In presenting the advantages of a process-based concept of personalization, this post summarizes and adapts ideas from a longer peer-reviewed piece  previously published in the journal Theory and Research in Education. Sections 1-2 lay out the whole argument (in TL;DR fashion); and sections 3-7 are “optional further reading” that elaborate the case and make further connections to practice for educational leaders.
(1) Introduction to the problem: Lack of confidence in “personalization” meaning anything in particular
“Personalization” is a popular idea in education right now. In fact, it’s so popular — and so widely used to describe diverse educational designs — that there’s some doubt that it really means anything in particular.
Consider the following critiques:
“Personalized learning has a big problem. Inside America’s schools, the term is used to mean just about anything.”– Benjamin Herold of EdWeek 
“Personalized instruction is like a chameleon.”– School change scholar Larry Cuban 
“While personalized learning (PL) may be a ‘thing,’ it is not a thing. There are so many approaches…”– Alex Hernandez of EdSurge 
Are they right?
And is the current enthusiasm for personalizing education perhaps … hopelessly confused? Is the word “personalization” a hollow, illusory rallying point?
If so, a big disappointment lies ahead for the many advocates who currently treat personalization as a shared priority in education. If “personalization” means nothing in particular: are we doomed always to squabble over an empty and unanswerable question — “What is ‘true’ personalization?” — and never to make any broadly satisfactory progress together, in a shared project?
And what a disaster this would be for research. If personalization is so loose a concept as to defy meaningful study by researchers, it will suffer from a perpetual lack of evidence.
(2) Summary of the recommended approach: A general, process-based understanding of personalization
I will propose that things are not quite so hopeless as that. While the skeptics are right that “to personalize” can indeed mean an intimidating number of things, it cannot mean just anything.
I contend that educational personalization is a real, identifiable, and researchable thing. Furthermore (although it’s far from the focus of this piece) it is also worthwhile. A variety of forms of personalization are notably effective. For instance:
- Personalized placement into pull-out tutoring programs in math yields impressive and cost-effective results. 
- Interest-based task personalization with the introduction of algebra has been shown to increase student performance even after personalization ceases in later years. 
- Personalized pacing of instruction to the rate of individual mastery has been shown over and over again to be effective. 
Moreover, a non-trivial, organizing concept for personalization is available — a concept which might bring a common meaning and focus to our current conversations in education.
What is this central concept? It’s preliminarily suggested by the verb “to personalize” itself: “to personalize” is a process.
What kind of process? Business and marketing researchers have long understood personalization as something which occurs in a kind of designable process within larger organizations or systems  . Specifically, in personalization processes, information is used about individual beneficiaries to variably output relevant results for those beneficiaries.
And this simple, organizationally-contextualized, information-centric process concept of personalization is easily transported to education.
Of course, when we speak of “personalizing education,” we aren’t talking of personalizing products or even services, in the traditional business sense.
No. Instead, we’re attempting to personalize, broadly, educational plans (that is, plans for learning) for individual students, through the individual decisions which add up to and select among those plans.
Educational personalization, then, is what happens in processes within systems or organizations (formal or informal) that … do just this: use information about individual students to output decisions related to those students’ individual educational plans.
Under this definition: those educational “decisions” generated through personalization may:
- Affect any conceivable juncture of the educational plan. (E.g.: content, objective, media, motivational strategy, etc.)
- Use any kind of student information. (E.g.: biofeedback, student self-directed learning objectives, achievement levels, etc.).
- Be of any granularity. That is, such decisions may have very long-range implications for the student’s educational agenda, very short range implications, or somewhere in between. Even a tiny educational plan can be personalized, if some part of it has been generated through personalized decision-making process. Of course, as plans are often nested within other plans, it is useful to remember that any personalized plan of limited scope may still be encompassed within the confines of a larger, unpersonalized plan.
And importantly: the personalization process need not be novel, interesting, ethical or effective to count as “truly personalized.”
Use of this technical, inclusive, process-based definition of personalization might accrue several practical advantages to designers and researchers. I’ll focus on four, to be elaborated on in a latter section. But in short, such an understanding of personalization might provide:
- Advantage 1: A concrete starting place for designers, by meeting the important question, “What kinds of things are we designing, when we design for personalization?” with a concrete answer. In large part, we are designing decision-making processes.
- Advantage 2: A basis for technical comparison across diverse designs, via use of the familiar analytical vocabulary (e.g., “trigger,” “agent,” “rules,” etc.) which is suggested by the process metaphor, which should enable better research.
- Advantage 3: Inducement to really learn from diverse design alternatives in personalization, including failed designs. A major temptation for reformers in any reform movement is to define away failed or unpopular designs as “not true examples” of the target reform. Unfortunately, because this allows for failed implementations to be swept aside as irrelevant, this tendency impedes the development of knowledge about what works and doesn’t, in the larger space around the reform. Meanwhile: a strictly technical definition, unpegged to further notions of quality or ethics, may help us overcome this temptation relative to “personalization.” It might encourage us to squarely confront and learn from diverse design alternatives in educational personalization, including instructive failures such as historical tracking.
- Advantage 4: A reframing of the initial debate, in a personalizing design project, away from lofty questions of definition, and towards more urgent ones regarding priority-setting and the concrete details of design. In the end, diverse styles of personalization may be best considered as complementary to each other within larger educational systems; there’s no need to tussle over whose is the “true” personalization, among interest-based, pace-based, program-based, etc. styles. The proposed understanding of personalization elevates more urgent tasks for discussion, such as setting priorities about what kinds of personalization to pursue first, and what personalizing processes should look like in their details.
This is the basic summary of my case for a process-based understanding of educational personalization education.
But if you’re interested in sticking around for details (do!): in the following sections, I’ll further elaborate on the appropriateness of a process-based understanding of “educational personalization.”
And I’ll start with a more in-depth treatment of the real definitional challenges faced by its advocates. If you’ve ever struggled to unite a group of educational professionals around a definition of personalization, you’ll find some sympathy ahead.
(3) Elaboration on the real differences of opinion over what counts as “personalized”
The skeptics are right that there are many different kinds of personalization. Advocates of personalization really do differ on important details about their ideal designs.
In a typical example: some advocates, like Barbara Bray and Kathleen McClaskey , prefer to reserve the term “personalized” for those designs that place students in the “drivers’ seats,” making plans and decisions about their own learning. Meanwhile, plenty of others — from the Chan Zuckerberg Initiative to Khan Academy — are content to use the term to also refer to the work of computers and algorithms in adapting lessons or materials for individual students.
A second deep point of debate concerns which parts or decision-junctures of the instructional process must be affected, for an educational design to count as truly “personalized.” For instance: some — such as Denis Littky and Eliot Levine of the Big Picture Learning Network  — hold that real personalization must affect what is to be learned by each student; say, what skills, or content. To these advocates, varying only the pacing of an otherwise one-size-fits all instructional unit to match an individual student’s time to mastery would not fully count as personalization — despite the common practice among curriculum companies, etc., to use the word “personalized” to refer to just such self-paced designs!
Still other differences in what is meant by “personalization” provocatively underscore shifts in meaning and in fashion over time. While historically, the placement of students into “tracks” and “ability-groups” was self-consciously an effort to personalize their educations, the modern Glossary of Educational Reform firmly opposes “personalization” to tracking . According to their line of thinking, schools must actually rid themselves of academic tracks (vocational tack, college prep track, etc.) if they are to “truly” personalize education. I’ll discuss this issue at length in a later section on the advantages of a broadly inclusive definition that cannot so easily define away unpopular historical implementations.
Naturally, formal definition attempts have sprung up to fill the vacuum we see here — especially as funding opportunities have attached themselves to interventions calling themselves “personalized”! These definitions, however, have their limitations.
(4) Limitations of a common exampled list approach to formal definition
Attempts at formal definition — a sampling of which is given below — have not done much to ease worries about the coherence of the concept of personalization.
“Personalized learning is tailoring learning for each student’s strengths, needs, and interests— including enabling student voice and choice in what, how, when, and where they learn—to provide ﬂexibility and supports to ensure mastery at the highest standards possible.”– Worthen, Frost, and Gentz for iNACOL 
“Personalized learning refers to instruction in which the pace of learning and the instructional approach are optimized for the needs of each learner. Learning objectives, instructional approaches, and instructional content (and its sequencing) may all vary based on learner needs. In addition, learning activities are made available that are meaningful and relevant to learners, driven by their interests and often self-initiated.”– US Department of Education 
“Personalized learning, learning that is connected to each individual’s development, background, interests, and experiences, provides an approach that broadly and equitably supports educators’ efforts to empower learners as individuals. Personalized learning offers a path to effectively support the growing diversity of the population of students by understanding how individual learners learn best and actively engage, motivate, and inspire them with the right resources at the right time, in the right medium, and at the right pace.”– Digital Promise 
“Personalized learning allows all children to receive a customized learning experience. Students learn at their own pace with structure and support in challenging areas. Learning aligns with interests, needs and skills, and takes place in an engaging environment where students gain a better understanding of their strengths.”– National Center for Learning Disabilities 
A reader wouldn’t be faulted for concluding that the most prominent thing these definitions have in common is … lists!
- ”strengths, needs, and interests”
- “what, how, when, and where”
- “learning objectives, instructional approaches, and instructional content”
- “right resources…right time, in the right medium, and at the right pace…”
One might even begin to doubt, a few commas into a new parade of examples: is there really a core idea of “personalization” that can bind all these loose lists together?
As promised, I will argue there is a core concept, and that such a despairing conclusion is unnecessary. But back to that in a moment.
Definitions which rely heavily on a smattering of examples and imprecisely-related concepts have notable weaknesses. Briefly, I’ll note three limitations:
- Apparent contradictions are easily created between definitions, even when they are written in the same spirit. Items which appear essential in one laundry-list-style definition may be absent from the next. (E.g.: Is a consideration for “where” students learn essential, as per iNACOL’s definition, but no one else’s? Is consideration for student “background and experience” definitive, as in Digital Promise’s account, but in no one else’s?). This raises some legitimate doubts about whether there really is a common concept beneath the variously-exampled term.
- It’s often unclear how many or which set of listed items within one definition are absolutely essential to that definition, and which are not. With the wide variety of items listed in even one such definition, what is the threshold set or number of those characteristics that must be met, in order for a given educational intervention or program to be considered “personalized”? Must all be met? Or just a majority? Or a key subset? It’s unclear.
- They may escape naming a central unit of analysis, resulting in our not knowing what real-world, designed “things” we are to hold up against the definition to see if they classify under the definition. For instance: is it appropriate to ask whether a whole “school” is personalized? What about a “program”? Must we focus on even smaller-grained designs, such as individual lesson plans and activities? A loose definition without a central unit for analysis doesn’t provide guidance on how to apply it against real designs.
These are serious conflicts, and serious questions to leave open! I’m sure these have plagued many a committee tasked with “personalization,” while provided with only a definition or two styled like those above.
(5) Is there a central unifying concept to “personalization”? Yes — from a process perspective
The happy news is that I believe there is a way of evaluating interventions as “personalized” or not, in education, that doesn’t involve holding up an intervention alongside a loose laundry-list of a characteristics, and trying to determine (again, often absent any more specific guidance) if it “more or less” matches up with that laundry-list.
This is because, despite my ragging on them, these “laundry list” definitions may not be quite as loose or center-less as they appear.
Once we squint past the highly specific lists of only partially-overlapping examples, I think most of us really can sense some simpler, central concept which binds these definitions together. Even if that central concept does loom frustratingly out of the frame and unarticulated.
Specifically: there is almost the sense of an underlying grammar to these definitions — as if their diverse examples have been plugged into a common underlying relational structure.
And of course, I am here to argue that that underlying structure or grammar is in the shape of … a process! Specifically, a sort of organization-bound decision-making process.
Consider, as evidence, the clear presence of at least three classic process elements in the definitions above: (1) informational input (e.g., information about the student, such as “development, background, interests and experiences,” or “learner needs”), with (2) agents (e.g., “student voice and choice”), acting on that informational input to output (3) results (here, of course, in the form of decisions about plans for learning — ranging from “learning objectives, instructional approaches, instructional content…” to “right resources, right time, right medium, right pace”).
And so: if we have common, process-inflected categories to summarize the contents of these laundry lists, and a common process grammar to put them into meaningful relation to one another … why not use that simple underlying process concept to form a new, streamlined definition?
I gave a simplified definition in the intro, just to set the stage. But getting more technical(ish), let’s try:
“Educational personalization is that which occurs in any process in which agents (who/which may or may not include the individual student beneficiary of the process) use information about an individual student to output individual decisions for them — specifically, decisions adding up to or choosing between actionable educational plans.”
I think this is a nice stab, at least. The process framework helps us get closer to some deeper idea of what we’re talking about, when we talk about personalization.
Also, in contrast to the style definitions previously discussed, this particular definition is clear about the unit for evaluation (i.e., what we are evaluating, to see whether or not it is truly personalized.) In this definition, it is not whole schools or “programs” which are themselves “personalized” in any precisely definitional sense. Instead: it is individual decisions about a student’s plans for learning which can be deemed to have been personalized or not. And these atomic decisions are personalized to the extent that they were generated through a personalizing process.
The obvious implication is that many such personalized decision-making processes might exist in a school or program, and personalizing processes might lurk sort of nested within other processes.
In fact, I suggest a little codicil to the definition as a reminder of this:
“Many such single personalizing processes may exist within larger educational organizations and systems; and they may indeed be nested within further plans and processes. To increase the personalization of an educational organization or system, then, is to increase the number and/or influence of personalizing processes within it.”
It may not be perfect. (I’ve tweaked the wording even since my original piece!) But I like to think that this process perspective helps us churn out definitions of educational personalization which get closer to the elusive “essence” of what we have so often tried to communicate principally with examples.
(6) Advantages of an inclusive, staunchly technical process-based perspective on educational personalization
I believe the point of defining things and cutting up the world in different ways is ultimately more than an aesthetic quest for “pretty” solutions. (Though of course though those are nice too, where you can get them!)
Meaning …. I suspect it’s only really worthwhile to make a new definition when that definition helps us do things better.
Accordingly, I’m only presuming to clutter the world with yet another definition of educational personalization on the strength of the belief that it can help us do things better, in both design and research on personalization.
So: what are the advantages of the proposed process-based perspective on personalization? I propose four to focus on.
Advantage 1: A concrete starting point for design — and research
First: the process approach to understanding educational personalization might help designers and researchers get started. This is because it answers a major question of designers tasked with educational personalization — namely ”What kinds of things are we designing, when we design for educational personalization?”
From a technical, process-based perspective, the answer to this fundamental question is clear: a major thing to be designed is educational decision-making processes, within larger systems or organizations, for outputting plans for individual students based on information about themselves. There’s something to be said for a definition which gives us such concrete “But what do we need to make?” guidance!
Of course, the same concreteness suggests a reasonable starting point for researchers. What is to be studied, when studying personalization? To a large degree, processes. A nice starting point.
Advantage 2: An analytic framework for comparison and learning across diverse designs
Not only does a process metaphor help us get started and bound our objects for research and design, but it also suggests familiar analytical tools for describing the further parts of any given process.
There exists familiar analytical vocabulary for describing the elements of processes which can provide a basis for direct technical comparison across diverse personalizing designs. For instance:
- We know that every process has a trigger — or a reason to begin.
- We know that every process has agents, who do the work of that process.
- We know that every process has central rule-guided activities, which begin the work of the process once triggered, and perform various operations on the relevant information fed into the process.
- We know that every process has a concrete, countable result with respect to the system or organization in which it occurs (in the case of an educational process, a decision relevant to an individual student’s plan for learning.)
With such process-analytic vocabulary available to precisely describe any individual personalizing process, we are better equipped to systematically compare diverse personalizing processes.
For example, we can quickly characterize the differences between two otherwise identical processes which use differently-sourced pieces of student information in their decision-making; or the differences between two otherwise identical processes, one of which uses a human decider as agent, and another of which uses a computer algorithm as agent.
This is obviously useful for describing different schools of thought in personalization. For instance, the distinguisher in Bray-and-McClaskey’s and Littky’s student-self-directed personalization is the requirement that students are deciding agents. Interest-based versus attainment-based personalizing processes use obviously different student information. The idiosyncratic rules and triggers of various self-paced learning implementations distinguish them neatly from one another: e.g., what level of student achievement, measured on what exact kind of assessment, merits a student moving on through the system? And so on.
The ability to put diverse processes into direct comparison to one another serves researchers interested in systematically comparing types and styles of personalizing processes towards uncovering their comparative efficacy, efficiency, etc.
And of course, a guide to the designable sub-parts of any given personalizing process is useful for designers as well. It frames a concrete list of considerations for design for any given personalizing process. (“What activates this process? What information do we use to activate the process, and to further inform the process once it starts? Who’s involved in the decision-making process? What are the guidelines of the process?” etc.)
Advantage 3: An inducement to look across diverse designs in the first place
This is perhaps my favorite advantage of the proposed definition: that it gives us inducement to look across and consider diverse personalizing designs in the first place.
The far-seeing reader will have noticed that any number of very uninteresting, unpopular, and even unethical designs qualify as “personalized” under the strictly technical definition given here.
For instance, when students get to choose their own elective classes in the comprehensive high school, that — dull as it is — is one form of personalization. After all, in a system with electives, we discern: a process using information from or about students (in this case: their interest-based choices) to output decisions about their educational plans.
So: it counts! By the same logic, letting individual students pick their summer reading books off a list is personalization. Yawn again!
These kinds of things have been around for ages! Is that really what we’re excited about and pushing for, in schools?
The association of “personalization” with such stolid designs may be objectionable to the reformer. After all: any red-blooded reformer naturally wants to distinguish their cause from what has come before. They are reforming exactly because what has gone before has not been enduringly successful or adequate!
And even more controversially: some ethically problematic designs count as “personalizing,” under the present definition.
A vivid example is the tracking of students into vocational and pre-college tracks. Historically, in many American high schools, tracking drew ineffectually (and discriminatorily) on racial and socioeconomic information about students for its decisions. Not to mention, it unnecessarily constrained students to rigid, multi-year paths difficult to escape from.
A temptation of the reformer is to carefully bound the definition of terms central to their reform so as to exclude both “boring” designs of old which they hope to transcend, and ethically objectionable instances, whether historical or contemporary.
This certainly makes sense from a rhetorical perspective. Ideally, the names of our preferred reforms would refer exclusively to novel, interesting, efficacious, and ethical instances! But what serves rhetorical purposes deeply undermines real progress on the ground.
Imagine trying to study something which as soon as it’s shown to be inefficacious or risky — or even just as soon as it’s seen as dated — no longer counts as a “true instance” of that thing. And then: having the results discounted, among interested parties, as therefore irrelevant to their cause!
That’s no way to build knowledge. It’s practically a plan to stick our heads in the sand and avoid learning about relevant issues in a reform project.
Meanwhile: an inclusive, strictly technical definition forces us to pay attention to one another, and to history. With a strictly technical definition, we can’t so easily distance ourselves from bad or unsuccessful designs — contemporary or historical. This challenges us to cohere as an inquiring community; to pay attention to and humbly learn from one another’s mistakes, rather than skip too quickly to distancing ourselves and explaining away problems via that convenient distance. It encourages fruitful collaboration among advocates of personalization, rather than just competition for exclusive definition of the term.
Advantage 4: A reframing of the initial debate in a personalization project: Not what is “true” personalization, but what exactly to make, and how to prioritize it
Finally: the present definition helps us frame better initial questions for debate, in a shared design effort, than the slippery question “What is true personalization?”
For instance, it suggests the more useful queries:
- “Which decision-junctures, in educational programs, should we personalize?”
- “For which students should those decisions be personalized, and on what triggers?”
- “What should be the further design details of those personalizing processes?”
- “How should we prioritize all the vast possibilities for personalization, within our educational programs?”
I’ll close on that final question, which is arguably the key initial question in a reframed initial debate around personalization.
After all: if personalization happens in any decision-making process which varies the plans it outputs for individual students based on information about themselves, then there are innumerable points in an educational agenda — of various granularities — into which such personalized decision-making processes could be inlaid.
Every tiny detail of a student’s educational plan — from exactly what hour of the day they participate in a given learning experience, to the tone of the facilitators’ voices, to the exact topics of deep and meaningful student interest which serve as the material: all these could, hypothetically, be selected via personalized decision-making process.
Not to mention, there are innumerable possibilities for the exact designs and styles of those processes. (E.g.: Will they be student-directed, or computer-driven, or faculty-directed? What information will be used to make the decision? What rules will guide the process? etc.)
The possibilities are staggering when confronted head on. A nod to the critics here, even if I have rebuffed their worry that there might be no essential commonality among “personalizations.”
So: what to make of this diversity of options?
In an encouraging note for the collaboration-minded reformer, it’s important to acknowledge that within the big picture of the curriculum, there’s room enough for many diverse kinds of personalization to exist side-by-side in complementary fashion. This is good news.
For instance: despite occasional efforts to ideologically oppose them, there’s plenty of reason to believe that self-directed personalizing processes could co-exist nicely alongside computer-driven personalizing processes within a larger system, simply serving different purposes for different students at different times. And motivationally-personalizing processes, on the one hand, can nicely complement personalized practice- and pacing-recommending processes on the other. They all add to the overall personalization of education. It’s just a question of where to put ’em all.
But this complementarity of diverse designs notwithstanding, the fact remains that even the most broadly appealing options for personalization are too numerous to design for all at once. And this is of course the more complicated news.
The options for personalization are infinite; but resources are limited. Each personalizing process takes time and money to think through and execute; and each personalized plan requires resources to support its carrying out. Technology does indeed chip away at this problem, but does not eliminate it.
I believe that the inclusive, contextualized perspective on personalization proposed in this piece makes us confront these real constraints regarding how much and what kinds of personalization we can achieve at any given time.
The only reasonable response under such constraints is to prioritize. Education providers have to make ethical, community-supported choices about what is worth prioritizing for personalization, in their programs.
The question of “What should we agree to prioritize for personalization?” is certainly not an easier question to answer than “What is ‘true’ educational personalization?” But, it is arguably a more useful and urgent question. It gets us closer to the real work of making good design decisions in real programs.
So: is educational “personalization” an empty rallying cry, with no core content? Is it just a popular buzzword, under the cover of which ambitious educators and organizations can sneak their own fuzzy scorecards of what makes for “good education”?
No, not quite. Or — it doesn’t have to be that way.
In this piece, I have argued that we can likely find agreement in a simple, technical definition based on a process understanding of personalization.
This process-based definitional approach leaves open the possibility that personalization can certainly refer to many things. Yet even while allowing for it to mean many things, this definition points a way towards orderly discourse and collaboration among those with different interests and priorities in personalization.
In short: a broad, inclusive definition with a staunchly technical basis might help us learn from each other’s successes and failures — and get down to the brass tacks of design for personalization in real educational programs.
To dismiss educational personalization with a shrug at its “faddishness,” and to infer from the diversity of views which present themselves under its banner that it is an incoherent concept and thus not worth pursuing, would be a grave mistake. I hope I have shown in this piece that educational personalization is a real and meaningful thing. And if — with analytical rigor and a willingness to learn from our mistakes — we can together discover and embed the right kinds of personalizing processes in the right places in our systems, then our students will demonstrably  benefit.