Last week economist Susan Dynarski published a pithy and compelling op-ed in the New York Times calling for the additional supports needed to get low-income students to and through college. The article came after a recent Atlantic Monthly piece on the importance of high school mentorship and on the heels of President Obama’s My Brother’s Keeper initiative announcing its new Success Mentors program, which will enlist over 250,000 mentors to combat chronic absenteeism. Mentorship, by all accounts, is coming back into vogue.

This is unsurprising to those of us keyed into the chronic network gaps that accompany persistent achievement gaps across the education system. As Robert Putnam recently highlighted in Our Kids—a tome on growing social capital gaps in our nation’s young people—nearly two thirds of affluent young people have access to mentoring beyond their extended family, whereas nearly two thirds of low-income young people do not.

And as Dynarski aptly pointed out in her article, relationships matter. Relationships can carry with them crucial guidance and information, not to mention emotional and even financial supports, all of which serve to support students to and through college. Similar research from America’s Promise Alliance has shown that even before college, young people are more likely to graduate high school if they have access to a web of supportive relationships, which may include parents, adults inside and outside of school, and peers. At least one stable, anchoring relationship, researchers found, can act as a gateway to this wider web of support.

But the rising recognition that new or stronger relationships can play a part curing opportunity gaps raises new questions about how we measure mentorship. Reading these articles got me thinking: should we be investing in interventions or relationships? Put differently, do growing mentorship efforts stand to sustain or disrupt the way our current education system organizes students’ networks?

In many cases like those above, we are witnessing mentorship as a sustaining innovation to the traditional education system: that is, an innovation that makes our current systems (hopefully) work better for more students. Investing in sustaining innovations can drive important improvements in performance. Just think of the incredible sustaining innovations we’ve witnessed in mobile phone and computer technology. Today, computing power that once filled entire rooms now resides in the palm of our hand.

Education reformers will often look for sustaining innovations, or interventions, that can help more students to be successful in the existing system. Dynarski’s article illustrates the point. She runs down a long list of the oft-hidden benefits that students with parents who attended college enjoy. She goes on to rightly suggest that it’s essential to make those same benefits available to low-income first-generation college students if we want to ensure their success. These supports include everything from intensive mentoring and advising, to subway passes and textbooks, to money to cover any shortfall between costs and financial aid. All of these interventions, she posits, should help more first generation students beat the odds and successfully graduate college.

We can trace similar patterns in how high schools are framing the value of mentoring programs among younger students. These programs are increasingly measured against their efficacy at helping students to graduate high school and go onto college. Mentors are essentially filling in for embarrassingly high student-to-counselor ratios in many high schools. In other words, we bring mentors to intervene in high schools in an effort to sustain the same outcomes we ask guidance counselors to drive: keep students in school, on track, and headed toward college.

Metrics matter. They define the trajectories along which we innovate. They also define the viability of business models. It’s not surprising that today most mentorship efforts join a host of other interventions we hope can move the needle on our existing metrics of high school and college success. But as such, we risk limiting mentorship to being only an intervention that might help our current systems work better, rather than a tool to fundamentally disrupt the system. We also risk belittling relationships as merely a means to an end, rather than an end unto themselves.

This by no means implies that college access and success are the wrong metrics, particularly for students we are trying to lift out of poverty. But they are admittedly narrow metrics when it comes to all of the ways relationships may matter to a student’s well-being and opportunities down the line. Focusing exclusively on mentorship as a sustaining innovation to our existing education system may in fact limit innovation in mentoring models that stand to radically expand low-income students’ access to diverse and powerful networks. Indeed, if we only see mentorship as sustaining to our existing education system, we risk underestimating the other value that relationships can offer unto themselves and underinvesting in the human element of the relationship. For example, we know that strong ties can lend important emotional supports to students even absent advice on how to get into college. We also know that ironically it’s often our weaker ties—mere acquaintances, rather than only wise enduring mentors—that can lend us vital access to new information for job prospects.

In other words, if we only invest in mentorship models that excel at meeting existing education metrics—high school graduation rates, test scores, college access and success—we may get better and better at intervening in educational systems but will not actually break open a system that lends itself to forging new relationships at scale, much less providing the same depth and breadth of useful and supportive relationships to poor students that their wealthier counterparts enjoy. This in turn is unlikely to cure chronic network gaps between rich and poor. Instead, we’d expect to see a narrowing of mentorship to deliver on just some of the value those relationships offer.

Any healthy ecosystem should contain both sustaining and disruptive innovations. Education and mentorship are no different. Although mentors may be a critical tool to improving our existing systems, they can also serve as a powerful engine to disrupt historically limited networks and connections between students and non-teacher adults. With growing recognition that relationships matter, we should be wary of metrics that narrow the project of forging relationships for students who need them most.