Tabular DI Fundamentals and Speed Reading

In SBI PO Prelims and Mains, five Data Interpretation questions can decide whether you clear the section cutoff or miss it by a single mark. The candidates who get all five right are rarely the fastest calculators — they are the fastest readers. They scan the table in 20 seconds, lock in the row-and-column logic, then attack the questions. Let us learn the discipline they use.

Definition: Tabular Data Interpretation (DI) is a set of questions based on data arranged in rows and columns, sometimes with totals, sub-totals, percentages, or multiple categories like years, products and cities.

Definition: A row typically lists one entity (a year, a city, a product line) and a column lists one attribute (revenue, units sold, share). The intersection is a single data cell.

The first 20 seconds — read the table, not the question

Counter-intuitively, the very first action when a tabular DI set appears is not to read all the data, and not to read the question yet. The first action is to read the table header and column titles for about 20 seconds with three specific goals:

1. What does each row represent? Years? Cities? Products? Branches?
2. What does each column represent? Absolute numbers, percentages, ratios? In thousands, lakhs or crores?
3. Are totals or averages given, or must they be derived?

Once you know these three things, you have a mental map of the table. Now, and only now, read the first question. Always return to the table for each new question — never re-read the whole table top to bottom.

Why it matters

SBI PO 2023 Mains had a DI set where Column 3 was labelled "% growth from previous year" while Column 2 was in absolute crores. Candidates who missed the unit confusion treated 12% as 12 crores and lost all five questions. The cost of misreading a header is the entire set — five marks plus four wasted minutes. Reading slowly for 20 seconds saves four minutes downstream.

Row vs column logic — which way do you read?

A row-wise question asks you to compare different attributes of the same entity. A column-wise question asks you to compare the same attribute across entities.

Examples:

• "Total revenue of City A across all four products" → row-wise read along City A.
• "Total revenue from Product 2 across all five cities" → column-wise read along Product 2.
• "Which product had the highest sales in 2023?" → column-wise along year 2023.
• "What is the average sale of Product 1 over five years?" → column-wise along Product 1.

A test mark on each question — "row?" or "col?" — saves seconds of re-orientation, especially when the question is phrased in convoluted English.

The absolute vs percentage trap

The single most common SBI PO trap is mixing absolute values with percentages. The exam shows two columns side by side: Column A = "Revenue (crores)", Column B = "% share of branch in total". A careless candidate might add Column B values across branches expecting them to give total revenue. Wrong — Column B sums to 100, not to the total revenue.

Before any computation, ask:

• Is this number an absolute quantity (rupees, kg, units)?
• Is it a percentage (always relative to a base — what is the base?)
• Is it a ratio (compared to which other quantity?)

Reading the question first — for the second pass

Once you have the table map, read the first question carefully and only then return to the relevant cells. Underline three classes of keywords:

• Quantifiers: "at least", "at most", "exactly", "approximately"
• Multiplicity tags: "respectively", "in the same order"
• Base words: "more than", "less than", "from", "compared to" — these decide which number is the denominator

Without underlining, the brain glides over "respectively" and you swap two answers. With underlining, you mechanically check that your final answer matches the order asked.

Percentage problems — picking the right base

The most-tested trap: picking the wrong base for percentage comparison.

• "A is what percent more than B" → base is B. Use (A - B) / B × 100.
• "A is what percent of B" → base is B. Use A / B × 100.
• "Increase from previous year" → base is previous year, not current.
• "Decrease over the base year" → base is first year of the series.

The wrong choice of base typically changes the answer by 10 to 30 percentage points, and the wrong option will be a distractor in the choices.

Worked example

Question: A table shows revenue of four companies in 2022 and 2023 in crores. Company P: 80 → 100. The question asks, "By what percent did the revenue of Company P increase from 2022 to 2023?"

Solution:
Step 1: Identify the base year — the year before the change is 2022, so the base is 80.
Step 2: Compute the change: 100 - 80 = 20.
Step 3: Compute the percentage: 20 / 80 × 100 = 25%.
Step 4: Double-check the direction: revenue went up from 80 to 100, so the percentage should be positive, which it is.
Conclusion: Revenue of Company P increased by 25%.

A common wrong move is to divide by 100 (the new year), which gives 20%. This is the kind of error SBI PO loves to plant as a distractor.

The five-linked-questions strategy

Tabular DI in SBI PO is almost always a five-question set built on one table. The trick: most questions reuse the same row sums or column sums. After computing the total of each row and the total of each column the first time you need them, jot them in the margin of the paper. The next four questions usually re-use those totals, and you do not recompute. This cuts the total time per set from 6 minutes to 3 minutes.

For ratio comparisons, when two ratios look similar (say 13/7 vs 19/11), cross-multiply mentally — 13 × 11 = 143 and 19 × 7 = 133 — and the bigger product gives the bigger ratio. This is faster than dividing.

Real-world example

A bank manager looking at a quarterly branch-performance report uses exactly the same logic — scanning headers for units (lakhs vs crores), reading row totals for branch performance, column totals for product-line performance, and flagging the row with the largest percentage growth. The DI section is essentially a bank manager's reading test, compressed to 60 seconds per question.

Common misconception

The most common misconception is that mental calculation speed is the bottleneck, so candidates spend coaching-class months drilling multiplication tables. In reality, the bottleneck is reading: misreading "thousand" as "lakh", confusing "more than" with "of", or applying a row percentage to a column total. Speed maths helps, but slow careful reading + medium-speed arithmetic beats fast careless reading + fast arithmetic every time.

Choosing your DI sets

In SBI PO Mains, four DI sets are usually offered and only the best three are attempted. The smart strategy: in the first minute, glance at all four sets and pick three based on which table layouts look the cleanest. A clean table — one with simple headers, no nested categories, fewer than five columns — is almost always faster than a complicated multi-tier one with footnotes.

In SBI PO, Data Interpretation is rarely about hard arithmetic — it is about doing easy arithmetic at speed, under pressure, with a forest of numbers in front of you. The aspirants who clear the sectional cut-off are not the ones who calculate the most accurately; they are the ones who recognise the calculation at a glance and convert it into a mental shortcut. This lesson builds that recognition layer.

Definition: A percentage is simply a fraction whose denominator is 100. Saying "37.5%" is exactly the same as saying "3 out of 8", which is exactly the same as saying "0.375".

Definition: A ratio a : b is a comparison of two quantities, mathematically equal to the fraction a/b. Comparing two ratios is therefore comparing two fractions.

Why memorising fraction-to-percent conversions matters

A typical SBI PO tabular DI set asks five questions like "By what percent did Branch B's sales increase from 2022 to 2023?" or "What is the ratio of male to female employees across all four cities?" Almost every such question hides one of these structures:

• a fraction of the form 1/n where n is small
• a percentage that maps back to a small fraction
• a comparison of two fractions

If you have already locked the small-fraction table into long-term memory, you skip the entire division step. The brain reads "12.5%" and instantly substitutes "1/8". You then multiply or divide using that 1/8, which is far faster and far less error-prone than punching 12.5 into mental arithmetic.

The fractions worth memorising cold:

• 1/2 = 50%
• 1/3 ≈ 33.33%
• 1/4 = 25%
• 1/5 = 20%
• 1/6 ≈ 16.67%
• 1/7 ≈ 14.28%
• 1/8 = 12.5%
• 1/9 ≈ 11.11%
• 1/10 = 10%
• 1/11 ≈ 9.09%
• 1/12 ≈ 8.33%

And the eighths and other useful multiples:

• 3/8 = 37.5%, 5/8 = 62.5%, 7/8 = 87.5%
• 2/3 ≈ 66.67%, 4/3 ≈ 133.33%
• 5/6 ≈ 83.33%
• 1/16 = 6.25%, 3/16 = 18.75%

Drill these until the response is automatic. A two-second recognition saves you twenty seconds across a DI set, which is the difference between attempting the last set and leaving it blank.

The core percentage formulae you will reuse

Every DI question is one of four patterns. Lock the formulae.

Pattern 1 — Percentage of a number: X% of Y = (X/100) × Y. Convert X% to its fraction whenever possible. 25% of 480 is not 0.25 × 480; it is 480/4 = 120, in one step.

Pattern 2 — What percent is X of Y: Answer is (X/Y) × 100. The trick: cancel before you multiply. 84 out of 240 is 84/240 = 7/20 = 35%.

Pattern 3 — Percentage change: ((New − Old) / Old) × 100. Always divide by the OLD value, never the new one. This is the single most common sign error in DI.

Pattern 4 — Reverse calculation (unitary method): If 12% of a quantity equals 96, then 1% = 8, so 100% = 800. This is faster than dividing 96/0.12 — you simply divide by 12 and multiply by 100.

Comparing ratios by cross-multiplication

You will often be asked which of two cities has a higher male-to-female ratio, or which year had a higher profit-to-revenue fraction. Do not convert each ratio to a decimal — that is two divisions, and decimals invite arithmetic slips.

Use cross-multiplication. To compare a/b with c/d (with b, d > 0):

• Compute a × d and b × c
• The fraction whose numerator-side product is larger is the larger fraction

Worked illustration: compare 7/12 with 9/16. Cross products: 7 × 16 = 112 and 12 × 9 = 108. Since 112 > 108, we conclude 7/12 > 9/16. No decimals, no division, just two small multiplications.

Why it works: multiplying both sides of a/b vs c/d by the positive number bd preserves the inequality, giving ad vs bc.

Approximation: your secret weapon in SBI PO

In SBI PO Mains, the answer options for DI are spaced widely enough that rounding intermediate values to two significant figures is almost always safe. If a question asks for 23.7% of 4,872, you can confidently compute 24% of 4,900 ≈ 1,176 and pick the closest option. The full computation 0.237 × 4,872 = 1,154.66 sits inside the same option range.

A safe approximation discipline:

• Round percentages to the nearest 0.5% or whole number
• Round large numbers to two or three significant figures
• Track the direction of your rounding so you know whether your answer is slightly high or slightly low
• Reserve exact calculation for questions where two options are within 1% of each other

Why it matters: The exam is timed at roughly 36 seconds per question. Exact calculation simply does not fit. The students who clear sectional cut-offs treat approximation as the default and exact arithmetic as the exception.

Real-world example

Real-world example: A typical SBI PO table shows that Branch A's loan disbursal was 4,250 crores in 2022 and 5,100 crores in 2023. The question asks for the percentage increase. Mentally: change = 850, original = 4,250. The fraction 850/4,250 simplifies to 850/4,250 = 1/5 = 20%. Done in under five seconds because you recognised that 850 × 5 = 4,250.

Common misconception

Common misconception: Many aspirants compute percentage change as (New − Old) / New, especially when "new" appears earlier in the sentence. This is wrong. The denominator is always the BASE, which is the original value you are comparing against. If revenue moves from 100 to 120, that is a 20% rise (20/100), not a 16.67% rise (20/120). The wrong denominator turns a positive trap into a negative score.

Question: A village has 3,500 voters. Of these, 14.28% supported Party X. How many voters is that?

Solution:
Step 1: Recognise 14.28% as the fraction 1/7.
Step 2: Compute 3,500 × (1/7) = 500.
Conclusion: 500 voters supported Party X. No division needed — pattern recognition turned a 14.28% multiplication into a single division by 7.

In SBI PO Prelims and Mains, every DI set is a set — usually 5 to 6 questions stacked on the same table or chart. Aspirants who treat each question independently end up reading the same numbers six times. Aspirants who pre-process the table once and treat the six questions as variations of one calculation finish the entire set in half the time. This lesson shows that "pre-process once, reuse everywhere" technique on a multi-column table of employees and gender percentages.

Definition: Tabular Data Interpretation (Tabular DI) is a category of DI questions where data is presented in a rectangular table — rows are entities (departments, products, branches) and columns are attributes (totals, percentages, ratios). Questions test percentage, ratio, average, and difference calculations across rows or columns.

Definition: Pre-processing in DI means computing all the obvious derived quantities — absolute counts, totals, gender splits, averages — once at the start of the set, even before reading the actual questions. The numbers are then plugged into each question without recomputing.

Anatomy of the set

The source table gives employees in 5 departments and the percentage that are female in each. A representative row says — Department A: 250 total employees, 40% female. That single row has three useful numbers buried inside it:

• Total = 250
• Female = 40% of 250 = 100
• Male = 250 − 100 = 150

If a question later asks "male in Dept A as a fraction of total male across all departments," you already have the 150. If it asks "ratio of male in A to female in B," you again already have the 150. The 12-minute set becomes a 6-minute set.

Doing the worked example end to end

Following the source: Dept A — 250 total, 40% female.

Step-by-step margin notes (write these down on the OMR rough sheet):

• Dept A: Total 250, Female 100, Male 150.
• Continue for B, C, D, E — say total female across all five departments is given as 520.

If the question now is "Female in A as percentage of total female across all departments":

Female in A / Total female × 100 = 100 / 520 × 100 = 19.23%.

That single calculation took 5 seconds because the 100 was already pre-computed. If you had to first read "40% of 250" all over again, set up the multiplication, and only then divide by 520, you would have spent 30 seconds for the same answer.

The structural insight — questions are linked

Across SBI PO papers, the six questions of a DI set rarely test six unrelated calculations. Typically:

• Q1 — pure percentage (e.g., Females in B as % of total in B).
• Q2 — ratio (e.g., Males in A : Females in C).
• Q3 — difference (e.g., Difference between males in A and females in D).
• Q4 — average (e.g., Average number of females across departments).
• Q5 — percentage between rows (e.g., Females in A as % of total female).
• Q6 — combined / "what if" question (e.g., If 20% of females in B resign, new ratio of males to females in B).

Every one of those questions reuses the same per-department male and female counts. So the rational strategy is — build a tiny grid of (Department, Total, Female, Male) once. Three columns, five rows — fifteen numbers. After that, every question is a one-line plug-in.

Common time leaks

Three habits drain the most time in tabular DI:

• Recomputing the absolute female count for the same department in three different questions.
• Reading the question stem from scratch instead of scanning for the "what is being asked" phrase first.
• Doing multiplications like 40% of 250 the long way (250 × 40 / 100) instead of using "10% of 250 = 25, so 40% = 100" — a 3-second mental hop.

Why it matters: SBI PO's DI section gives 30 minutes for 30 questions, typically organised as five sets of six. If you can shave 5 to 6 minutes by pre-processing each set, you bank time for Quant and Reasoning where every extra minute is worth half a mark. Pre-processing is also the foundation for caselet DI, pie-bar combinations, and missing data DI — all SBI Mains staples.

Real-world example: A branch manager at SBI reviewing monthly KPIs across five product lines does exactly this — total customers, conversion percentage, ticket size — and writes the absolute conversions next to each row before even looking at the questions head office sent. Without this pre-processing, comparing product lines on the fly takes hours; with it, the same review takes minutes. The DI section is testing this real banking skill.

Common misconception: Many students assume "pre-processing wastes time at the start of a set." In reality, computing five percentages takes about 60 to 90 seconds and saves 5 to 6 minutes downstream because every question reuses those exact five numbers. The break-even point is around question 2 — after that, every saved question is pure profit.

Question: Dept A has 250 total employees, 40% of whom are female. If the total number of females across all 5 departments is 520, what is the female count in A as a percentage of the total female count?
Solution:
Step 1: Compute the absolute female count in Dept A. Female in A = 40% of 250 = 0.40 × 250 = 100. (Mental shortcut: 10% of 250 is 25, so 40% is 100.)
Step 2: Compute males in A for any later question: Male in A = 250 − 100 = 150.
Step 3: Compute the required percentage: (Female in A / Total female) × 100 = (100 / 520) × 100.
Step 4: 100/520 = 5/26 ≈ 0.1923, so the percentage = 19.23%.
Conclusion: Females in Dept A form approximately 19.23% of the total females across all departments. Critically, the 100 and 150 are now logged on your rough sheet for the rest of the set.

Cover the answer, recall, then check. 12 cards on reading and solving tabular DI at SBI PO speed.

Q1. In tabular DI, what three things do you read BEFORE looking at the questions?
A1. (1) The title and units of each column (actual numbers vs %, lakh vs crore), (2) whether any row/column is a Total, (3) any stated relation such as "Sold = Produced − Unsold". Reading these first prevents unit traps.

Q2. A column reads "Employees (in '00)" and a cell shows 12. How many employees?
A2. 12 × 100 = 1,200. The "in '00" multiplier is a classic SBI unit trap — always apply it.

Q3. Fastest way to find what percent 90 is of 450?
A3. 90/450 = 1/5 = 20%. Simplify to a known fraction before converting.

Q4. The percentage-change formula used in almost every DI set?
A4. (New − Old)/Old × 100. From 250 to 300: 50/250 = 20%. The denominator is always the OLD/base value.

Q5. Total = 8,000 and a share is 17.5%. Find the value fast.
A5. 17.5% = 7/40, so 8,000 × 7/40 = 1,400.

Q6. When must you use a weighted average, not a simple average, in a table?
A6. Whenever the groups differ in size — the combined average of two branches' averages must weight by employee count, not just (avg1 + avg2)/2.

Q7. Simplify the ratio of two cells, 336 : 224.
A7. Divide both by 112 → 3 : 2. Cancel the largest common factor you can spot.

Q8. What solving mode does "approximately what percent" signal?
A8. Estimation — round to the nearest friendly fraction (e.g. 33.1% → 1/3) and pick the closest option; don't compute exactly.

Q9. Compare two years' growth without full calculation.
A9. Compare the ratios New/Old. 360/300 = 1.2 (20%) beats 550/500 = 1.1 (10%).

Q10. Which DI questions do you attempt first?
A10. Single-cell / single-lookup questions. Defer grand-total and overall-average questions (which need every cell) to the end.

Q11. Most common careless mistake in tabular DI?
A11. Reading the wrong row/column or ignoring a unit. Underline the exact asked cell and its unit before computing.

Q12. Shortcut to average 210, 190, 205, 195, 200?
A12. Use base 200: deviations +10, −10, +5, −5, 0 sum to 0 → average = 200. Averaging small deviations beats adding big numbers.

Tabular DI is the backbone of every SBI PO paper. In Prelims, DI sets make up roughly 15–20 of the 35 Quant questions; in Mains, Data Analysis & Interpretation is its own 35-question, 60-mark section solved in 45 minutes. Tables also hide inside caselets and mixed graphs, so speed-reading a table is the single most transferable DI skill.

How to read a table (before you touch the questions)

1. Read the caption and every column unit — actual numbers, percentages, ratios, or "in '00/lakh/crore". Mixed units are the number-one trap.
2. Spot any Total row/column and any stated relation (e.g. Sold = Produced − Unsold).
3. Note whether the rows are years, branches, or products — this decides what "growth" and "share" mean.

Core operations

• Share of total = part/total × 100 (convert to a known fraction).
• Percentage change = (New − Old)/Old × 100 — the base is always the old value.
• Combined average of unequal groups = weighted average.

Branch	Employees	Deposits (Rs cr)	Loans (Rs cr)
A	120	480	360
B	80	400	300
C	150	600	540
Total	350	1480	1200

Example: Branch C's share of total deposits = 600/1480 ≈ 40.5%. Loan-to-deposit ratio for A = 360/480 = 3/4 = 0.75.

Exam Tricks & Tips

• 🎯 Read units column-by-column first; a single "in '00" multiplier changes every answer in the set.
• 🎯 Attempt single-cell lookups before grand-total questions — bank the easy marks, then return.
• 🎯 Average large numbers by base-deviation: subtract a round base, average the small deviations, add back.
• 🎯 For "approximately what percent", round to the nearest friendly fraction and pick the closest option.
• 🎯 Simplify a ratio of two cells by cancelling the largest common factor before dividing.
• 🎯 Underline the exact asked cell and its unit to avoid row/column misreads under time pressure.
• ❌ Don't take a simple average of two branch averages — if the branches differ in size you must weight them by count.

Expected exam pattern

A 5-question set built on one table (branches × metrics, or years × products). Typical asks: percentage share, ratio between two cells, percentage growth, a combined/weighted average, and one "which is highest/lowest" comparison.

Quick recap

Read units and totals first, work single-cell questions first, convert shares to fractions, always use the OLD value as the base for percentage change, and weight averages by group size.