The Nonprofit Quarterly is dedicated to circulating ideas within the nonprofit sector, to testing assumptions and to moving conversations forward that might not otherwise progress. Paul Brest, the President of the William and Flora Hewlett Foundation, has taken the opportunity to throw out this salvo for response from nonprofits. He has, in his last paragraph, invited feedback; are his assumptions on the mark? You can reply to [email protected] and we will pass your comments along with or without attribution.

Sally is the CEO of a nonprofit organization. She devotes her weekends to a major home improvement project–adding a sun porch to her house. Before she started the project, she made a sketch of the finished structure, drafted a set of plans, outlined the stages of construction — from laying the foundation to putting up the frame and roof — and drew up a list of the resources necessary to undertake the project.

On Monday, when Sally returns to the office, the chances are slim that she will find an equivalent plan in her desk drawer to guide the organization’s work. Of course, she may have a plan in her head. But if Sally is like most of us, she can’t be sure how thorough or robust the plan is, or whether important pieces are missing, until she has committed it to paper, describing each step necessary to achieve the intended outcome.

Although most nonprofit organizations have a mission statement and some have a general description of their strategy, relatively few have a written plan implementing the organization’s mission and strategy. At the Hewlett Foundation, we have found it striking how often a CEO who believes that he or she has such a plan firmly in mind is unable to articulate it to a program officer. The problem is not unique to nonprofit organizations; many funders share it as well.

What is usually missing is a causal model: a clear description of how the organization plans to deploy its resources to achieve its social, environmental or other objectives. A causal model–sometimes called a “logic model”–specifies the organization’s desired outcomes, describes each step of the process necessary to achieve them and identifies milestones on the way to success. In its simplest form, a causal model looks like this:

inputs –> activities & outputs –> outcomes

Inputs consist of the organization’s human and financial resources, activities and outputs are what the organization actually does and delivers, and outcomes are the ultimate results it plans to achieve. For example, a youth mentorship program may seek the outcome of students’ graduating from high school. Its main input is the organization’s personnel, whose activities include selecting volunteer mentors and scheduling their meetings with the students. The outputs are the meetings of the mentors and students. Evaluation takes place at each stage of the process–from the selection of mentors through their meetings with students–and is ultimately concerned with the outcome: an improvement in graduation rates.

I believe that if causal models were to pervade the nonprofit sector, they would increase the efficacy of organizations many times over. Conversely, their absence is a huge barrier to an organization’s achieving its goals and to anyone’s–the organization’s CEO, board, or funders–being able to evaluate its success.

Earlier, I contrasted Sally’s plans for her home improvement project with the absence of an equivalent plan for her organization to make the point that causal models (by whatever name) are commonplace in our everyday lives, not to mention in most activities in the private and public sectors. Processes of social change are much more complex, and progress more difficult to assess, than in building a sun porch–sometimes even more complex than in building a bridge or skyscraper. But if this explains why developing plans for social change is a daunting prospect, it also indicates why it is essential. If Sally wings it with the home improvement project, the sun porch may turn out somewhat rickety but nonetheless usable. If she wings it with a social change agenda, the enterprise is more likely than not to fail miserably.

So what are the barriers to the greater use of causal models in the nonprofit sector? One obvious barrier is that many nonprofit executives are overworked, and building a causal model is time-consuming work that makes demands on their time and resources. This is true, and it is incumbent on funders to provide the resources necessary to allow nonprofits to engage in strategic planning and evaluation. But for an executive to give the organization’s other day-to-day activities higher priority than developing a causal model is like an airplane pilot’s deciding that it is more important to get off the ground and up in the air than to know where he’s going. Most of us would prefer not to be passengers on hat flight.

Second, I’ve heard it said that causal models deprive organizations of the flexibility to exploit unanticipated opportunities and challenges–sometimes arising from unintended consequences of their work–and, more fundamentally, that they are soulless and take the passion out of social change. Of course, organizations must have the flexibility to respond to the unexpected. But while it’s one thing to make midcourse corrections, it’s quite another never to have charted a course at all. Passion is incredibly important; it’s what makes those committed to the nonprofit sector go to work early and come home late. But without the capacity to move beyond passion to effective planning and execution, the nonprofit sector would be left largely with well meaning efforts that confuse good intentions with effect.

I believe that a more significant barrier to the development of causal models in the nonprofit sector is their
unfamiliarity, exacerbated by seemingly arcane differentiations among “outputs,” “outcomes,” and the like, and further aggravated by the difficulties of evaluation. If so, then we need to demystify these concepts and incorporate them into the everyday parlance of the nonprofit sector. And we also need to unpack the role of evaluation in causal models–to make evaluation both less scary and more pervasive. How can we accomplish these goals?

First, it would help to develop a common vocabulary for the concepts we have been discussing. When the Hewlett Foundation was designing a form that asked applicants to describe their proposals in terms of causal models, we looked to the available literature and noticed that, while the essential structure was the same, the models were described in somewhat different ways with different terms. The scheme we adopted–illustrated by the diagram in the text above–is a hybrid, designed to make the concepts and language as simple and accessible as possible. Though it may not be feasible to get the wonderfully diverse group of organizations that comprise the nonprofit sector to adopt a standardized vocabulary, we can at least develop a dictionary of translations.

Second, we can highlight examples of organizations’ successful use of causal models, so that people like Sally can learn from their peers. The Nature Conservancy, United Way and World Resources Institute are just a few of the organizations that systematically use causal models as the basis for their work.

Third, we can help organizations design causal models to guide their work. Foundation program officers who have carefully thought through their own programs’ strategies can provide a degree of such assistance to grantees. In addition, several entities specialize in providing capacity-building assistance, including the design of strategic plans. For example, Innovation Network offers free help on its Internet site,, and provides consultation for a fee. TheoryofChange.Org, a joint venture of the Aspen Roundtable on Comprehensive Community Initiatives and ActKnowledge, offers online guidance as well as training and workshops. Organizational Research Services is a for-profit enterprise that provides such consulting. Increasing the outreach and availability of such organizations seems a promising strategy for teaching nonprofits how to construct causal models.

Fourth, funders can require grantees to present clear and robust strategic plans. The Hewlett Foundation’s application format is centered on the proposal’s causal model and requires indicators of progress at each stage. We do not try to impose a particular substantive model on a grantee. Rather, we try to ascertain how the organization articulates its own causal model. While a number of funders, including some newer small foundations and giving circles, have similar requirements, the large majority of foundations do not. To the extent that such requirements become pervasive, they could have a significant effect on the sector.

Fifth, there are several nascent projects that seek to rate nonprofit organizations, with the aims of assisting funders in making wise investments and increasing the flow of capital to high-performing organizations. Some rating schemes focus primarily on the IRS 990 tax returns. Although such ratings take into account an organization’s administrative and fundraising expenses, reserves, and the like, they provide no information about its impact. To base a funding decision on high ratings only on these dimensions is tantamount to buying a used car at a bargain pr ice without knowing whether it runs, or purchasing an energy-efficient air conditioner without knowing whether it has the capacity to cool the room. The nonprofit sector has no small number of organizations that are efficiently run but actually accomplish little.

Though it is often difficult to assess an organization’s actual impact, a clear causal model is often the best available proxy. In any event, the causal model allows the funder and, indeed, the organization itsel f to examine each step of the process and assess the empirical assumptions underlying it . I know of at least one rating project–Donor Edge, under development by the Greater Kansas City Community Foundation and the Bridgespan Group–that takes account of an organization’s causal model. If taken to scale, whether at the community or national level, a project such as this has the potential to transform the nonprofit sector by motivating organizations to develop clear and plausible causal models.

Finally, we need to clarify the role of evaluation so that organizations see it as an aid to their operations rather than as an imposition or burden. To begin with, it is important to distinguish between the different stages at which evaluation takes place. Evaluating activities and outputs tends to be fairly simple and well within the reach of every organization. In our mentorship example, it is not difficult to keep track of the number of youths assigned mentors and the constancy and duration of their relationships. Of course, this requires some sort of tracking system and a commitment of human resources. But it is hard to imagine how the organization could function effectively without doing this.

Evaluating the mentorship program’s effect on graduation rates is much more difficult, both because the data may be harder to obtain and because good evaluation of outcomes depends on comparisons, for example, with simi larly situated students who did not receive mentoring. Indeed, the assessment of outcomes may often require resources and expertise beyond many nonprofits’ capacity. Funders committed to a particular field, such as youth development, should stand ready to pay for such evaluations and, indeed, to support social science studies that can inform the field as a whole. Independent of such studies, however, an organization should at least be able to describe how it would evaluate its intended outcomes in principle, striving where possible to use quantifiable or objectively measurable indicators. Specifying one’s objectives in measurable terms is extraordinarily helpful in clarifying those objectives — just as specifying indicators of progress at each stage clarifies the structure of the causal model.

Let me close with the observation that this essay embodies an implicit causal model. It assumes that organizations would be more effective in achieving their aims if they developed and articulated causal models, including evaluation criteria, and that a major barrier to doing this is the fact that much of the nonprofit sector has not internalized the underlying concepts and vocabulary. Each of my proposals for overcoming this barrier also makes implicit assumptions that are subject to question–for example, that many organizations in the sector would respond positively to assistance or the inducement of ratings. The essay will have been effective if it stimulates discussion of these important issues.

Paul Brest is the president of the William and Flora Hewlett Foundation.