I think the first thing to recognize is that running out of time on DI is usually a symptom, not the root problem. If you are regularly spending 3 to 4.5 minutes on questions that are classified as Medium in your review, the solution is not simply to become more disciplined about cutting yourself off. The bigger question is: why are those questions taking you 3 to 4.5 minutes in the first place?
A few causes show up repeatedly.
You may be trying to understand everything before answering anything. This is especially common on MSR. Students read every tab linearly, try to hold the entire situation in their heads, and only then attack the question. That is usually unnecessary. Your job is not to master the entire dataset. Your job is to understand enough of its structure to know where to retrieve information when the question asks for it. Before diving into the details, identify what each tab contains and how the tabs relate. Then let the question tell you where to go back.
You may be doing more calculation than the question requires. In Table Analysis and TPA, students often start calculating every possible value because they have not yet identified the constraint that actually determines the answer. Before calculating, ask: What exactly has to be true for this answer to work? What information would eliminate an option? Can I compare values rather than calculate them precisely? DI often rewards selective calculation.
Your underlying Quant, data interpretation, or reading/reasoning skills may not yet be automatic enough. DI combines arithmetic, algebra, percentages, ratios, statistics, logical reasoning, and reading comprehension. If solving a basic percent change or translating a relationship takes 45 seconds longer than it should, DI magnifies that weakness because you also have to navigate the data. A low DI score can therefore be telling you that some underlying skills need more work, not merely that your DI strategy is poor.
You may be rereading because you do not trust your first interpretation. This is another major time sink. Students read a condition, use it, then go back and reread it two or three times because they are afraid they missed something. That often comes from insufficient practice recognizing the structure of the question. With enough targeted practice, interpreting those conditions becomes much more automatic.
You may also be treating a question classified as Medium as one you are obligated to solve. But if a particular TPA hits one of your weaker areas, it may effectively be a difficult question for you. The goal is not to prove that you can solve every question later classified as Medium. The goal is to maximize your performance across the section.
For MSR specifically, I would stop trying to “read the passage efficiently” and instead learn to map the information. Know what Tab 1 is about, what Tab 2 contains, what Tab 3 contains, and where the numerical versus qualitative information lives. Then retrieve what you need.
For Table Analysis, focus on filtering and comparison before computation. Ask which rows actually matter. Look for extremes, thresholds, ordering, and relationships. Many TA questions become much easier once you stop treating the entire table as relevant.
For TPA, get very clear on the shared constraint connecting the two columns. Students often solve each side independently and create twice as much work. In many TPA questions, the two answers are connected by a central equation, condition, or logical relationship. Find that relationship first.
I would also review your slow questions with a different goal than simply understanding the solution. Ask yourself: Where did my extra two minutes go? Was it reading? Rereading? Setting up? Arithmetic? Not seeing the key constraint? Testing too many possibilities? Staying with an approach that was not working for too long?
If you categorize 20 or 30 slow DI questions this way, you will probably discover that your “timing problem” is really two or three specific skill problems. Fix those, and your pacing will usually improve naturally.
That is where I would focus before trying to impose a rigid timing system on top of the underlying issues.
Good luck!