for several years, with its Recommendations capability for recommended products and content and its Automated Personalization capability for personalized offers. Last year we released Auto-Target, a capability within the A/B testing workflow of Adobe Target that lets you use AI to personalize entire experiences.
PRO TIP: Not sure which Adobe Sensei AI capability in Adobe Target is right for you? Review this handy infographic to figure out the right AI-driven personalization capability for your use case. Then read the Beyond A/B Testing whitepaper—it goes into greater detail on this topic, even giving an example of the relative levels of ROI you can obtain from each.
Although many organizations express difficulties leveraging the data they have about their customers, we’ve recently provided some guidance on how to use that data to identify important audiences for personalization and how to bring data into and push it out of Adobe Target.
Adobe Target empowers brands to personalize as much and wherever they like. The intuitive visual experience composer and three-step guided workflow make it easy for non-technical marketers to test and personalize independently, and because you can implement Target both client- and server-side, there’s no limit to where you can personalize. Really. You can personalize flavors from a “smart” drink machine, the response Alexa gives you, your website—any touchpoint that can make a web request and receive back a response.
PRO TIP: Ready to personalize everywhere? Read this blog post to learn how Adobe Target lets you personalize any touchpoint that can make a web request and receive back a response.
Perceived benefits of for using AI for personalization
When asked about the perceived benefits of using AI for personalization, survey participants cite improving the customer experience (82 percent) most; deliver better performing content, offers, and experiences (64 percent); and improving performance metrics (57 percent). Almost a third believe it would help their organization remain competitive, and around a quarter believe AI would increase work quality (26 percent) and remove repetitive or tedious tasks (25 percent). (See Figure 2 below.)
Figure 2. Areas in which the organization wants to use AI for personalization.
Only 1 percent see no benefits to using AI for personalization, which means that 99 percent see benefits to using it. That begs the question: How much are respondents investing in AI now or in the near future? As Figure 3 below reveals, there appears to be a disconnect between perceived benefits and investment in AI. Just nine percent show high levels of current investment, and only 20 percent plan for high investment in the next 18 months. What’s keeping them from investing? More on that a little later.
Figure 3. Current investment in AI for personalization versus plans for investment in the next 18 months.
Regardless of what survey participants want to use AI for, we wanted to know what opportunities they see for using it to automate personalization. The highest number of respondents — two-thirds — cite audience identification/ segmentation as an opportunity. This is followed by automation of cross-channel personalization (51 percent); personalizing content, offers, and customer experiences (49 percent); and auto-generation of content for personalization (45 percent). Again, it’s surprising that only 32 percent see the opportunity to use AI to personalize recommendations for products and content.
Barriers to using AI in personalization
We’ve all heard concerns about automation and AI putting humans out of work. Is that fear lessening? And what other beliefs or circumstances are preventing companies from diving into using AI for personalization? Figure 4 below gives some insights into what might be holding back organizations.
Figure 4. Risks, concerns and barriers to using AI for personalization.
At the top of the list, identified by 45 percent of respondents, is uncertainty about how to integrate human work processes with AI, which they perceive as complex.
PRO TIP: Getting started with AI doesn’t have to be complex. Did you know that you can turn any A/B test into AI-powered personalization with one click? Adobe Target introduced Auto-target last summer, so if you can run an A/B test, you can run AI with no other limitations.
Tied with the complexity issue at 45 percent is the age-old “lack of budget” issue. Adobe Target customer Swisscom used Auto-Target one-click personalization on its homepage and Customer Center banners and saw a 40 percent increase in clickthroughs. Sometimes it’s big uplifts in revenue or conversion rates like this that can justify allocating more budget.
PRO TIP: Feel like using AI is going to change all your work processes? Not really. Like we said earlier, give Auto-Target a try. Just set up your A/B test, give it a few experiences to consider, click Auto-Target during the A/B test setup, and then let it run.
Concerns about the maturity of AI technology and a lack of knowledge on how to train the algorithm tie for third and fourth place.
PRO TIP: Don’t know how to train the algorithm, or concerned about the maturity of AI technology? The higher the quality of data you feed the algorithm, the more predictive the model it builds.
Only 30 percent express concerns about their ability to interpret and apply AI output — what’s known as the “black box” issue of AI. Though few tools help with this, it’s actually pretty important so that you can apply that understanding to future testing and personalization activities and explain to stakeholders and leadership why AI worked.
PRO TIP: New Personalization Insights reports out in beta now and releasing to Adobe Target Premium customers this fall reveal which visitor attributes were most influential in the machine-learning model that Adobe Target built, and the model grouped customers together into the audience segments it used.
Only 17 percent are concerned about job loss from AI, so perhaps people are beginning to worry less about AI, or that concern has been overstated.
So what would respondents have to do to start using AI for personalization?
Actions to take to start using AI to personalize
To overcome these barriers and start using AI to personalize, the most commonly identified need, at 53 percent, is to centralize data sources for AI (see Figure 5). This need seems related to the earlier discussed limited ability to leverage all the data our respondents have about their customers.
Figure 5. Actions that must be taken to leverage AI for personalization.
PRO TIP: A great first step to unifying data sources is to build a data layer. What better way to do that than with industry-leading Tag Management capabilities powered by Adobe Launch? Check out this blog post to learn about it.
Getting buy-in, support for, and education on AI appears to be a significant barrier to overcome as well. More than half say that to use AI, they’d need to educate leadership about AI, educate the workforce about it (47 percent), create an organizational culture accepting of using AI (47 percent), and generate organizational support for AI (38 percent).
PRO TIP: Need to get educated about AI quickly? Listen to these two great Summit sessions: The first explains the algorithms that Adobe Target uses in Auto-Target and Automated Personalization. The second describes the future Adobe sees for personalization and experience optimization.
While only 30 percent indicate a barrier of lacking an understanding of where to apply AI (see Figure 4), when asked about what actions they’d have to take, 47 percent say they’d need to identify appropriate use cases and projects for AI.
PRO TIP: Discover how some of the world’s top brands have used AI to power personalization at scale. Watch the Adobe Summit sessions for Swisscom and Sky, and read the Swisscom customer success story.
Plans for using AI in personalization
Harvard Business Review found that organizations that personalize reduce customer acquisition costs by as much as 50 percent, increase revenue by up to 15 percent, and improve marketing spend efficiency from between 10 percent and 30 percent. Clearly personalization is critical. That’s why it’s good to see that two-thirds of respondents say that AI is important or very important to their future personalization plans.
Earlier, we noted that only 20 percent have plans for high levels of investment in AI in the next 18 months. Of course, many brands are likely to test the waters first before diving right in to AI. When you add in those with plans for moderate (38 percent) and high investment (20 percent) over that same time period, you see a much more solid 58 percent moving into AI for personalization. That’s a noteworthy increase from the current 37 percent planning moderate or high investment.
Figure 6 below shows where survey participants say their organization is currently using or planning for those investments.
Figure 6. Use of or plans for use of AI in personalization
When you combine currently using and plans to use, you see that 80 percent are using or plan to use AI for personalized content, offers, and customer experiences; 79 percent are using or plan to use AI to auto-generate content for personalization; and 78 percent are using or plan to use AI to personalize product and content recommendations. Audience identification — which actually has the highest current use of AI at 21 percent — is a close fourth with current/plan to use at 76 percent.
Take your personalization to the next level
Are you interested in taking your personalization to the next level, but don’t know where to start? Let me give you some easy next steps: Join us on the 2018 Adobe Insider Tour, coming to a city near you. At these events, you can meet personalization experts and discover more ways that companies are leveraging AI to personalize their digital touchpoints. Then learn more about how AI can help you personalize experiences by reading “AI-Powered Personalization — Above Expectation, Beyond Imagination.”
Maybe you’re already personalizing digital touchpoints in exceptional ways using AI or other Adobe Target capabilities. If so, we want to hear about it. Submit your story to our 2019 Experience Business Excellence (ExBE) awards contest. Then come and share that story with your peers at one of our Adobe Insider Tour events for a chance to win some great prizes.
Background of the survey, including select business demographic information
We ran the survey globally for slightly more than two months. The bulk (72 percent) of respondents use Adobe Target as their primary solution for customer experience optimization and personalization. Three-quarters are located in North America, with the remainder from Europe (7 percent) or Asia Pacific (17 percent). A full 84 percent are either B2C (37 percent) or both B2C and B2B (47 percent) companies. Industries with the most survey participants come from financial services (22 percent), retail (18 percent), technology (13 percent), and travel and hospitality (9 percent). Most respondents work for highly successful companies, with 63 percent of respondents coming from companies that make $150 million to $1 billion (25 percent) or more than $1 billion (38 percent) in annual revenue.