The Generalized Differentiation Strategy Model

Estimated time to read: minutes

In this series I have made the case for why a higher education institution needs a strategy. Then I highlighted the importance of understanding an institution’s competitive position within a market and, frankly, to underscore that market positioning should be paramount in any discussion of the design and launch of educational programs.

Higher Education Strategy Series: This is entry #5 of this series. This page describes the whole series and provides links to all of the articles.

While Michael Porter’s generic competitive strategies are important and useful, further work can be done to provide guidance for the leader of an institution in relation to defining its generic strategy. Similarly, this advice applies to the leaders of a program who need to define its strategy (almost certainly within the context of the institution). In this essay I define the Generalized Differentiation (GD) model. In the next one I highlight the insights it provides.

1. Purpose of this model

The GD model adds details to Porter’s framework, and the end result is that it is easier to draw conclusions that drive a leader’s decision-making related to an institution’s strategy. After a leadership team completes the analysis for a program following the GD model, leaders will have established a language for talking about the program, agreed on the target clusters the program is attempting to reach, and defined the program features that will figure prominently in the institution’s investment and marketing plans.

Note carefully that, similar to Porter’s generic strategies, a GD strategy is not a fully specified strategy; I address such a strategy in parts 9-14 of this series. A GD strategy focuses on the prospect’s decision, an intermediate step that is one piece of, but not the same thing as, the overall success that an organization hopes to achieve. This model merits time and space here because it emphasizes points that are important to the organization’s eventual success as well as blind spots for many higher education institutions.

2. The model

The mathematical model is presented in detail in my white paper on this topic. It uses the language of mathematics to ensure clarity; while true, this can also be an obstacle for comprehension. So I provide a general text description below; the interested reader should use the appendix in the white paper for a more detailed description.

2.1 Overview

This model is structured around the prospect’s decision to enroll at an institution. It is a maximization model in which a prospect chooses to enroll in a program that provides the maximum perceived value compared to all other organizations that he/she is considering.

The basic assumptions underlying the GD model are fairly unproblematic:

  • Prospects choose a program based on features of the organization.
  • The prospect is assumed to have imperfect information about the features of the organizations that he/she is considering.
  • Prospects can be usefully grouped into clusters such that all of the prospects within a specific cluster value features in a similar way.
  • Prospects choose the program that provides the most value among those organizations considered, and that value primarily comes from those features that the prospect both values the most and knows the most about.

Examples of features are cost, convenience of attending, quality of business school, quality of gym, quality of the institution’s online programs, U.S. News & World Report ranking of the institution, etc. By including cost within the feature set, this model integrates the cost leadership generic strategy within the differentiation generic strategy; that is, Porter’s cost leadership generic strategy is equivalent to a Generalized Differentiation strategy in which cost is one of the features that the institution focuses on. This change is based on the insights that prospects choose to attend institutions with varying comparative price points, the relative price is just one of the dimensions considered, and the relative price is traded off against the relative scores of other features.

It is critically important that an institution include the appropriate features within this list. This knowledge can only come by getting familiar with target prospects and how they decide between programs. Including inappropriate features in this list could lead the institution to invest in a program that has no realistic chance of succeeding; the investment would raise the cost (and, probably, the price) of deploying the program while simultaneously failing to make it more attractive to prospects. It must be understood at the beginning that investing in a process to improve the program’s knowledge of the target market should never end. Prospects and the market evolve over time, and both can change enough that what was once an attractive program can become one that is easily dismissed.

2.2 Related concepts

A program is assumed to have a limited budget to spend on features, so it must allocate this money effectively. It needs to guide its investments with its understanding of the features that target prospects value and how prospects make decisions. This raises three concepts that need further investigation: target prospects, decision rules, and value.

Target prospects

For all but the largest and most ambitious institutions, the program is not going to be designed for and marketed to the whole market. The program will focus on a subset of the market. It is generally assumed that prospects can be grouped into clusters such that all of the prospects within a specific cluster have relatively similar value profiles and minimal values across the set of features; that is, each prospect within a cluster mostly values the same features at the high end and the same features at the low end. Of course, this is not strictly true, but it helps us think about the market in a productive manner. Different organizations define different clusters and different numbers/sizes of clusters depending on how they think of the market.

This foundational integration of a program’s chosen target market into the model removes Porter’s two focus strategies from consideration; actually, other than having a different label, the analysis for whole-market strategies is the same as for focus strategies (just as it was with Porter’s initial analysis). Further, since we have already seen that a cost leadership focus is just another type of differentiation focus, this model is, at its core, an assertion of the following:

All strategies are differentiation strategies aimed at some particular target market.

Decision rules

In order to simplify the discussion and analysis, I assume that prospects are rational, choosing the program that provides the prospect with the most value. Further, I assume that the prospect uses a satisficing-type rule. This means that the prospect has separated features into two groups:

  • Satisficed: A prospect only mandates that a satisficed feature meets a minimum level of quality for the program to be included for consideration — that is, once the prospect is satisfied relative to that feature, then any further improvement of the feature would be irrelevant to the prospect. If it does not meet the minimum level, then that program would be ruled out. On the other hand, a program does not get credit for how much it exceeds the minimum level of quality for any satisficed feature.
  • Maximized: Prospects want the non-satisficed, or what we’ll call maximized, features of a program to be as high-quality as possible.


I have used the term value without clarifying it; I’ll address that now. Prospects are assumed to place a certain value on each feature. The prospect first chooses which features are satisficed and which are maximized. For the satisficed features, the prospect decides the minimum quality that is acceptable. The prospect disqualifies any programs in which the satisficed features do not meet the minimum requirement. Next, for all of the maximized features, the prospect allocates 100 value points across them all—think of allocating 100 percentage points—so that the more important features receive more points than others and the least important features receive the fewest. For example, suppose that prospect 4 has five maximized features when comparing programs. If V(4, ‘library’) = 0.05 (that is, assigning 5 value points to the library), then it can be inferred that a library is not important to this prospect.

I do not assume that prospects have perfect knowledge of all of the features of all programs under consideration. This is reflected separately by the knowledge factor in the model. Knowledge can be thought of as the percentage awareness that a particular prospect has for a particular feature at a particular program; e.g., a 0.5 would reflect that a prospect thinks he/she is only partially knowledgeable about the quality of a specific feature of a particular program. Of course, it is probable (or even certain) that a prospect does not have a full understanding of what he/she knows; that is, it is not possible to know what one does not know. This means that decisions can be sub-optimal. The institution has to work to minimize errors of this sort but cannot eliminate the possibility.

One of the jobs of marketing is to increase the knowledge of prospects within the program’s target clusters related to the quality of features that are important to those prospects. This requires that the institution know the following:

  • Who those prospects are,
  • How they can be reached,
  • The features that are maximized and those that are satisficed, and
  • The relative value that the average prospect within that cluster places on the maximized features.

The program will never have perfect answers to all of these questions, but better information should lead to better decision-making (both by the prospects and by the institution).

3 Maximization goal

All of this leads to the following equation, which we’re not going to be able to ignore, for the overall value that a prospect t places on program p :

O(t,p) = AP(p,t) ∑i=1n ((1 - Q(t,i)) × (F(p,i) - M(t,i)) × L(t,p,i) x V(t,i))

Here is a simplified verbal form of that equation:

The value that a prospect places on a program is equal to the total (across maximized features) of the quality of the feature times the prospect’s knowledge of the feature times how much the prospect values that feature. Only those programs that exceed minimum requirements on satisficed features are even considered.

Thus, if a program wants to be chosen, then the program’s leadership should focus on improving those features that prospects value the most, improving the knowledge of prospects about those features, and marketing a program to prospects who value its strengths. This figure walks the reader through this equation term by term and considers how each term affects the overall value that the prospect places on a particular program.

After determining the overall value for each program, the prospect then would choose the program with the greatest overall value.

Figure: Explanation of the variables within the Generalized Differentiation equation
  • AP(p,t): Only those programs for which the satisficed features have reached the minimum value are included in the analysis.
  • i=1…n: i is a specific feature among the n total features; thus, this is a summation over all possible features.
  • (1 - Q(t,i)): Since Q(t,i) equals 0 for maximized features and 1 for satisficed features, then 1-Q(t,i) equals 1 for the maximized features. By using this term instead of Q(t,i), maximized features are included in the calculation and satisficed features are ignored.
  • (F(p,i) - M(t,i)): This term specifies by how much a program’s feature exceeds the prospect’s minimum requirement for that feature (again, just for the maximized features). Increasing the feature quality would increase the overall value. This value can be negative if the feature is not good enough.
  • L(t,p,i): The more the prospect knows about the feature, the greater the possible overall value. If the prospect does not know about the feature, then the quality of the feature, no matter how good or bad, would have no effect on the overall value. Similarly, if a prospect knows a lot about the maximized feature but the feature quality is less than the minimum required, then this shortcoming can have a large negative effect on the overall value.
  • V(t,i): This term specifies how much the prospect values the feature. The more the prospect values the feature, the bigger the impact on the overall value.

4. Actual decision process

I want to be clear that we are not going to be doing any computations when choosing between strategies. When I am asking you to assign value points or knowledge points, I am not doing it as a prelude to calculating overall value. The value of assigning these points comes from the discipline that it places on the thinking process and, more specifically, the group decision-making process. Everyone has to put their cards on the table, making their arguments about the following questions:

  1. What are the target clusters of prospects and what are their distinguishing characteristics,
  2. Which features they think are the most important,
  3. Which are satisficing and which are maximizing,
  4. Which satisficed features are not at the needed level, and
  5. Which features prospects know the least about.

Having done this, and come to some type of compromise on all of the above, then the leadership team would look for the following problems that the above analysis should have raised:

  • Which satisficed features are not at the needed level? What are the complexity and time required to get them to the needed level? Does the size of the target cluster justify the investment? Can we communicate the improvement to the target clusters effectively?
  • Which important maximized features can you make the most improvement on, either by improving the quality or improving knowledge in the market? What are the complexity and time required to improve the quality and/or knowledge? Does the size of the related target cluster(s) justify the investment? Can we communicate the improvement to the target cluster(s) effectively?
  • Come up with a list of improvement projects from the above two bullet points. Prioritize them. Tackle a few at a time, choosing some that have shorter time horizons combined with others that have longer time horizons. Assign the necessary budget and timeline to each. Assign a project manager and an administrative or faculty champion to ensure that each gets done.

Every year this whole process should be repeated so that leadership's knowledge of the market is maintained. It might be the case that different features become more or less important so that the prioritized list has to be shuffled.


For this series, I am posing activities for an educational leader to complete. The unifying project for these activities is to define a medium- and long-term plan for competing and winning online. For all of the following, think of a specific program that your are thinking of taking online.

  1. What are the features of a program that prospects consider? Be as broad in your thinking as you can be. Also, talk with your marketing and, especially, admissions professionals to get their insights while building this list.
  2. What different clusters of prospects can you identify within your target market? Are they differentiated in any way by the features that they consider more or less important?
  3. Which of the features in the list are satisficing and which are maximizing? This might differ depending on the cluster. Again, discussing this with marketing and admissions professionals might help you in this process.
    1. What are the minimum requirements for each cluster for each satisficed feature?
    2. For each cluster, assign 100 total value points across all of the maximized features. Important features would receive more points; unimportant features would receive fewer. 
  4. For each one of the features on the list, assign between 0 (min)–100 (max) knowledge points to each one of them to reflect the knowledge that prospects have about that feature. 

We will use all of this information in the next article in this series. I wanted to give you a bit of time to gather this information before you need it for the analysis.

Feel free to reach out to me if you have any questions or comments. 

Keep Learning

Define and Act on Your Institution’s Strategy

Dr. Scott Moore has written a 15-part series on defining and acting on a higher education strategy to guide leaders during these difficult times. It is targeted at educational leaders who are participating in shaping their school's actions during and after the COVID-19 pandemic.


Dr. Scott Moore

Dr. Scott Moore is a former Principal Learning Strategist at Extension Engine. In this role, he led the global Custom Learning Experience practice. He worked with dozens of nonprofit, higher education, and learning business organizations as they considered using online learning to support their mission and margin, using his deep understanding of organizational dynamics, online learning, strategic differentiation, decision-making, and more. Prior to joining Extension Engine, he was a faculty member, administrator, and dean at Michigan Ross and Babson College for 20+ years. He holds an M.B.A. from Georgia Institute of Technology and a Ph.D. in Decision Sciences from the Wharton School of the University of Pennsylvania.

Let us teach you about learning.

We'll send you an occasional email with resources from our team of learning experts.

Subscribe to Updates