Despite considerable investment, no biomarkers are regularly in use to help people with mental health and substance use problems. Yet, there may be opportunities to evaluate biomarkers in the clinic, and potentially change clinical practice and care. Abi-Dargham et al1 provide a comprehensive account of the current state of candidate biomarkers in psychiatric disorders. Their effort is wide-reaching and impressive, describing promising candidates and gaps in autism spectrum disorder, schizophrenia, anxiety, post-traumatic stress disorder, mood disorders, and substance use disorders. It is now time for our field to stop letting perfect be the enemy of the good. We are often divided (still sometimes even about whether some of the disorders we take care of have biological underpinnings at all) and highly self-critical. This is often healthy, but at times self-defeating. Prediction of outcomes, events, or treatment response does not need to be perfect. It needs to be better than chance, to provide more benefit than harm, and, where possible, to be cost-effective. Cost-effective where possible, because sometimes certain societal or other benefits, even if more expensive, may be considered worthwhile depending on value systems. Here I propose a path forward for the most promising biomarkers identified by Abi-Dargham et al, using the example of the striatal connectivity index (SCI) in schizophrenia. The authors identify the SCI as a particularly promising biomarker for evaluating treatment response (and resistance) to antipsychotic medication in first episode psychosis2, 3. Concerning Step 1 of biomarker development (the targeting of the clinical question), response to antipsychotics is of undebatable importance, because non-response often leads to hospitalization, and sometimes death. As to Step 2 (internal validation), it is clear that this biomarker is targeting the underlying process, due to rigorous study design that utilized algorithmic treatment, participants with limited or no antipsychotic exposure, and testing using different antipsychotics. Step 3 (external validation) was successful, using an even higher bar, testing in a chronic schizophrenia sample. In addition, the original papers examining the SCI showed that it was associated with days of inpatient hospitalization, creating an even stronger argument for its potential, via association with a “high stakes outcome”. There are implications with this biomarker for getting first-episode patients to clozapine faster, which can be life-saving for some, given that those who do not respond to conventional antipsychotics could be identified at their first episode of psychosis. So, what is left to do? Evaluate clinical utility, i.e. Step 4. The time is now to conduct randomized controlled trials that directly assess the utility of promising biomarkers in ordinary clinical practice. Sticking with the SCI example, this could mean conducting a randomized controlled trial of early psychosis patients stratified via their SCI. This trial could explore whether early use of clozapine in those identified by the SCI as likely non-responders to conventional antipsychotics actually has a positive impact in terms of clinical improvement, e.g. by reducing hospitalization days, thus helping understand whether the SCI is clinically useful. That is, does the use of the SCI in clinical practice lead to better outcomes for this group of patients than usual care? While this could be interpreted as a question simply of earlier clozapine utilization, it is not about earlier utilization for all. Rather, the biomarker would guide earlier utilization by clinicians only for that group of patients with a SCI value that indexes likely non-response to conventional antipsychotics. A trial such as this one, if successful, would have the potential to change prescribing and regulatory guidelines specifically for patients assessed by the SCI as likely not to respond to conventional antipsychotics. This could mean that a biomarker in psychiatry would have real-world impact, when currently there is no such case. An opportunity could exist in the same trial to incorporate melanocortin 4 receptor genotype, which confers a nearly five-fold increased risk of weight gain in relation to antipsychotic exposure4. Similarly, while agranulocytosis is rare, an allele in the HLA-DQB1 gene carries a ~15 fold increased risk of this potentially lethal event5. Therefore, one could stratify patients on multiple biomarkers, maximizing potential gains and minimizing potential harms, in the same clinical trial. Cost-effectiveness analyses could further strengthen the case. Saving even one day in hospital would likely offset the costs of the magnetic resonance imaging (MRI) and genetic tests. If we are serious about getting biomarkers into clinical practice, the biomarker/biological field and the psychiatric services field should work together for successful implementation. Engaging patients and family members with lived experience would be important. A parent who has witnessed his/her teenager or young adult child recovering from early psychosis thanks to the use of a given biomarker would be a powerful advocate, providing a lived experience voice that could help support the scale and spread (i.e., the implementation) of that biomarker into clinical practice. In addition, engaging policy makers who may have a say in health system incentives early in the process, as well as the relevant regulatory agencies, would be wise. Practice change is notoriously difficult. Even the implementation of measurement-based care in mental health clinics, e.g. using a scale routinely to guide treatment decisions, may be a challenge. If we have the data to bring MRI results or a genetic test into the clinic, the challenge of implementation may be even greater. Fortunately, in relation to the SCI biomarker example in early psychosis, the presence of networks of clinics that are part of a learning health system – e.g. via EPI-NET in the US, as well as similar initiatives in Canada, Australia and elsewhere – could be as good of an environment as we might hope for to propel successful translational efforts into clinical practice. In that sense, the time is now as well, and the broader notion of precision medicine, implementation science, and a learning health care system has been described6, and could be applied in psychiatry. Much of the focus of the comprehensive review by Abi-Dhargam et al is on adult psychiatry. However, the peak age of onset of mental illness is 14.5 years of age7. When describing or planning for biomarker evaluation in anxiety or depression, where many cases have their onset during adolescence, one could argue that most studies should be conducted at that timepoint in the lifespan. The dynamic evolution of mental illness at that timepoint also offers primary and secondary prevention opportunities. For instance, 75% of all index psychotic episodes have already presented for mental health care earlier in life for other mental illness8. Therefore, capitalizing on our knowledge of the windows of brain development that are paired with windows of onset of mental illness (and substance use) could assist in more optimal study design related to timing. Furthermore, funders might consider investing in “master observational trials”, which have been dubbed “a new class of master protocol to advance precision medicine”9, currently emerging in oncology. The master observational trial is a prospective, observational trial that broadly accepts patients and collects comprehensive data on each. All of the information is tied together in a prospective observational registry using standardized reporting and metrics. The goal of these trials, which would be of tremendous benefit to psychiatry, is to harness the power of real world data to advance biomarker discovery and test the clinical utility of precision-based and personalized medicine.