Impact-Based Scaling Exercise: Move the needle by going deep, not wide

TL;DR — Behavioral health is flooded with capital and obsessed with scaling wide, yet the prevalence of addiction in our communities is barely moving. This piece argues the opposite strategy: go deep in one place, across the full spectrum of prevention, treatment, and recovery, rather than thin and wide across many.

Using St. Lawrence County, NY as a worked example, it builds a community-level scoreboard in four parts: size the need where it actually lives; measure the supply of help against it; fix the measurement basis so it captures the presence of good (recovery, connection) and not only the absence of bad (deaths, crime); and commit to falsifiable ten-year goals.

The central position: a coordinated, full-spectrum effort could measurably shrink addiction in a community over a decade to the point where the treatment system contracts on purpose and the same conditions that reduce addiction (work, connection, stability) are what make a community thrive. It's an intellectual exercise, but a serious one with real numbers and the conditions under which I'd admit where I might be wrong.

We must be bold. President John F. Kennedy said those words during his “moon speech” at Rice University in 1962. Seven years later, the boldness of the American people landed a man on the moon.

Since then, we have had various people and entities pitch their “moon shot” projects. Google even created a “moonshot factory”.

Lately, I have been wondering when America will have another bold project we can all rally behind. The nation’s 250th anniversary has me thinking about the next big thing for America. We’re in sort of a tough place. It seems impossible that we could gather around a bold vision, something that implants a desire to collectively strive for something great. But, I’m optimistic that something is there for us.

Not so long ago, I worked for an organization that had the bold vision that we could “solve addiction.” We didn’t exactly know what it meant to “solve addiction,” other than we believed that most people and communities suffered needlessly from that affliction.

When we think of substance addiction, the quirk in the equation is the prevention of use. You cannot become addicted to alcohol if you never drink alcohol. Maybe you develop a process or behavioral addiction, e.g., gambling, but then again, if you never gamble, that cannot progress to an addiction.

Why is this a quirky thing? The majority of people who drink alcohol or gamble do not develop an addiction to that thing. Therefore, society, and especially our independence and freedom-minded society, says, “Keep your damn hands off my alcohol and my gambling.”

But the negative impacts of addiction are enormous: by some estimates, up to $1 trillion a year in economic costs and over 173,000 combined alcohol, drug, and suicide related deaths in 2024, of which, roughly 80,000 were drug overdoses.

Given this enormity, mental and behavioral health has seen an influx of billions of dollars of investment capital ($3.1 billion in Q3 2021 alone); and over 60% of all behavioral health operator acquisitions were private equity backed.

It is very tempting now, to build a practice or a piece of software that scales to hundreds of thousands and millions of lives. But even as we see a decrease in the mortality rates, the extent of the problem is still significant in our communities.

What I observe happening around scale are efforts that go 20 miles wide and 20 feet deep.

Scale isn’t translating to impact.

The field defaults to scaling wide because reimbursement rewards episodic activity, not outcomes; because territoriality fragments communities; and because no one owns what happens beyond the four walls of their facility.

What we need are scaling efforts that only go as wide as our community and as deep as the problem demands. We must design a system that is the opposite of what we currently have.

The problem here becomes obvious, you cannot be in the eradication or prevention business and the treatment or recovery service business at the same time. The latter requires that people experience addiction. Held against a ten-year time-to-impact horizon, we can shrink the treatment segment on purpose without affecting access and quality of care.

But that’s a way of looking at the problem through a scarcity mindset.

Unfortunately, the size and revolving nature of the problem mean that even if you prevented one person from addiction and if you treated one person with addiction and you kept one person in recovery, 4-5 more people would enter the problem scope.

What follows is mostly an intellectual exercise. But it traces my working hypothesis that addiction is a solvable problem in our communities. I do believe that an organization or a collective of organizations could engage in a multi-year strategy of going deep that measurably neutralizes the scale and impact of addiction in that community.

This rests on a premise with a growing evidence base from social determinants of health to the ‘deaths of despair’ literature that addiction tracks environment, not just neurochemistry. Iceland’s Planet Youth model is the clearest proof it’s actionable: a generation of environmental and structural change measurably cut adolescent substance use over roughly two decades.

Impact-based scaling

If our premise holds — that depth in one place beats breadth across many — then the burden shifts onto us to prove it. "Going deep" is only a better strategy than "scaling wide" if we can say, in numbers, what deep success looks like and when we'd expect to see it. 

Scaling wide has an easy scoreboard: lives touched, clients admitted, program completions, markets entered. It is a scoreboard of activity, and it is the reason capital (including grants and philanthropy) flows toward it. Activity is legible to investors on a fund's timeline. Going deep needs a harder scoreboard, one that measures whether a place or a people are actually getting better. 

What follows explores a theoretical scoreboard for a single place: St. Lawrence County, New York, and uses it to argue that a ten-year commitment to a small set of measured goals would produce not just less addiction, but a more economically and socially thriving community.

Why St. Lawrence County, NY? It’s where I live so I can sort through the knowns and unknowns to get a more realistic analysis. And, although it’s quite homogenous, it’s rurality is not unlike most of America.

The argument below has four elements: 

1. Size the need where it actually lives

National prevalence figures are useless for designing a local solution. The relevant question is not how many Americans have a substance use disorder, but how many people here do — and at what severity. Using the federal NSDUH small-area estimates for the North Country region, applied to the resident populations of the village, county, and region, the affected population resolves to something a community can actually hold in mind.

Two features of this picture matter more than the headline counts. The first is severity concentration: the county does not look dramatically worse than New York State on substance use overall, but it runs meaningfully higher on the severe and co-occurring end: serious mental illness, co-occurring disorder, major depression. That is the profile a fragmented, single-modality, scaled-wide service handles worst, because severity and co-occurrence are exactly what defies a one-product solution.

The second is the Potsdam paradox, which is where the case against naive scaling becomes concrete. The village has two populations sharing one ZIP code: a large transient student population layered over a smaller, lower-income, year-round core. The first group's dominant risk is high-risk drinking and emerging mental illness, which is a prevention and early-intervention problem. 

[Relatedly, my wife is a professor at one of the two schools, and since 2020, the number of students experiencing mental health conditions that lead to taking semesters off is not insignificant.] 

The second's burden skews toward chronic disorder, co-occurrence, and the conditions of economic precarity — a treatment and recovery problem. A point solution optimized for either is structurally blind to the other. You cannot serve this village (or other small communities) by doing one thing well at scale. You serve it by coordinating several things across the spectrum. Most organizations get and do this, but their primary orientation is being a single-point solution or intervention.

2. Measure supply against need, not in isolation

A need figure alone tells you nothing about whether help exists. Next, we must lay out the supply of recovery infrastructure directly against the demand. We must ask not "how many people need help" but "how thin is the scaffolding around them." The NORC/ETSU/Fletcher Group Recovery Ecosystem Index (REI) scores each rural county's recovery ecosystem on a 1-to-5 scale (1 strongest, 5 weakest) from nine indicators spanning treatment access, harm reduction, peer recovery, and civic connection.

Pairing these two instruments is deliberately a juxtaposition of two independent lenses, not a single validated metric. Demand is modeled on one axis with administrative supply on the other. But this is what makes the gap visible. A community can have high need and thin scaffolding at the same time, and when it does, the implication is sharp: adding clinical seats moves one slider on the supply side while leaving the ecosystem of harm reduction, peer recovery and civic density untouched. Scaling a single service does not strengthen an ecosystem. Often, it creates a lopsided supply-side that becomes under utilized and then underfunded as being unnecessary.

Now we are exposing a deeper measurement problem that the third element has to solve. The REI counts structures — facilities per capita, distance to providers, meetings per capita. It is a census of inputs. It can tell you a county has scaffolding; it cannot tell you whether any individual standing inside that scaffolding is actually getting better. Demand instruments count need. Supply instruments count structures. Neither one counts recovery.

3. Fix the measurement basis: absence of bad and presence of good

Here is the flaw that quietly defeats most community health scorecards, including the intuitive one a thoughtful funder would first reach for. Almost every available indicator measures the absence of bad: fewer overdoses, fewer suicides, less crime, less unemployment, less homelessness. These are necessary, and they are evidence. But a scoreboard built only from them cannot distinguish a community where people are thriving from one where they have merely stopped dying. 

A place can post falling mortality and rising despair at the same time. If the goal is a thriving community — not just a less-fatal one — the scoreboard has to measure the presence of good as well.

The presence of good has two population-level signals worth instrumenting. The first is recovery capital — the personal, social, and cultural resources that actually sustain a recovery, measured longitudinally at the individual level and aggregated. 

Of course, Commonly Well does this through our Recovery Capital Index.

This is the missing counterpart to overdose mortality: where the death rate tells you how many people the system failed, recovery capital tells you whether the people still in it are accumulating the resources that predict they will stay, or better yet, leave and thrive. 

The second is social connection and civic density, or civic and associational membership, volunteer rates, participation. This is the closest population-level proxy for the environmental mechanism at the heart of our thesis: if healthy environments reduce addiction, then connection rising is the signal that should appear first, before any death rate falls.

That timing insight forces the scoreboard into tiers. Indicators do not all move on the same clock, and a flat list of eight or ten will mislead a funder or community about whether the strategy is working, because the lagging indicators — the ones everyone instinctively watches — are the last to move.

Tier 1 — Leading (expect movement in years 1–3). Age of first use; teen perception of risk; social connection and civic density; recovery-housing capacity; unemployment; household income. If the theory of change is right, these indicators will move first. If they are flat by year three, the theory is wrong for this place and the capital should be redirected before it is sunk.

Tier 2 — Proximal (years 2–6). SUD and mental illness prevalence; treatment engagement and retention; aggregated recovery-capital scores; SUD-related crime; homelessness. These are the mechanisms doing its work. We see the need falling and recovery accumulating as the leading conditions improve.

Tier 3 — Lag (years 4–10). Overdose mortality; suicide mortality; alcohol-related traffic deaths. These are the ultimate scoreboard and the hardest to read, because at a county this size they are rarer events. And often, these smaller public health departments aggregate and publish the data slower. 

Single-year counts swing for reasons unrelated to any intervention. The county's own overdose deaths moved from 12 to 5 in a single year, a change that is mostly statistical noise. These indicators are credible only as multi-year rolling averages, and even then must be reported with humility about confidence intervals.

The power of the tiered structure is that it is an early-warning system, not just a report card. It tells a funder and community when to expect each signal and therefore when the absence of a signal is real news. A flat scoreboard says "wait a decade and see." A tiered scoreboard says "if connection and first-use age haven't moved by year three, stop and rethink." That is the accountability mechanism a serious ten-year strategy requires.

Critically, the same instrument that monitors progress also sets the baseline that directs the money. Built once at the outset across all three tiers, the scorecard reveals where the gap is widest in this specific place; and that, not a prior commitment to any single tier, is what should determine where the first tranche of dollars go. 

The thesis informs what we measure; the data decides where we spend. 

This is also why the measurement layer is the first money spent, not an afterthought: for a region this size, a credible multi-tier baseline is plausibly a low-six-figure effort, a small fraction of costs plaguing the community when left unaddressed.

4. Set the ten-year goals and vision for a thriving-community

A monitoring framework becomes more than a thesis when it commits to numbers and a date. The ten-year goals are deliberately stated as absolute thresholds, not vague improvements, so that success or failure is unambiguous:

Our connective claim is that these are not separate wins. A ten-year projection shows why. When upstream conditions improve and the inflow of new cases slows, the existing prevalent population works through at its own pace, and the clinical-treatment tier shrinks not because care was withdrawn but because it becomes needed less. This is the resolution to the contradiction we identified above: the prevention arm doesn’t starve the treatment arm, it slowly makes it smaller, on purpose, and that shrinkage is the success metric, not a casualty.

A community that hits these thresholds has not merely suppressed a set of bad outcomes. It has fewer people entering disorder (Tier 1 working), more people accumulating the resources that sustain recovery (presence of good rising), a treatment system relieved of crisis-scale load (the projection's shrinking clinical tier), and — because the leading indicators that drive all of this are themselves economic and social — a population that is working, connected, and sustainably housed at higher rates than when the decade began. 

The reduction in addiction and the rise in thriving are the same phenomenon measured at different points in the causal chain. That is the thriving-community claim, and it is falsifiable: if the absence of bad improves while the presence of good stays flat, the thesis has failed on its own terms, and the scoreboard will tell us so.

Putting it all together

Almost everyone that gets into the SUD, mental health, and recovery support space as owners and operators of solutions do so with the best of intentions. Many experienced the horrors and pain of addiction and mental illness themselves. They get it and want to provide a resource that gives an offramp for more people. Sadly, the pressures of the reimbursement and funding structure create perverse incentives and misaligned models of care.

Every day, we are bombarded with the reality that the demand far exceeds the supply. But on the other side of that reality is the tension of stigma and shame, creating a psychological barrier for too many people to timely engage care.

Nonetheless, we build a practice, gain some traction, then think, “Hey, I could do this in 10 more locations.” We pitch a few investors, and now we’re off to the races – scaling wide, when in reality, we barely scratched the surface at facility number one.

We’ve made a lot of progress in the 20 years since I started working in the field. A major barrier that remains is territoriality. We are a field that has identical operations working in a single community, but one is a for-profit endeavor and the other is a non-profit organization. They may be doing the same work, but not for the same reasons. Oftentimes, consumers have no idea which is which or why it might matter to their care.

That treatment episode is just that, episodic and generally isolated. Providers have vastly improved their aftercare programming and community connectedness, but most clients are left to their own devices beyond the four walls of that provider.

The externalities of the world often impose outsized negative impact, undoing or reducing the net positive effect of the treatment received. This is a known problem among providers and a reason why many limit outcomes measurement to the end of their care – that and insurers do not incentivize longitudinal care stickiness.

We want people to get better. We create an intervention we believe works. But we do not fundamentally engage or change the environment to promote extended recovery, wellbeing, and thriving. And thus, we maintain the status quo of a revolving door of treatment and recovery. The same people cycling in and out for years.

The field is obsessed with scaling wide – mostly a symptom of the tech industry’s addiction to scaling. Scaling wide is visible through activity. Going deep in a community, however, offers the opportunity to scale outcomes and impact. 

The business model is a separate essay. But I’m convinced the “going deep” model isn’t a trade of impact against sustainability and financial success. When done right and thoughtfully, it compounds both into provable and generational impact.

Above I stated that addiction was a solvable problem but that we never quite knew what that meant. I think I do now: it’s not to cure every case, but to change one community’s conditions so that addiction loses its grip on its people and resources, and to prove it with numbers others can copy.

Commonly Well partners with behavioral health providers, community organizations, and health systems to conduct rigorous, actionable research that drives real outcomes.

Interested in similar strategy, analysis, and support for your organization?

Contact Commonly Well | Schedule Call