The sad truth is, even though MBC is shown to improve therapy outcomes across modalities, and clinicians know it, some have been slow to adopt (Chung & Buchanan, 2018; Jensen-Doss et al., 2026). 

Nevertheless, those that are using MBC are seeing the benefits. NovoPsych helps thousands of clinicians each week engaged in MBC. 

The work the NovoPsych team has done is incredible. Here’s what I told our internal team in last week’s staff meeting:

• Each month, about 100,000 clients receive measurement-based care facilitated by NovoPsych (e.g., progress monitoring with tools like DASS-21).
• Because MBC helps more clients meaningful benefit from therapy, that’s a lot more people experiencing relief that they otherwise might not have had.  Using Cohen’s d as the primary metric, our internal “happiness adjusted” estimate is that the MBC impact is about 20,000 therapy-equivalent clients per month. That’s 240,000 people per year who have meaningful improvement because of NovoPsych’s tools. 

That’s the part that still hits me emotionally.

Because it means: even though I’m not in the therapy room anymore, the work is still deeply clinical—it’s just clinical at scale.

Improving therapy outcomes, 10% more effective each year:

I’m proud that NovoPsych has helped so many clinicians improve client outcomes. And what if each theoretical and technological advancement improved therapy outcomes by a further 10%, and clinicians were to adopt one of these innovations each year? After a decade, we’d double therapy’s effectiveness.

The next 10%: Better diagnosis helps client’s receive the right treatment at the right time

MBC is powerful and makes therapy about more effective —but it can still measure the wrong thing. 

If we miss autism, bipolar disorder, ADHD, PTSD, a personality disorder, or a trauma presentation, then our measures can become a beautifully plotted graph of a misunderstanding.

One statistic that keeps me up at night is bipolar disorder:

A large systematic review/meta-analysis found median delays of:

• 3.5 years from symptom onset to help-seeking, and
• 6.7 years from symptom onset to correct diagnosis.

That roughly 3.2 years from first help-seeking to correct diagnosis—years where people can be treated for the wrong thing, destabilised by the wrong medication, or simply bounce through services. 

Our internal ambition with a broad-spectrum screening questionnaire is simple: compress that multi-year diagnostic drift into something closer to minutes—a structured, evidence-based intake that flags key differential diagnoses early. That ambition is explicitly part of our roadmap. 

I think back to clients I worked with where I realised, years later, “I framed this wrong.” I missed the bipolar, autism, ADHD, or trauma. Not because I didn’t have enough knowledge about these presentations — but because the space of psychopathology is huge, and I was focussing on the wrong thing. 

Technology can help, and NovoPsych will turn comprehensive, accurate assessments that typically take years into something that takes minutes. We’ll do the science, and I hope that when we package this psychometric assessment in a way that actually makes clinicians’ jobs easier, it will make a big impact. Research and theory aren’t enough; we need this to impacts therapy reality. 

Our hope is that clients are given this adaptive assessment before the first therapy session, which means the clinician knows from session one key areas of focus. Making every first session one that hits the mark. Improving engagement in session one, which could reduce dropout rates, a factor that has outsized impact on poor outcomes. 

Another step further: Ambient psychometric assessments 

One emerging direction in psychometric assessment is to move beyond relying on self-report questions and move towards formal assessment that occurs ambiently. Wright et al. (2026) found AI could rate people’s Big Five personality simply by listening to them talk about random topics.

Instead of only asking people to rate statements on a scale from 1 to 5, this broader AI first approach allows psychometric assessment to happen in the background during a standard clinical session. That matters because client sessions reveal nuance, context, priorities, contradictions, and personally salient concerns that standard questionnaires can miss.

My view is that this kind of assessment could plausibly improve outcomes by 10% if it helps clinicians understand the person more accurately, assess more constructs, and more quickly. Better assessment can lead to better formulation, earlier identification of the key maintaining factors, more precise treatment planning, and more meaningful monitoring over time. Even modest improvements at each of those steps could compound across the course of therapy into a meaningful gain in outcomes.

Another 10%: Dynamic Biopsychsocial Formulations  

I think the future of assessment is about going beyond diagnosis toward more dynamic, biopsychosocial formulations. That means not just asking what diagnosis fits, but what problems are present, what predisposed them, what precipitated them, what perpetuates them, and what protective factors can support recovery. 

 Fried (2026) argues that mental disorders are better understood as overlapping biopsychosocial property clusters than as fixed natural categories, which helps explain heterogeneity, comorbidity, and blurry diagnostic boundaries. Instead of searching for one perfect classification system, he proposes building a richer “mental health atlas” that maps symptoms, traits, context, and dynamics over time.

At NovoPsych, we are increasingly interested in helping clinicians move in this direction: beyond narrow categories and towards richer, more evidence-based dimensional ways of understanding people. Measures like B-HiTOP are a step forward, but still just consider symptoms. The bigger opportunity is assessment that helps build a more complete map of the person in front of us, their context, social determinants of health, beliefs, and symptoms. Because people are richer than diagnoses, and our assessment models should be too.

The next 10%: reflective and deliberate practice, at scale (and AI’s role)

Mental health treatment remains deeply human. Clinicians can have access to all the tools described above, but therapist skill matters more than all of that. There is body of work suggesting that what separates highly effective therapists from the rest of us is not just experience, but how they improve — particularly through intentional training behaviours such as deliberate practice (Chow et al., 2015).

Deliberate practice is not the same as simply reflecting on your work. Reflection helps clinicians make sense of what happened: what they missed, what they did well, where they felt stuck, and what they might do differently next time. Deliberate practice goes a step further. It is a structured method for improving performance through focused work on specific skills, with feedback, repetition, and an appropriate level of challenge.

Reflection builds insight. Deliberate practice builds capability.

In my own development, both have mattered. Critically reflecting on my work helped me identify blind spots; feedback and intentional skill rehearsal helped me improve. 

What particularly interests me is what happens when reflective feedback becomes cheaper, faster, and more accessible.

Historically, getting better as a therapist has depended heavily on expensive supervision, formal training, and limited human feedback loops. AI changes the economics of feedback. Clinicians are already using clinical copilots like Just Ask NovoNote to get timely, specific feedback on their clinical work, and they’re loving it. 

The evidence is not yet mature enough to claim that AI-delivered feedback improves client outcomes directly. But the direction is clear: if clinicians can access high-quality feedback that is safe, ethical, specific, and under their control, it is reasonable to expect this will support better practice over time. Used well, AI can help surface blind spots and provide targeted corrective feedback for a few cents, rather than requiring a few hundred dollars of supervision every time.