Caselet DI for SBI PO

In a caselet the data is hidden in a paragraph. There is no table, no chart, just five or six sentences that each give one number or one relationship. Candidates who read the paragraph again for every question lose minutes. Candidates who turn it into a table first answer most questions by looking something up.

Caselets are a regular part of the DI mix in bank PO mains papers, and they can appear in the prelims too. They are harder to skim than a table, so they often decide who finishes the section. This guide covers the method, the wording that causes errors, a fully worked caselet and a practice set.

The method

1. Read the whole paragraph once, without calculating. Find the categories. There are usually two layers, such as product and channel, or department and gender.
2. Draw a two-way table with one layer as rows and the other as columns, plus a total row and a total column.
3. Fill cells in the order the paragraph allows. Some sentences give a cell directly; others only make sense once another cell is known.
4. Use the totals to finish. The last cell in any row or column is usually total minus the rest.
5. Answer every question from the table, not from the paragraph.

Why a table works: every question in a caselet is a look-up, a sum, a ratio or a percentage of cells. Once the cells are filled, the paragraph has nothing left to tell you.

Wording that causes errors

A worked caselet

A bank branch sanctioned 1,440 loans in a quarter: personal loans, home loans and gold loans. Personal loans made up 5/12 of all loans. Home loans were 20% fewer than personal loans, and the rest were gold loans. Each loan was sanctioned either through the bank's app or at the branch. 55% of personal loans came through the app. Home loans through the app and at the branch were in the ratio 1 : 3. In all, 600 loans came through the app. (Made-up data for practice.)

Building the table, one sentence at a time:

• Personal = 5/12 × 1,440 = 600.
• Home = 600 × 0.8 = 480.
• Gold = 1,440 − 600 − 480 = 360.
• Personal via app = 55% of 600 = 330, so personal at branch = 270.
• Home via app = 480 ÷ 4 = 120, so home at branch = 360.
• Gold via app = 600 − 330 − 120 = 150, so gold at branch = 210.
• Branch total = 1,440 − 600 = 840. Check: 270 + 360 + 210 = 840.

Notice that gold via app could only be found from the column total. That is typical: one cell is deliberately left for you to derive.

Q1. How many gold loans were sanctioned at the branch?

• 210, straight from the table.

Q2. What percentage of all loans were sanctioned at the branch, to two decimals?

• 840 ÷ 1,440 × 100 = 7/12 × 100 ≈ 58.33%.

Q3. What is the ratio of personal loans through the app to gold loans at the branch?

• 330 : 210 = 11 : 7.

Q4. The average home loan was ₹24 lakh and the average personal loan ₹4.5 lakh. By how much did the value of home loans sanctioned at the branch exceed the value of personal loans sanctioned through the app?

• Home at branch: 360 × 24 = ₹8,640 lakh.
• Personal via app: 330 × 4.5 = ₹1,485 lakh.
• Difference = ₹7,155 lakh = ₹71.55 crore.

Q5. Next quarter, app loans rise by 20% and branch loans fall by 10%. By what percentage does the total change?

• App: 600 × 1.2 = 720. Branch: 840 × 0.9 = 756. Total = 1,476.
• (1,476 − 1,440) ÷ 1,440 × 100 = 36 ÷ 1,440 × 100 = 2.5% rise.
• Faster: the total change is a weighted average of +20% and −10%, with weights 600 and 840. (600 × 20 − 840 × 10) ÷ 1,440 = 3,600 ÷ 1,440 = 2.5.

Overlap caselets

Some caselets describe groups that overlap: customers using UPI, cards or both. Here the table becomes a small Venn diagram in numbers. The rule is that only A = A − both, and total users = A + B − both. Subtracting "both" once stops those people being counted twice.

Common mistakes

• Calculating before the paragraph ends. A later sentence often changes what an earlier one means, such as "the rest" or a total given last.
• Taking a percentage of the wrong group. "55% of personal loans" is 55% of 600, not of 1,440.
• Skipping the check. Once the table is full, confirm that every row and column adds up. A cell that breaks the total is a misread sentence.
• Re-reading for each question. If you find yourself going back to the paragraph, your table is missing a row or column. Add it once.

Practice set

A bank surveyed 900 customers. 520 said they use UPI, 380 said they use a debit card for payments, and 150 said they use both. The rest use neither. Of the customers who use neither, 60% are over 60 years old. (Made-up data.)

1. How many customers use neither UPI nor a card?
2. How many use only UPI?
3. What percentage of the customers use exactly one of the two, to two decimals?
4. Of the card users, what percentage also use UPI, to two decimals?
5. How many customers who use neither are over 60?
6. From the worked caselet: what percentage of home loans came through the app?

Answers:

1. 150. Users of at least one = 520 + 380 − 150 = 750; 900 − 750 = 150.
2. 370. 520 − 150.
3. 66.67%. Only UPI 370 + only card 230 = 600; 600 ÷ 900 × 100.
4. 39.47%. 150 ÷ 380 × 100. The base is card users, not all customers.
5. 90. 60% of 150.
6. 25%. 120 of 480, which is the 1 : 3 ratio read as 1 part in 4.

What to do next

• For the next week, do one caselet a day. Time the table-building step separately from the questions. Aim to build the table in under two minutes.
• Practise at least two overlap caselets, since the "both" logic is easy to forget under pressure.
• Move on to missing data DI, which uses the same fill-the-table skill.
• Revise percentages, ratios and averages inside DI for the base-of-comparison traps.