Still Below Your Benchmark? Focus Your Prep Where It Matters Most

Maya Brooks

Hi everyone, and welcome back to the AI Med Tutor podcast. I’m Maya Brooks, your AI-generated fourth-year medical student co-host, and as always, I’m joined by Dr. Randy Clinch. Today, we’re talking to those of you who have been preparing for COMLEX Level 2 or USMLE Step 2, but you’re still working toward whatever benchmark your medical school requires before you’re eligible to take your examination. You’ve been studying for weeks. Maybe longer. You’ve worked through hundreds—or thousands—of questions. You’ve reviewed weak areas. You’ve taken a COMSAE, an NBME self-assessment, or another practice examination. And then you get another score back. You’re still below the benchmark your school requires before you can sit for COMLEX Level 2 or USMLE Step 2. At this point, the pressure can become intense. You may be watching classmates schedule their exams or move forward with rotations while you’re still trying to demonstrate readiness. So what should you do now? That’s what we’re talking about today. Dr. Clinch, where should a student in this position begin?

Dr. Randy Clinch

Hi everyone. I’m Dr. Randy Clinch, a DO family medicine physician and medical educator. And if you’re in this situation right now, I want to start with something important. A score below your school’s benchmark deserves your attention. But it does not automatically mean that you need to start your entire board-preparation process over again. In fact, if you’ve already been preparing diligently, doing more of everything may be the wrong response. At this stage, I want you to become more diagnostic. Your question is no longer simply, “What should I study?” Your question should be, “What is still preventing my performance from reaching the level I need?” Those are very different questions.

Maya Brooks

So this sounds a lot like one of the ideas we’ve returned to throughout AI Med Tutor: diagnose before you prescribe.

Dr. Randy Clinch

That’s the idea. If you’ve taken several practice assessments, you now have something you didn’t have when you began preparing. You have data. Maybe you have several COMSAE reports. Maybe you have NBME self-assessment reports. You probably have question-bank analytics. You may also have a Miss Log or at least a sense of the kinds of questions that continue to give you trouble. At this point, I don’t want you treating each new assessment as an isolated event. I want you to put the information together. And that brings us to a simple four-step process I’m going to call the Benchmark Recovery Loop. The four steps are: find the signal, diagnose the bottleneck, prescribe the repair, and prove the repair.

Maya Brooks

Let’s start with finding the signal, because I can imagine a student looking at several assessment reports and seeing weaknesses everywhere.

Dr. Randy Clinch

And that’s one of the traps. If you look at every below-average category and tell yourself, “I need to fix all of this,” you can quickly end up rebuilding your entire study plan. Instead, compare your last two or three assessments. Then compare those results with what you’re seeing in your question bank. Ask yourself: What keeps showing up? Maybe cardiovascular performance has repeatedly been weak. Maybe you keep struggling with gastrointestinal questions. Maybe several reports point toward problems with diagnosis or management. But don’t stop at the category name. That broad category is only the beginning. If cardiovascular performance is low, for example, what is actually happening? Are you missing heart-failure questions because you don’t understand the physiology? Are you confusing heart failure with COPD when patients present with dyspnea? Are you identifying the diagnosis correctly but missing the next best step? Those are three very different problems. Your assessment report gives you the map. Your question-bank performance helps provide the microscope.

Maya Brooks

And that leads naturally into step two: diagnose the bottleneck.

Dr. Randy Clinch

Right. Go back to the three categories we’ve used throughout AI Med Tutor: content, reasoning, and mechanics. A content problem means there’s something important that you genuinely don’t know or don’t understand well enough. Maybe you don’t understand the mechanism. Maybe you’ve forgotten an important association. Maybe you cannot retrieve a drug effect or physiologic relationship when you need it. A reasoning problem is different. You may know the underlying facts, but you’re not assembling them into the correct clinical story. You recognize the individual findings, but you don’t identify the hinge clue that separates two competing diagnoses. And then there are mechanics problems. Maybe you know the medicine and understand the pattern, but you’re answering the wrong task. You’re choosing the diagnosis when the question is asking for the next step. You’re overlooking a timing or urgency qualifier. You’re spending too long on difficult questions and rushing through easier questions at the end of the block. Or you’re allowing the answer choices to pull you away from a reasonable interpretation of the stem. Those problems will not necessarily improve because you watched another six hours of videos.

Maya Brooks

And I think this is where the pressure students are feeling could actually make things worse. They see another score below the benchmark and think, “I need to study harder.”

Dr. Randy Clinch

I would change one thing in that sentence. At this stage, you probably don’t need to study harder. You need to study more precisely. If you’re below your school’s benchmark right now, the answer usually isn’t to make your study plan bigger. Your job is to make the problem smaller. Find the patterns that are still costing you points. Identify why they’re happening. Repair them deliberately. Then prove that the repair is working. And that takes us to step three: prescribe the repair.

Maya Brooks

Let’s make that concrete. Suppose I decide my primary bottleneck is content.

Dr. Randy Clinch

Then use precision content review. Don’t tell yourself, “I need to review cardiology.” That is far too broad. Instead, perhaps you realize that you continue to miss questions involving systolic versus diastolic heart failure because you cannot reliably connect the physiology to the clinical findings. That is a specific problem. Spend a short, defined period repairing that concept. Maybe that’s twenty or thirty minutes. Then immediately retrieve it. Do five or ten targeted questions. Explain the mechanism without looking at your notes. If appropriate, build one Pattern Card. Then see whether you can use that information.

Maya Brooks

What if it’s reasoning?

Dr. Randy Clinch

That’s where I would lean heavily on discrimination. Suppose you keep missing dyspnea questions, but when you review the explanations, you realize you actually knew most of the individual facts. Your problem may be distinguishing the competing stories. So create a Pattern Card around the presentation. Maybe the title is, “Heart failure versus COPD in a patient presenting with dyspnea and wheezing.” Identify the key clues. Identify the mechanism. Then ask yourself what the hinge findings are that move you toward one diagnosis and away from the other. After that, do a small interleaved set. Mix heart failure, COPD, pneumonia, pulmonary embolism, and perhaps asthma. Now you’re practicing the decision you were actually getting wrong.

Maya Brooks

And if it’s mechanics?

Dr. Randy Clinch

Choose one behavior. Not five. If you’re repeatedly answering a different question from the one being asked, use the task-sentence-first approach. Read the final sentence before you begin working through the stem. Ask: What is my job here? Diagnosis? Mechanism? Next step? Treatment? Complication? Then read the stem with that task in mind. If your problem is rumination, your repair might be a mark-and-move rule. If you’re missing urgency, deliberately perform your stability screen while reading the stem. Pick one mechanics correction and apply it across enough questions that it becomes part of your workflow.

Maya Brooks

So far we’ve found the signal, diagnosed the bottleneck, and prescribed a repair. What’s step four?

Dr. Randy Clinch

Prove the repair. And this is where I want students in this situation to be particularly thoughtful. Another COMSAE or NBME assessment should not be the first time you find out whether your latest study strategy worked. Those assessments are valuable resources, and they can also carry a lot of emotional weight when you’re waiting to reach an institutional benchmark. So before you take another major assessment, look for smaller evidence that your repair is working. If you repaired a content problem, can you retrieve the information without your notes? Can you answer targeted questions correctly? If you repaired a reasoning problem, can you distinguish that pattern from its closest competitors? If you repaired a mechanics problem, does the correction hold across an entire block? Then take it one step further. Can the repair survive a mixed timed set? That matters because your real examination is not going to announce, “Here come the eight questions on the topic you repaired yesterday.” You have to recognize the pattern when it appears unexpectedly. As a reminder, since your question bank can quickly run out of very specific questions you want to test, search our prior episode titles for the one that includes the free resource, OpenEvidence, so you can use that resource as a supplement to your question bank and generate a variety of very specific questions for yourself.

Maya Brooks

So there’s a difference between feeling better about the topic and demonstrating better performance.

Dr. Randy Clinch

A very important difference. Reading an explanation and thinking, “Now I understand it,” is useful. But it is not proof. Proof comes when you retrieve it again. Then when you recognize it in a slightly different presentation. Then when you distinguish it from something similar. And eventually when you can do that during a timed mixed block. That is the progression we want. Targeted repair followed by transfer.

Maya Brooks

I want to talk about something else these students may be experiencing. Some of them have been studying for a long time. They may have expected to test weeks ago. They may be seeing classmates move forward while they’re still preparing. And they’re tired.

Dr. Randy Clinch

That matters. One of the worst responses to a disappointing practice score is automatically deciding that tomorrow you’re going to study twelve hours. If you’ve been preparing for an extended period, fatigue can become part of the performance problem. Your concentration declines. Your review becomes superficial. You start reading explanations instead of actively working with them. You become more likely to ruminate on difficult questions. And eventually you can spend more hours studying while getting less useful learning from those hours. This is where I want you to return to the idea of the Daily Engine. You need retrieval. You need targeted repair. And you need reinforcement. But the size of those components can change. On a strong day, perhaps you can complete a full timed block, review it effectively, complete a targeted repair, and revisit some older learning. On a more fatigued day, you may need a smaller question set, one important repair, and a brief retest. Preserve the learning loop. Reduce the load when necessary.

Maya Brooks

That also sounds like a reason not to suddenly add three new resources.

Dr. Randy Clinch

Yes. If you’ve already completed substantial board preparation, another resource is rarely valuable simply because it is new. Before adding something, ask: What specific problem will this resource solve that my current tools cannot? If you cannot answer that question clearly, adding it may simply increase cognitive load. Your question bank already contains performance information. Your previous assessments contain information. Your Miss Log, if you’re using one, contains information. Your job now is to extract the signal from the information you already have.

Maya Brooks

Could you give students a very simple way to remember the Benchmark Recovery Loop when they finish this episode?

Dr. Randy Clinch

Sure. Four questions. First: What keeps showing up? That finds the signal. Second: Why am I getting it wrong? That diagnoses the bottleneck. Third: What is the smallest repair that addresses that problem? That gives you the prescription. And fourth: What evidence will show me that I actually fixed it? That gives you the proof. So: What keeps showing up? Why is it happening? What is the smallest repair? How will I prove it?

Maya Brooks

And presumably the next formal assessment comes after you’ve had a chance to work through that cycle.

Dr. Randy Clinch

Yes, while following whatever assessment schedule and eligibility process your medical school has established. Every school handles board eligibility somewhat differently. Your institution may require a particular score, a particular assessment, multiple assessments, or other criteria before authorizing you to sit for COMLEX Level 2 or USMLE Step 2. Follow those requirements and work with the faculty or advisors responsible for your school’s process. What I want to help you control is what happens between those assessments. Don’t simply wait for the next score. Use that time to generate evidence of improvement.

Maya Brooks

Before we finish, Dr. Clinch, what would you say to the student who has worked hard, has taken multiple assessments, and is starting to wonder whether they’re ever going to reach that benchmark?

Dr. Randy Clinch

I would tell that student to resist the temptation to turn one number—or even several numbers—into a judgment about their future. But I also wouldn’t tell them to ignore those numbers. Use them. If your performance isn’t where it needs to be yet, there is still something to learn from that. Your task is to make the problem increasingly specific. Maybe there are two clinical presentations that continue to hurt you. Maybe your knowledge is improving but your next-step reasoning is lagging behind. Maybe your biggest remaining issue is pacing. Whatever it is, naming the problem gives you something you can work on. You don’t need to solve all of medicine before your next assessment. You need to identify the problems that are still producing lost points and systematically reduce them. Find the signal. Diagnose the bottleneck. Prescribe the repair. Prove the repair. Then reassess. That is a much more useful strategy than simply telling yourself to study more.

Maya Brooks

And I think that’s a good place to end. If you’re working toward your school’s benchmark right now, remember that your next practice score isn’t the only evidence that you’re making progress. Look for evidence in the work you’re doing every day. Are recurring misses becoming less frequent? Are you recognizing patterns faster? Can you explain why the correct answer is correct? Can you distinguish it from the closest alternative? Are you making better decisions under timed conditions? Those are signs that the underlying system is changing. That’s it for today’s episode of the AI Med Tutor podcast, everyone. Thanks so much for listening. Keep using your performance as feedback, make your repairs specific, and give yourself the opportunity to prove that those repairs are working. And remember, stay curious, and keep learning.