GMAT multi-source reasoning strategy: timing and triage first.

The complete Data Insights guide covers all five DI question types, but Multi-Source Reasoning gets only a few paragraphs there because it deserves its own playbook. It is the one format on the GMAT Focus Edition where a single item can quietly consume a huge share of your section clock and leave you scrambling on the other questions. Most students lose points on MSR not because the reasoning is hard, but because they manage the clock badly — they read everything, then read it again, and run out of time. This post is about not doing that.

On Multi-Source Reasoning, the question is never “can I solve this?” first. It is “is this set worth the time it will cost me?” You decide that before you read a single tab in full.

What a Multi-Source Reasoning set actually is

Multi-Source Reasoning is one of five Data Insights question types, alongside Data Sufficiency, Table Analysis, Graphics Interpretation, and Two-Part Analysis. What makes MSR different from every one of them is the source structure. Instead of a single prompt, you get a panel of two or three tabs at the top of the screen, and you can only view one tab at a time. Click a tab and its contents fill the panel; click another and the first disappears.

The tabs mix formats deliberately. One tab might be a block of text — an email, a memo, a short report. Another might be a table of figures. A third might be a chart or a diagram. A classic set pairs, say, a price table, an order form, and a chain of emails between two managers. The information you need for any one question is usually scattered: the question depends on combining a number from the table with a condition stated in an email.

Each set comes with a small number of questions about that shared pool of information — typically three, sometimes two. The formats are mixed too. A common pattern is one standard five-option multiple-choice question plus two “either-or” questions, each presenting three statements you answer Yes/No, True/False, or Inferable/Not. That second format matters enormously for timing, because the three-statement items are scored all-or-nothing — you must get all three statements right to earn the point. There is no partial credit for two of three.

You will typically see one or two MSR sets in a Data Insights section. GMAC does not publish a guaranteed per-type count, so treat that as a typical range rather than a promise, but it shapes the math: one or two sets of two-to-three questions each accounts for a meaningful slice of the 20-question section.

Pre-read the tabs, do not read them

Here is the mistake almost everyone makes the first time: they open tab one, read it carefully end to end, open tab two, read it carefully, open tab three, read it carefully — and only then look at the first question. By the time they reach the questions they have spent three or four minutes absorbing detail they will never use, and they have forgotten half of it anyway.

The right move is to build a navigation map, not to read for comprehension. Spend roughly thirty to forty-five seconds doing a fast orientation pass:

• Read the tab labels. The tab names tell you what kind of source each one is — “Pricing,” “Emails,” “Q3 Report.” That alone tells you where a given fact probably lives.
• Skim each tab's shape, not its content. Note that tab two is a table with columns for region and cost; that tab three is three short emails in date order. You are cataloguing where information sits, not memorizing it.
• Note the cross-references. If an email says “per the attached schedule,” flag that the email and the table talk to each other. Those links are where the questions will live.

Then go to the questions. For each question, identify which tab (or which two tabs) it depends on, return to exactly that tab, and read only the part you need. You read per question, against your map — never the whole panel up front. Reading all tabs in full detail before looking at a single question is the named trap of this format. Your map tells you where to look; the question tells you what to look for.

Treat the tabs like a filing cabinet, not a book. You do not read a filing cabinet cover to cover. You learn which drawer holds what, then you open only the drawer the question sends you to.

When to mark-and-skip an entire set

This is the part of a multi-source reasoning strategy that actually moves your score, and it is the part nobody wants to do. A poorly managed MSR set can balloon past ten minutes if you let it — ten or more minutes for what is often just three questions, two of which are all-or-nothing and easy to miss anyway. On a 45-minute, 20-question section, that single set has eaten a fifth of your clock and may have bought you one point.

GMAT Focus gives you the tool to handle this. You can bookmark questions throughout the section, and on the review and edit screen at the end you can revisit your flagged questions and change up to three answers per section. In practice that means you can mark a question, move on, and come back if time allows. The standard heuristic from the prep literature is to give a question thirty to forty-five seconds; if you have no path to an answer, guess, bookmark it, and move.

Apply that ruthlessly to MSR. If your orientation pass tells you the set is dense — three tabs, lots of cross-referencing, an all-or-nothing item that hinges on reconciling three sources — and you are already behind on the section, the disciplined move is to put a defensible guess on each question, bookmark the set, and bank the time for cleaner questions elsewhere. A clean Data Sufficiency item might take you under a minute; trading an unmanaged MSR set for three of those is almost always the better expected value.

The trap is the sunk-cost spiral: you have already spent four minutes, so you feel you have to finish. You do not. The four minutes are gone either way. The only question is what the next two minutes are worth, and on an oversized set they are usually worth more spent elsewhere.

A worked example

Here is an invented two-tab set so you can see the navigation approach in action. Imagine the panel has two tabs.

Tab 1 — “Vendor Quotes” is a table. It lists three suppliers and their per-unit price for a component, plus a minimum order quantity:

• Supplier A: $4.00 per unit, minimum order 500 units
• Supplier B: $3.60 per unit, minimum order 1,000 units
• Supplier C: $3.20 per unit, minimum order 2,000 units

Tab 2 — “Emails” is a short exchange. A procurement manager writes: “We need exactly 1,200 units this quarter, and budget approval caps total component spend at $4,200. We cannot split the order across suppliers.” A reply adds: “Note that any supplier whose minimum order exceeds our 1,200-unit need is off the table — we will not pay for units we cannot use.”

Now a question: Which suppliers could fill this order within budget and within the stated constraints? This is where the map pays off. The question depends on both tabs — the prices and minimums from Tab 1, the quantity, budget, and no-split rule from Tab 2. Work it source by source:

• Apply the minimum-order rule first (Tab 2 × Tab 1). The reply email says any supplier whose minimum exceeds the 1,200-unit need is out. Supplier C's minimum is 2,000 — above 1,200 — so C is eliminated before any arithmetic.
• Cost-check the survivors against the $4,200 cap. At 1,200 units: Supplier A costs 1,200 × $4.00 = $4,800, which is over budget. Supplier B costs 1,200 × $3.60 = $4,320, also over the $4,200 cap.
• Conclusion. C is excluded by the minimum-order rule, and both A and B exceed the budget at the required quantity. No single supplier satisfies all constraints — which is exactly the kind of clean, defensible answer an MSR question is built to reward, and exactly the kind a panicked full-read reader talks themselves out of.

Notice what made this fast: you did not need to memorize either tab. You needed to know that the prices lived in Tab 1 and the rules lived in Tab 2, then pull the two specific facts each step required. That is the navigation map doing its job.

The recurring traps

Trap 1 — Full-reading every tab up front

The headline mistake. You burn three or four minutes absorbing detail before you know which detail matters, then forget most of it. Fix: orientation pass only — labels, shapes, cross-references — then read per question.

Trap 2 — Using only one tab when a question needs two

MSR questions are engineered to require combining sources. If a question feels answerable from one tab alone, you have probably missed a constraint sitting in another tab. Fix: before you answer, ask which other tab might hold a qualifier. The no-split rule in the worked example is exactly that kind of buried qualifier.

Trap 3 — Treating the three-statement items as three questions

The either-or items look like three small judgments, but they score as one all-or-nothing unit. Getting two of three right earns nothing. Fix: verify each of the three statements against the source with the same care; do not coast on the third because the first two felt easy.

Trap 4 — The sunk-cost spiral

You have already invested four minutes, so you stay to “finish.” The invested time is gone regardless. Fix: judge only the value of the next two minutes. If the set is oversized and you are behind, bookmark and move.

Trap 5 — Re-reading instead of re-locating

When a question stumps you, the instinct is to re-read a whole tab. Usually you do not need more reading — you need the one fact you skipped. Fix: go back to your map, identify the specific cell or sentence, and read only that.

A time budget within the DI section

Data Insights is 20 questions in 45 minutes — an average of about 2 minutes 15 seconds per question. But that average is a trap, because the five formats are wildly uneven in weight. A clean Data Sufficiency question can run well under a minute; an MSR set runs long. The way to survive is to budget the set, not the average. The published prep-source heuristics put a three-question set at roughly 5 to 7.5 minutes. Here is a workable plan inside the 45 minutes:

• Orientation pass: 30–45 seconds. Map the tabs — labels, shapes, cross-references — before you touch a question.
• Per-set target: about 5–6 minutes for a three-question set. That is the aggressive end of the published range and a defensible personal target; if you prefer the looser consensus, allow up to ~7.5 minutes, but know that every extra minute here is a minute stolen from the rest of the section.
• Per-question check inside the set: ~30–45 seconds to a path. If a question gives you no route in that window, put a defensible guess, bookmark it, and move to the next question in the set.
• Hard stop: if a single set crosses ~8 minutes, stop. Bookmark whatever is unanswered, place your best guesses, and bank the remaining time for the cleaner single-prompt questions.
• Reserve the review screen. You can revisit bookmarked questions and change up to three answers per section at the end. Leave yourself a couple of minutes so that feature is actually usable, not theoretical.

The 2:15 average is a section-wide number, not a per-question instruction. Spend it where it earns points. An MSR set that runs to six tight minutes is fine if a string of fast Data Sufficiency questions paid for it.

Why this matters for your score

Data Insights is scored on the same 60-to-90 scale as Quant and Verbal, and it counts equally toward your Total Score — one of three sections, each weighted the same. The Total Score runs 205 to 805 in 10-point steps, and at the top the curve compresses hard: a 645 sits at roughly the 87th percentile, the same competitive tier a 700 occupied on the old GMAT, which is why people call it “the new 700.” That is a percentile equivalence, not a score one — by the score-scale concordance a 645 Focus converts to about 680 old, which you can check on the score converter. The practical point: a couple of MSR sets mismanaged into a section-wide time crunch can pull DI down enough to cost you a percentile band on the total. Triage protects the whole score, not just the section.

The short version

A Multi-Source Reasoning set is two or three tabs you view one at a time, with two-to-three questions drawing on the shared pool, some of them all-or-nothing. Do not read the tabs — map them in a thirty-second orientation pass, then read per question against the map. Combine sources, because the questions are built to require it. Budget the set at five to six minutes, hard-stop near eight, and use bookmark-and-skip without guilt: an oversized set is worth far less than the three clean questions its minutes could buy. Triage first, solve second.

The platform

Zakarian GMAT's Data Insights chapters teach Multi-Source Reasoning as a navigation skill, not a reading-speed contest, with the tab-mapping and source-combining habits built into the worked examples and problem sets. The practice runner tracks per-question time, so you can see exactly where an MSR set is bleeding your clock and whether your triage discipline is holding under pressure. And the error log's six-tag taxonomy — Conceptual, Careless, Time Pressure, Misread, Strategy, Other — tells you whether a missed set was a reasoning gap or, as it usually is on MSR, a Time Pressure or Strategy failure you can drill away. I went from 565 to 735 on exactly this loop. The sample chapter is free if you want to see the teaching first.