The average is a useful tool, Sara Wilson writes. But when we build systems around what is statistically typical, we routinely fail the people at the edges.
My son did not respond to his name on schedule. When we fell outside the developmental milestone chart in his pediatric clinic, a nurse smiled and said, “Well, you knew parenting was hard.” The remark was not unkind. His divergence was absorbed into a general truth, categorized, and set aside.
Years later, working with non-profits and funders across Canada, I see the same dynamic operating at scale. We build systems around what is statistically typical. And in doing so, we routinely fail the people those systems exist to serve. That failure is rarely visible to the people running the systems. It is most visible to the people who cannot get through the door. Most of us in this sector have known this for years. The harder question is what we are willing to change.
How the average became architecture
In the 19th century, statistician Adolphe Quetelet proposed the “average man,” a composite figure produced by layering population measurements. What began as a mathematical tool became an institutional standard. Developmental charts, intake forms, employment metrics, funding frameworks: all treat the average as both normal and desirable. Description became expectation. The further someone sits from the centre of the curve, the more they are asked to explain themselves.
Non-profit organizations genuinely want to serve everyone. But intention does not determine architecture. Consider your own organization. Your intake process almost certainly assumes a fixed address, a reliable phone number, fluency in formal language, and a life story that fits your categories of need. It assumes someone who can attend programming on your schedule and demonstrate progress in ways your funders recognize. The people furthest from that centre are the ones your system will consistently struggle to reach. They get labelled “hard to serve.” The label sounds clinical. What it means, structurally, is that the system has decided the problem is theirs.
The people most difficult to serve, who are almost always those most in need, become a liability in a funding model that prizes legibility.
Funders are not exempt. When we fund outcomes that aggregate cleanly, we reward programs designed for people who were already close to the outcomes we’re measuring. When we require organizations to demonstrate scale, we incentivize the smoothing of difference. The people most difficult to serve, who are almost always those most in need, become a liability in a funding model that prizes legibility. We have built a sector that is structurally better at serving the people who need it least. The people running these systems are not indifferent. They are optimizing for the incentives in front of them. Changing outcomes requires changing the incentives, and that starts with naming the problem clearly.
From servicing the centre to governing for singularity
The shift required is not primarily technical. It is conceptual. The statistical centre tells us what is common. It does not tell us what is complete. Yet we have allowed it to function as a quiet measure of human worth and built our governance models accordingly.
I want to be clear that I am not writing this from outside the problem. I have worked inside government communications offices during crises and watched us choose the story that was easiest to tell. I have sat in funding conversations and felt the pull to smooth a complicated outcome into something a donor could celebrate. I have written the report that made a program look more successful than it was, because the people it failed most were the ones least likely to be in the room when the report was presented. I know what it feels like to optimize for legibility. Most of us in this sector do. That is not a reason for despair. It is a reason to take the structural critique seriously, because the problem is not individual failure; it is the water we are all swimming in.
The organizations doing this work differently are not staffed by better people. They have built different incentives, different questions, different rooms. Here are five places to start.
Audit your intake for hidden norms
Not to confirm they are fair, but to find where they are not. Track who starts your application and disappears. Track who attends once and doesn’t return. These patterns reveal where the system is failing, not the individual. Ask which populations are consistently underrepresented in your programs, and whether that absence reflects lack of need or lack of access. The answer is almost always the latter. Then look at the form itself. Count the assumptions embedded in each field. A question like “Who is your emergency contact?” presupposes a stable social network. “What is your current address?” presupposes a home. These are not neutral questions. They are filters.
Rewrite intake as an act of welcome rather than screening and accept that this will require more staff time. That is not inefficiency. It is the cost of genuine access. One practical starting point: assign a staff member to conduct brief, non-judgmental exit conversations with anyone who disengages within the first 30 days. The patterns that emerge will tell you more about your system than any client satisfaction survey. The people who previously disappeared will stay long enough to be seen. That is not a small thing.
Redefine what counts as progress
Traditional outcome frameworks celebrate convergence toward the norm: stable housing, employment, self-sufficiency. These are valuable goals, but when they are the only measures, they render invisible many forms of meaningful change. For people whose lives unfold non-linearly, progress looks like sustained engagement, reduced harm, preserved dignity, deepened trust.
Build dual measurement systems: one set of conventional metrics for accountability, and a parallel set of edge metrics that capture relational and non-linear gains. Edge metrics might include sustained engagement beyond 90 days, reduction in crisis contacts, client-reported sense of belonging, or successful navigation of a system that previously excluded someone. Present both to funders in every report. Over time, this trains donors to see the full picture rather than rewarding only the part that was already legible to them. The goal is not to abandon accountability; it is to expand what we hold ourselves accountable for. Some funders will resist. Others will find it the most interesting conversation you have brought them in years.
Invest in relational capacity
Systems built for the centre can scale efficiently. Systems built for the edges require relationships: sustained, flexible, deeply individualized ones. This costs more per person. Boards and funders need to hear that plainly, not as an apology, but as a fact about what genuine care requires.
Create dedicated budget lines for relationship time, flexible staffing, and trauma-informed supervision. Organizations that make this investment typically see lower long-term crisis costs and stronger retention. The cost is not a design flaw; it is the design. The alternative – cheap intake with high dropout – produces impressive volume numbers and quietly abandons the people who needed the most. That trade-off deserves to be named out loud in your next budget conversation.
Change who holds authority
A board drawn entirely from the centre – educationally, professionally, experientially – will reproduce centrist assumptions no matter how equity-minded its members are. This is not a statement about intent. It is a statement about epistemology.
Revise your board skills matrix to explicitly value edge expertise: direct lived experience of disability, poverty, system navigation, non-linear life paths. Aim for at least 30% to 40% of board members to bring this perspective and offer stipends, structured onboarding, and genuine decision-making authority so that knowledge shapes outcomes rather than being filtered through consultation after the fact. People who have navigated systems from the outside carry knowledge that cannot be acquired through proximity or goodwill. That knowledge belongs in the room where decisions are made, not in an advisory circle around it.
The mechanics matter here. Stipends signal that the time has value. Onboarding that explains rather than assumes signals that you want people to be effective, not decorative. And genuine decision-making authority means that when lived-experience board members raise concerns about a program design, those concerns alter the design. If the only result is a warmer acknowledgement in the minutes, you have built a more diverse board without building a more equitable organization. Diversity of presence is not the same as diversity of power.
Resist the pressure to aggregate too quickly
Funders and boards want clean narratives: we served X people, Y percent achieved Z. That demand is understandable. At its worst, it is a mechanism for erasing the people whose stories do not resolve. Insist on including full, complex individual stories alongside your aggregate data, narratives that show progress without tidy endings.
Push for multi-year, adaptive funding models that reward persistence with complex clients rather than penalizing it. Some Canadian community foundations and United Ways have begun piloting this approach and report stronger long-term impact and deeper organizational insight. If your accountability frameworks cannot accommodate ambiguity, the problem is the frameworks. A practical ask to bring to your next funder conversation: request permission to include two or three narrative case studies in your annual report that document meaningful change without a tidy resolution. Most funders, when asked directly, will say yes. The problem is we rarely ask.
None of this requires doing everything at once. The leader reading this who is under-resourced, board-constrained, and already exhausted does not need another complete renovation plan. Pick the door that is slightly open. If governance feels impossible this year, start with the intake audit. If the intake audit feels too exposing, start by asking one staff member to track who disappears in the first 30 days and why. The data will do the rest of the arguing for you. The point is not perfection; it is direction.
Holding the edges
The average is a useful tool. It helps us see patterns, allocate resources, identify disparities. The error is not in having it. The error is in allowing a statistical centre to function as a moral one, in letting frequency become virtue, in mistaking what is common for what is complete.
A sector that makes this mistake will always be more comfortable with the people it was already comfortable with. It will interpret its own limitations as the limitations of the people it fails. It will speak fluently about inclusion while building for the included.
The organizations doing this work differently are not waiting for a better funding climate or a more permissive board. They have looked at who is missing and decided that absence is information. They have taken that information seriously enough to change something structural. Not everything. Something.
The measure of a sector is not how efficiently it serves the centre, but how genuinely it holds the edges. Somewhere in your own organization, there is someone like my son: someone whose development did not follow the script, whose difference has required explanation at every door. The average will never fully see them. It was not designed to. But you were.
