The Better Angels of Personalization
Surveying the Diverse Software Agents Behind Targeted Experience
Every day, the digital world becomes more personalized, and we experience that change as a simultaneous increase in both utility, concerns about privacy, and at times, a bit of frustration in poorly targeted messages that seem to be hounding us. Contrary to the “consolidation” narratives that some software vendors sell on, the systems and data that make personalized experience happen are usually far from unified. In fact, many different strata of software are producing those experiences behind the scenes and it isn’t uncommon for end-users to experience those systems as fragmented and poorly coordinated. For organizations trying to provide more consistent and better orchestrated customer experience (CX), the first step is to understand what all of these systems are, how they are different, and how they run as parallel streams to inflect the user’s overall experience.
The Continuum of User Agency in Digital Experience
A continuum exists between user-directed experience and personalized, automated experience, which is far from a binary opposition. To orient ourselves, let’s think of this in terms of what TBG calls “the Continuum of User Agency in Digital Experience” (a mouthful, we know).
The point of this diagram is to examine the range between a user navigating a “vanilla” website, through designing a rough-hewn but personalized dashboard on an intranet, through complex personalizations and orchestrated multi-channel targeting that may be based on a large amount of data about the user. (Note that the model focuses on normative cases; digital is flexible enough that “anything” is possible with many of these tools.) The extent that user input drives personalization or change is one factor—how much data can support personalization and targeting rules of what level of complexity is another factor. It is a question of resolution and how proactive the personalizing systems are (which, by the way, is also corollary to how much trouble they can inadvertently cause).
Welcome to the “Personalization-o-sphere”
We’ll use this blog today as a space to quickly inventory the different types of systems that co-exist to collectively form the “personalization-o-sphere,” describe what they really do best, and what elements of experience they personalize. It is not so much a matter of “what systems inherently are supposed to perform what functions”—but rather how systems are used, what the edge cases are, and how functions overlap. Inherently, these systems overlap in many areas and produce what may ultimately be compound (and not always well-orchestrated) experiences.
These are some of the most common personalized layers:
One of the peculiarities today in assessing the “angels” of personalization is that all of our expectations have been set by the experiences provided by larger companies—Google, Facebook, Amazon, Uber, and the like—that have almost entirely custom systems (which are continuously refactored by large groups of programmers) and the ability to address and orchestrate longer-term interactions with their customers. These organizations have a huge advantage over users of commercial systems, as they can react at a meta-level to their evolving sense of customer behavior. All the other personalization systems in play today are essentially playing a kind of catchup to the leading parts of these major players’ strategies—and necessarily doing so within a great deal more constraint because of the need to support more varied, more generic use cases.
When we call our order in at P.F. Chang’s, and they ask us if we want the same lettuce wraps we had last time, we know we are in the world of CRM. Likewise, when our vet calls us to remind us to bring our dog in because it has been a while, that is CRM driving real world interaction. Getting the same sort of prompts and contextual targeting into the digital realm is the goal of marketers everywhere, and in some cases the tools associated with CRM include tools for sending targeted email or triggering other digital communications. Further, in a perfect world, CRM acts as a hub for tracking all customer interactions—and a source of data for the systems below that provide further personalization and targeting through other layers. However, those integrations are a strategic matter and contrary to what people understandably wish for, very rarely “out of the box.”
3rd Party Optimization Systems
Multi-Channel Orchestration Systems
Recently, a newer stratum of systems has emerged which are more tuned to the larger, multi-channel world of personalization and which promise a broader “orchestration” of personalized interactions across a broader range of channels. Parts of the Adobe suite make this claim; also, freestanding platforms that fall into this category, like Thunderhead, have emerged more recently. Although these systems represent some of the stronger options for organizations to create a broader field of personalized interactions, realistically, a great deal is needed operationally and in terms of technical integrations to get the full value out of these systems.
Content Management Systems
Content management systems provide what is today probably only a small, though growing part, of the personalized experience landscape. Lower end systems like Sitefinity, Kentico, and Episerver provide limited offerings, which are best suited to small amounts of page-based, non-overlapping personalized experiences, relying on very limited audience profiling. More sophisticated systems like Sitecore and Adobe (which is a content management system, but also a set of freestanding orchestration tools) have offerings which rival the best parts of the 3rd party optimizers and Orchestration Systems. They generally require a trade of a bit more up-front and development work for significant dividends in reporting clarity, governance, and unified QA—as well as allowing for a single system to manage both base and personalized content.
Marketing Automation Systems
Examples of marketing automation systems are Marketo, Pardot, Eloqua, Exact Target, and others. Although they overlap with orchestration systems to some extent, the heart of marketing automation is email, and such systems generally provide management of outbound email campaigns and communication, but also at least two levels of personalization. First, the systems allow for user behavior and characteristics (on sites, apps, and captured in CRM records) to trigger programmatic, targeted outgoing emails. Second, the content of those emails can usually be further personalized to contain supporting content that further targets the user. (Some marketing automation systems also include systems for managing and personalizing landing pages, a sort of mini-CMS).
Microservice Rules Engine
As some portion of marketing software moves to a “microservices” architecture, it becomes possible to break up the part of various systems into component parts, and theoretically mix and match vendors. It is because of this, that a vendor like Conductrics can exist as a freestanding “rules engine” providing neither the data nor personalized delivery, but just the mechanism for managing and executing rules that capture what should be shown to what user. As with all personalization platforms of any sophistication, Machine Learning (ML) also now potentially plays a role—as a new, but not always effective, way of auto-generating optimized personalization rules.
A user’s location is a key variable that can drive personalization, and this factor often reaches its high point in mobile apps, which can be highly organized around that variable. Beyond the obvious functionality of Google Maps and Uber, the idea that phones could be presenting content based on a user’s proximity to a specific venue or store represents a growing, and more pervasive use case for geo-personalization. The inherent size and mobility of phones potentially allows personalization to further align with real-world activities more readily than desktop and tablet.
Sophisticated eCommerce systems also have the potential to personalize content shown as a separate silo, for instance in the case of cross-selling based on previous purchases or browsing, as has been made famous by Amazon. Systems like Sitecore and Episerver inherently blend personalized content management with eCommerce; more sophisticated systems like Hybris contain their own provisions for personalized offerings. However, much of the personalization in eCommerce for major brands arrives served by 3rd party optimizer Monetate, which has a large market share in that area.
Another source of personalized experience is found in loyalty programs that track customer loyalty points. Such programs often act as further repository of user data—as in the famous example of Harrah’s Casino’s loyalty program which tracked the gambling of customers, delivering them a steak dinner right before they would have statistically stopped gambling as a tactic to keep them going. The same principle applies to the digital realm, where buying pattern behaviors across loyalty programs can drive personalized experience, and targeted emails—with or without centralized data via CRM.
All social media is inherently personalized; it matches one-to-one content with users, based on both gross and finely grained aspects of the user’s profiles. Inserted advertising within social media, based on those characteristics and others, constitutes a different level of personalization.
Systems such as Google Remarketing cause another level of personalized content to show up in a user’s world, in the form of advertising that is triggered by past user behavior.
Recommender Tools & Chatbots
Although they may have a limited view of the user’s history and characteristics, recommender tools and chatbots can be thought of as kinds of “rapid profiling” tools that deliver a personalized result or recommendation. In more sophisticated versions, previous interactions with the same user may have an impact on the final output of the system. With Facebook Messenger trending towards becoming a digital channel for chatbots, it seems likely that the longer-horizon personalization of chatbots will become even more intense.
Successfully Aligning the “Angels”
Probably an incomplete list, but a suggestive one. The reason for marshalling these examples is precisely to show how—for a large organization that might use between half and all of them—the opportunities for personalization are both diverse and ungovernably complex, with fragmented data silos, content regimes, presentation regimes, tracking codes, and incompatible rule sets. Bringing these elements together is the ad hoc marketer’s Swiss Army knife of the current period, but the structure is inherently ill-suited for truly coordinated long-term personalization and customer engagement. Some consolidation is possible (as in Sitecore and the Adobe suite), and some systems like Thunderhead provide unique opportunities for centralized orchestration, but for the time being, the personalization of any brand’s digital offerings is likely to be the province of more than one system, often many competing systems.
Getting these systems to work together is possible—in a limited way—through data integrations, but the varied data models, ROI considerations and complexity of integrations against rapidly changing technologies creates some limitations. In the absence of true system alignment, any successful alignment among these “angels” will only come from a strict (and in some cases, unworkably complex) human management discipline. This is the status of where we are currently with the evolution of technology, and for anyone trying to reap the benefits, it is beneficial to start from a realistic assessment of the fragmentation of all of the agents trying to personalize the user’s experience.
About the Author
Leave A Reply