Late last year, Mark Zuckerberg and his wife, Priscilla Chan, announced they are starting a private PK–8 school in East Palo Alto, Calif. Chan will serve as CEO of the new school, called The Primary School, set to open this coming fall. The effort builds on a number of investments that the couple has made to improve education in the Bay Area and beyond. But unlike their previous investments in school districts, charter schools, and technology firms, this single school stands for a different model for trying to drive outcomes: the school will provide a bevy of wraparound services, including health care, to students starting before they even enter kindergarten.

This all sounds good in theory. But history suggests that the school’s success will not merely hinge on its ability to deliver more services to poor students who need them most. Rather, for full service schools to truly drive performance, they need to embrace a deliberate structure that integrates the delivery and dosage of non-academic services in a manner that actually drives student outcomes. In our current education system, that’s often easier said than done.

The Primary School is not the first reform attempt to target the detrimental effects that poverty can have on students’ ability to learn. Over the past decade, several educational and nonprofit institutions serving low-income students have begun to attack the effects of poverty by integrating services beyond schools’ traditional academic domain to embrace the sorts of supports—mental health services, pediatric care, and mentoring—for which poverty relief advocates have long called. For example, Geoffrey Canada established the Harlem Children’s Zone to provide a “conveyor belt” of birth to college supports to students within his native Harlem. Others have extended schools’ reach into poor students’ lives by lengthening the school day, popularized by charter organizations like KIPP, or even providing boarding school options for poor students to receive round-the-clock structure, such as D.C.-based SEED schools.

These organizations have seen positive results, but face an uphill battle. The conundrum that the U.S. education system faces is that society is asking it to deliver breakthrough academic results for the highest need students, but in a world in which we still don’t understand the precise solutions that can drive these outcomes. Moreover, historically we have constrained our ability to succeed by often delivering basic services that poor students likely need—pediatric and mental health, extended day care, or vision exams, to name a few—on a fragmented or modular basis. Rarely are these services truly interwoven into the project of driving academic outcomes. And often our public policies—such as discrete social services funding streams and Health Insurance Portability and Accountability Act (HIPPA) privacy requirements—can make it tricky, if not impossible, for schools to oversee fully and deliver non-academic services in ways that would best support their students’ progress.

The Primary School would be wise to look at industries outside of education and health care that have managed to use highly integrated models to drive breakthrough performance. For example, in order to succeed in the manufacture and sale of the first mainframe computers, IBM had to integrate backward through all of the parts of the value chain of its production that were not yet well understood and established. IBM had to build the constituent parts of mainframe computers: the company designed the logic circuitry, the application software, the memory systems, and so on. Each one of those systems had to be designed interdependently with the other systems. A change in one part of a memory system might necessitate a tweak in the application software, which could in turn cause a change in how all of the pieces fit together. All of this was unpredictable, and fixing or specifying any one interface would have detracted from IBM’s iterative design and innovation process. As a result, IBM essentially had to do everything in order to do anything.

Children are not widgets; their development and success hinges on more than engineering the right circuitry or software. But those invested in curing the achievement gap should take note: to help low-income students who are underserved by existing schooling options succeed academically, our research suggests that schools must integrate backward in an interdependent way into the nonacademic realms of low-income children’s lives. A successful integrated effort is not about layering one system—health care—on top of another—education. Nor should it be about running parallel efforts to drive access to care and learning alongside one another. Rather, if IBM’s experience is any guide, controlling and tweaking the design of the entire delivery model is vital to driving successful outcomes. Only once that design reliably drives outcomes should The Primary School—and full service models in general—modularize services by partnering with third-party clinics, school providers, or community organizations and their associated funding streams.

Zuckerberg and Chan’s effort stands to be a powerful driver to answering chronic questions about precisely what sorts of health supports our highest need students need to truly drive forward their ability to learn. This will depend, in no small part, on their ability to design a school that truly integrates academic and non-academic services.