Stephanie Bates traces a decision most operators made for good reasons: adding an operations manager to every clinic, and giving every employee two bosses.
Stephanie Bates has a word for the territory most operating decisions live in, and behavior analysts do not like it. The word is squishy. Real life, she told a Boston audience this month, does not have rigid edges. Her audience is trained in a science built on clean ones.
Bates spent years on the clinical side of ABA companies before moving into the business side, and now consults independently. Her account of how the field arrived at its current management structure is unusually specific, and it begins with a decision nearly every growing provider made for defensible reasons.
How ABA Clinics Ended Up With Two Managers
Clinics historically ran on a clinical director, usually a Board Certified Behavior Analyst, who also handled scheduling, authorizations, callouts, and whatever else arrived that morning. It was too much, and organizations fixed it the obvious way: they added an operations manager to every clinic. The clinical director got her clinical job back. Every employee got two managers.
What went unwritten was which manager owned what. In practice, Bates says, staff resolve that by asking whoever answers fastest, which means the same question can get two answers in a week and neither manager knows the other one weighed in. The consequence she emphasizes is not conflict but opacity. “There’s no longer a cohesive focus,” she said. “We don’t know where we’re going half of the time because we’re not always working in the same direction, and we grow bigger and bigger and bigger blind spots.”
She borrows a line from a colleague speaking elsewhere at the same event, which she says explains the whole pattern: our systems are perfectly designed for the outcomes they give us. The formulation is a quality-improvement adage usually credited to Paul Batalden, who adapted it from a Procter and Gamble colleague, and it has been carried into healthcare for decades. In her reading, ABA organizations designed themselves for a clinical silo and a business silo, then expressed surprise at the resulting animosity. Nobody intended it. The structure produced it.
Why Siloed Decisions Move Problems Instead of Solving Them
Bates is blunt about what separation does to decision quality. “When we operate in silos, we are not solving problems; we are moving problems,” she said, because each function optimizes for its own outcome and books the win. Her example is a billable-hours requirement rising from 24 to 28 because the business side prevailed, with nobody in the room asking what it costs on the other side of the wall. She notes the clinical side does exactly the same thing when it wins, and that clinicians pursuing research-grade intervention in a practice setting can create financial problems just as blindly.
Her illustration of the blind spot is a process every ABA employee touches and almost none can describe. The authorization cycle runs through every department in the company, and Bates estimates that 90 percent of employees cannot walk it start to finish. She is skeptical that the obvious exceptions can either: the analysts know their piece, the insurance team knows the output, and neither sees the whole. This is the same structural gap that shows up when organizations scale, where the rules that separate providers who scale cleanly from those who do not get discovered rather than designed, and it is why the largest platforms can operate at four-figure clinic counts and still struggle to answer basic questions consistently across them.
Why Behavior-Analytic Training Makes This Harder
Part of what makes the structure hard to fix is that the people inside it were trained to distrust ambiguity. Behavior analysts come out of a discipline built on controlled conditions and clean measurement, and the operating decisions in front of them have neither. Bates is careful about what she is not saying: embracing the squishy does not mean discarding the science or lowering the standard of evidence. It means accommodating more variables than a study design allows.
Her example is a clinician who intended to lift an intervention directly out of a research article and run it in a classroom, when the published procedure specified a student alone in a room with a table and an instructor. The procedure was sound. The conditions that produced the result were not available, and treating the article as an instruction manual guaranteed a failure that would then be blamed on the intervention rather than on the transfer.
The same instinct shapes how organizations handle being wrong. Bates announces her own mistakes deliberately, including replying to all fifty people on a thread to say she got it wrong, because staff will not surface errors in a culture where nobody senior models it. Comfort with mistakes, in her framing, is a precondition for confidence rather than the opposite of it, because people who cannot afford to be wrong will not commit to a decision at all.
The Tools Bates Recommends for Ambiguous Decisions
Her remedies are procedural, which is what makes them usable. For decisions with no obvious owner, she builds impact matrices and decision trees: how many departments does this touch, how many measures does it move, and what is known to be inherently urgent. Gaps in authorizations, she notes, are always urgent. She is explicit that the tools are not algorithms and have to be taught flexibly, because the skill is context-dependent and only develops through repetition.
She also wants the affected parties present, and identifies a specific politeness that keeps them out. Colleagues routinely tell her they did not want to clutter her calendar with a meeting, when the thing being decided was her responsibility. Excluding the affected party to protect their schedule is how organizations manufacture the decisions nobody can explain later.
Before a decision reaches a group, she recommends a smaller rehearsal: take it to one trusted colleague, talk it through, and ask whether it holds together, so the version arriving in front of six people from three departments has already survived a first pass. The point is not approval. Reasoning out loud once exposes the step you skipped.
For confidence, she recommends rehearsal, and practices what she prescribes. She role-plays hard conversations with an executive who knows she says yes when she means no, throwing difficult questions at her and debriefing afterward on what landed. The measure of success was the day that executive called to say she had not needed to practice first.
The cost of getting this wrong scales with size in one direction and severity in the other. Large platforms can absorb a bad decision routed to the wrong manager; practices that shut their doors under administrative weight rarely leave a public record of what actually broke, and the industry learns nothing from them. Meanwhile the senior hires arriving to fix these structures are increasingly operating and financial executives hired away from other industries, which treats the symptom at the top of the organization and leaves the ambiguity in every clinic untouched.
Debriefs, KPIs, and the Case for Blameless Review
When something breaks, Bates wants a structured debrief that establishes facts rather than fault. Her worked example: an authorization goes in late. Was that a single miss, or were the last seven or eight also late and it went unnoticed because none were denied? If it is systemic, the next question is what happened upstream, which might be one analyst underperforming or might be families canceling assessment sessions. Those are entirely different problems, and an organization that only records the denial never learns which one it has.
She tells a story about a colleague responsible for a state Medicaid line who spent years navigating a claims process she assumed was simply how that payer worked. It had gone unsolved because she did not know it was raisable. A cross-functional group eventually worked out the best order in which to ask the payer’s questions, which did not fix everything and made the job materially better. The point is not the sequence. It is that the problem was invisible until someone from outside the function looked at it.
Her last structural argument concerns measurement. Organizations that let finance and clinical each define their own indicators end up with goals pointing in different directions, so she works backward from business outcomes to drivers to metrics. She wants clinical measures that reflect what the work is for, including graduation defined by where a client started and where they need to go rather than by discharge to no services, which fits the broader push toward standardized outcome measurement in ABA. Getting there requires leaders fluent in both columns, which is exactly what the two-manager structure stopped producing.
