I was in DC last January to participate in the J.Boye CMS expert group. During that event, we discussed many things and one of the topics was personalization, driven by Lars Petersen from Sitecore. The workshop involved folks from all the families - software vendors, end users, service providers and analysts - and the goal was to explore why it was hard for personalization technology to really take off. It’s true that in our industry, we’ve been talking about personalization for at least 15 years but it’s still known by many as the “phase 2 that never happened.” I don’t remember who coined that expression, but that’s so true! Indeed, for many organizations that started CMS overhaul projects in recent years, personalization was never really prioritized, simply put into the ever-growing bucket of phase 2 features and functionality, or postponed indefinitely.
Technology has been maturing and improving, and vendors, including eZ, are starting to provide very viable personalization solutions but still, at large, it is fair to say that personalization technology has generated more discussion than results. So, during this workshop, we tried to identify reasons why it’s taken so much time for personalization technology to really go mainstream and deliver on its promises.
The question Lars asked the participants was something like:
“What prevents us from delivering relevant user experiences?”
The discussion was very interesting, and quite insightful. Here are a few thoughts and takeaways from my side after participating in the workshop.
The reasons why organizations haven’t adopted personalization technology widely
We tried to identify the main causes our industry has failed at delivering on the promises of personalization. Very clearly and perhaps surprisingly to some, all of us agreed that many of the reasons are not technical. We mostly identified organizational and business reasons such as:
Marketing teams are under-resourced
Because personalization involves many things - most notably a combination of design, technology, and a solid understanding of the user journey - it takes a real effort to set up, implement and optimize a strategy. Even some of the most talented marketing teams often don’t have the bandwidth and alignment with other internal departments to drive and sustain the effort it takes to implement a personalization strategy. Marketing’s focus tends to be on short and mid-term projects and all too often the more strategic, time-intensive initiatives have to take a backseat.
Organizations are siloed
For personalization technology to deliver value, the user experience needs to be personalized across all touchpoints. Personalization should be a priority for many departments in the organization, not just marketing. Without clear alignment on the objectives, approach, implementation and optimization strategy, organizations struggle to begin bringing personalization into their online experience.
Personalization is perceived as expensive
It’s true, personalization is not a simple endeavor, and for it to be effective, organizations need a sizable collection of content - we typically say 2,000 pieces of content per site but this is, of course, not a hard and fast rule. And personalization also means conducting more analysis and, of course, implementing new technology.
But if personalization technology is implemented on a well-defined scope of products or content, it’s not necessarily expensive, and some vendors, including eZ, provide performance-based pricing which allows you to pay after the technology delivers value in terms of newly generated revenue.
Business benefits of personalization aren’t clear enough
Personalization has been raised as a buzzword, a trend, a fashion. In most cases, when you hear the word personalization, it’s discussed in broad terms and there’s not a lot of substance behind it. How many organizations do you know that are toying with the idea of a personalization project just because it’s in style, without knowing what it really means, what their goals are, what the expected benefits are and, of course, the technical requirements and constraints?
This is perhaps first and foremost the responsibility of thoughts leaders in the space and vendors like us - to more clearly convey how personalization can lead to better user experiences and better business. We, and especially us vendors, need to be more specific and more educational when we speak about what personalization is and what it can deliver. We need to avoid generalizations if we want personalization technology to be widely adopted.
Move on ‘personalized experiences,’ we’re talking about content relevance
During our workshop with the J.Boye group but also in many other situations, I hear we are moving from delivering “personalized experiences” to delivering “relevant experiences.” “Relevant” is definitely one of 2016’s buzzwords. Our industry loves to change the words we use to point at things that are actually not changing much - think ASP, Cloud and Hosting or user experiences, customer experiences and digital experiences.
Though I don’t really stick to what’s in fashion, I actually think this change in terminology is a good move. The word relevance focuses more on driving business value by improving content relevance. Building “personalized” experiences for the sake of it isn’t going to be very useful to your business. Building “personalized” experiences because they provide a way for your brand, products and services to be relevant makes much more business sense!
Still, let’s try not to mix everything up or we’ll only get even more confused by the latest buzzwords rippling through the industry. Personalization is not a synonym for relevancy. There are many ingredients that can help us build relevant experiences and personalization is only one of them. Relevant experiences first and foremost depend on clear information architecture, valuable content and the ability to really understand the context the user is in.
Let’s look at how personalization technology can play a role in delivering relevant user experiences.
Implicit vs explicit personalization
Implicit personalization means presenting personalized content to a user based on an educated guess of which content is best for them. In most cases, this is based on behavioral tracking (ie. which web pages a user visited, which products they viewed, which products they placed in a shopping cart, which topics they search and which they spend time reading, etc.). That is what we use, for instance, in eZ Personalization. That is also what Sitecore calls Predictive Personalization. This is also similar to what some advertising companies such as Outbrain are doing, but they apply these methods to a more traditional advertising business model.
There are different types of algorithms and techniques used for implicit personalization. Depending on who you talk to, one may be better than the next to optimize for specific goals, such as conversions, visit length or lead generation.
Implicit personalization can provide great results. For instance, personalized recommendations in a check-out process are an effective way to upsell customers and today’s personalization technology can quite accurately identify which kind of shopper you are and what kind of additional products would interest you. In fact, approximately 50 percent of all purchasing decisions can be predicted in advance with recommendation technology, according to research from the 2015 SysRec Conference.
There is a tendency to think that, to be relevant, you HAVE to use implicit personalization. We should not make this kind of generalization. Implicit personalization is very good for some use cases, but not for all.
There is another kind of personalization, what we used to call “explicit” personalization. It’s very simple to explain. Explicit personalization is about letting the user say to his go-to sports site, “I’m interested in rugby and I really don’t care about football. I want to see rugby news on my personalized home page, my feed and in my newsletter, and please not a word about football.” Or you might sign up for a new car rental service online and tell them, “I have no need for large family cars, I mostly rent with my wife for weekend escapes and I prefer convertibles and coupes.”
Explicit personalization enables you to provide your users extremely relevant products and content. It’s something that needs to be part of the design phase of the site, application or service - it’s not a small gadget that you just add on top. Explicit personalization requires a platform that allows you to build these kind of connections between users and content. At eZ, we provide this capability to developers in the core of our platform.
And last but certainly not least, personalization, whether it is implicit or explicit, is not the only way to be relevant. We should not forget that relevance is and has always been a core mission of information architecture. Through proper information architecture, we can provide users with a relevant online experience, and we can do it without using any demographic, behavioral or personal data whatsoever.
If being relevant can be achieved in different ways, using implicit or explicit personalization or other recommendation techniques, the great news is that none of them are exclusive and the best example for that is probably the first reference that comes to mind in this domain: Amazon. The master of product recommendations, Amazon provides a very deep and complex set of techniques in order to optimize their business, and doesn’t hesitate to blend different approaches on its pages.
Personalization can generate clear ROI for performance metrics
Looking at it in general, benefits from personalization can seem ambiguous and indirect. In our J.Boye workshop, a lack of clear business benefits was identified as a factor. But if you focus on specific performance metrics such as revenue generated, the benefits can actually be extremely clear, measurable and direct.
At eZ, we use personalization technology for product and content recommendations, in different use cases such as e-commerce and publishing. In the case of e-commerce, it’s very simple to measure the impact of personalization on KPIs. Because of that, we can apply a performance-based pricing. We pay ourselves from a share of the additional revenue our solution generates for the customer.
Of course, some other cases are much more difficult to evaluate. For instance, using personalization to increase lead generation is far less obvious. It’s relatively easy to evaluate the increase in lead generation thanks to personalization, but it’s always difficult to evaluate the quality of the additional lead flow generated, especially when the lead-to-purchase journey is long and complex. As we all know, the number of leads generated is a great metric but the real KPI we should look at is the final business generated.
Tread carefully because if used improperly, personalization can hurt
When people shop in their favorite supermarket, they like to know where to go to find what they’re looking for. They don’t want the store’s organization to change. That’s why retailers try to replicate store plans in all their stores, to provide a familiar and smooth user experience, and to reinforce their brand. The Apple Store is probably one of the best examples. Every detail is uniform across their stores worldwide, making you feel right at home whichever Apple store you’re in.
Digital experiences are very similar to physical experiences - users like them to be predictable. This is also a foundation of good information architecture. But one issue with personalization is that, if used improperly, it can make websites, online services and apps less predictable and sometimes downright confusing. And to tell you the truth, it can be easy to fall into this trap, as very few of the elements of context used for personalization are truly reliable. There’s room for many mistakes, especially when you’re relying on pixel tracking alone.
Imagine you’re a store owner and you had an algorithm that shuffles the layout of your store for each customer based on their age, gender and marital status so a 35-year-old single male who regularly buys imported beer would find beer and wine right at the entrance (and not in the back as usual). Then one day this 35-year-old single man walks into a different store dressed as Mary Poppins for Halloween. If this store had the same personalization technology, our friend might be welcomed into the store by dresses, handbags and cosmetics.
This is of course a ridiculous example, but that is what can happen in digital life when personalization is not used properly. The exact same thing can happen to your home page simply because the indicators we rely upon to identify a user’s preferences aren’t fully reliable. Personalization is not an exact science and it's sometime hard to predict what users want or need, as Karen McGrane reminded us during her talk about adaptive and predictive content at the eZ Content Technology Conference 2015.
So there are places where personalization makes sense, and there are places where it doesn’t make much sense but no matter what, personalization techniques aren’t fool proof. Personalizing product recommendations based on on-site behavior usually works well, but what if a male user spent two weeks on your online shop searching for a dress for his sister. Do you really want to keep recommending dresses and related clothing after he’s purchased the gift? Be sure to take this into account when planning your personalization strategy.
It’s all about information architecture
Most of what we say above revolves around one thing: personalization can provide a lot of value, but it needs to be treated as part of your information architecture and content strategy. Information architecture as a discipline encompasses personalization. Information architecture includes organizing content, the way it’s discovered and the way it’s understood - spot on with personalization (If you want to learn more about the information architecture discipline, I’d simply recommend the 4th edition of the O’reilly reference book on the topic).
So if you treat personalization as something of a different nature, not related to information architecture, you’re going to get it wrong. Personalization isn’t an extra feature your vendor brings to the table that you just turn on, like adding a turbo booster to your car..
Personalization technology isn’t without challenges
During the J.Boye workshop, points were raised about the readiness of marketers to take on a personalization strategy effectively. It’s a fact that it’s more complex to deliver personalized content than a single stream of content for all users, or a few streams for different industry verticals or personas.
Personalization puts pressure on many facets of the organization, and there are a great deal of challenges above and beyond the recommendation block on your home page. Just think about issues related to:
- Site performance
- Accuracy of content and product recommendations
- Tracking of user behavior
- Privacy concerns
We, the vendors, have made a lot of progress in all of these aspects but it remains complex to serve personalized content online. Vendors tend to focus on the bright side of things (mea culpa, this includes eZ).
Looking to implement personalization technology?
Personalization is a sophisticated, complex topic and you need to be prepared if you’re going to purchase and implement it. So be informed, understand the topic and before making your digital infrastructure more complex, make sure you have a clear view of how your personalization strategy aligns with your information architecture and how personalization will support your overall business strategy.
If you have personalization on your roadmap for 2016, instead of rushing to your software vendor to buy the technology, make sure you have a solid grasp on what you want to achieve first. Be sure to have a strong UX expert and Information Architect in your team who will drive the project to integrate personalization into the core of your content management platform. If not, start to build such a skillset in your team or consider finding a partner who can help.
Then, don’t think about personalization as a massive thing. Start small, introduce it where it makes the most sense, analyze and experiment, and then expand to other areas of your site. For instance, if you have e-commerce on your site, start by simply adding personalized product recommendations at check-out. Once you validate this use case, you can extend and introduce it deeper in the user experience such as on product category pages.
I have to give credit to the J. Boye community. Those kind of events where vendors, users, service providers, analysts all sit at the same table and discuss the opportunities and issues of our industry in a non-commercial setting are invaluable. I’d recommend anyone working closely with content management technology to join the community. It’s a great community to be in, and the US expert group is a great group of people.
As for personalization, my prediction is that, when it will be widely adopted, it will be transparent, just a core part of the content delivery platform. Personalization is a feature, it is in no way a product. Then maybe we will stop talking about it as “the phase 2 that never happens.”
I hope this blog post was relevant, it surely wasn’t personalized! Please feel free to comment below if you want to add to the story.