Sanity-Check Against Something You Know in Guesstimates
In the previous move, you learned why rounded numbers beat false precision in a guesstimate. But clean arithmetic creates a new risk: a number can be mathematically neat and still be wrong in the real world. Sanity-checking is the final move that shows your estimate is anchored to something the interviewer can trust, not just to a tidy multiplication chain.
- A sanity check compares your final answer with an external benchmark such as a comparable market, named player capacity, or per-unit reality.
- It comes after rounded calculation and before you present the final range, so it acts as a trust layer on your estimate.
- The best sanity checks use something concrete: Ola target volumes, TVS+Bajaj+Ather combined volumes, Delhi IGI passenger movement, or JNPT plus Mundra throughput.
- A sanity check is different from a cross-check: a cross-check solves the problem another way, while a sanity check tests whether the result feels plausible against reality.
- For market-sizing answers, always convert the result into an interpretable unit such as lakh units, crore rupees, passengers per day, or revenue per MW.
- End with a range, not only a point estimate, because guesstimates are judged on directional accuracy and reasoning quality.
The Big Picture: Sanity Is the Trust Layer
In the universal guesstimate recipe, sanity-checking sits after rounded calculation and before the final range. You first clarify scope, choose an approach, sketch the tree, state assumptions, and calculate with rounded numbers. Then you ask: does this output match anything observable in the real world?
What a Sanity Check Actually Means
A sanity check is not extra decoration at the end of an answer. It is a separate reasoning step where you compare your final estimate to an external benchmark. That benchmark can be per-capita, a comparable market, a known company target, a public data anchor, or a bottom-up named-player build.
For example, if your estimate says India will sell around 38 lakh electric two-wheelers in FY28, the interviewer needs to know whether that number is plausible. The sanity check uses the source's named-player view: Ola alone targets 10 lakh per year by FY27, and TVS+Bajaj+Ather together add another 15-20 lakh. That creates around 30 lakh organised volume before adding the long tail.
Sanity check: compare your final answer to an external benchmark, such as a per-capita figure or comparable market, to test whether the estimate is believable.
Sanity Check vs Cross-Check
Candidates often use these terms interchangeably, but they are not identical. A cross-check means solving the same problem two ways, such as demand-side versus supply-side. A sanity check means comparing your result with a benchmark that already has real-world meaning.
The FY28 electric two-wheeler estimate gives around 38 lakh units from a top-down calculation. The sanity check is that Ola alone targets 10 lakh per year by FY27, while TVS+Bajaj+Ather contribute another 15-20 lakh, giving around 30 lakh organised units before tail brands. The strategic so what: the estimate is not just a percentage applied to a market, it is consistent with the scale of real named players.
The Three Benchmarks You Can Use
A strong sanity check does not need ten extra calculations. It needs one benchmark that is relevant to the output unit. In many interviews, one of these three benchmark types is enough.
- Named-player benchmark: Use known company volumes or capacity. In the EV-2W example, Ola, TVS, Bajaj, and Ather provide a bottom-up organised-market view.
- Unit-economics benchmark: Convert the answer into revenue per unit, cost per unit, or throughput per unit. For data centres, the source uses operational IT load capacity of around 900-1,000 MW, revenue per MW of around ₹12-16 crore per year, and 70-75 percent occupancy.
- Public-anchor benchmark: Compare with a known external anchor. For Indian airports, annual passenger traffic of around 37-40 crore converts to around 1-1.1 crore passengers per day, and Delhi IGI alone at around 1.9 lakh per day helps test whether the all-airport number is plausible.
Worked Example: FY28 Electric Two-Wheeler Market
Here is the complete sanity-check flow using the electric two-wheeler estimate from the source. FY means Fiscal Year, EV means Electric Vehicle, 2W means two-wheeler, and ASP means Average Selling Price, or the average price per unit sold.
The decision is to present 38 lakh units as the central estimate, with a range of 32-45 lakh units and revenue of around ₹42,000 crore. The outcome is stronger than a bare calculation because the interviewer can see both the top-down market logic and the named-player reality. The learning is simple: a sanity check converts a calculated answer into a defensible answer.
Why Sanity Checks Matter in Consulting Interviews
In a case interview, the final number is rarely the only thing being tested. Interviewers want to see whether you can move from arithmetic to business judgement. A candidate who says "38 lakh units" has completed the math; a candidate who says "38 lakh units, and that is plausible because Ola, TVS, Bajaj, Ather, and tail brands can explain the order of magnitude" has shown market sense.
This is especially important when your assumptions are rough. Guesstimates use rounded inputs by design, so the answer must survive a reality test. If the sanity check fails, you should revisit the assumption that drives the error: penetration, frequency, price, occupancy, conversion, or total base.
How to Speak the Sanity Check Out Loud
The best delivery is short and explicit. Do not hide the benchmark inside a long explanation. Say what your estimate is, name the benchmark, compare the two, then state whether you are comfortable with the result.
"My estimate is [X]. To sanity-check it, I will compare it with [known benchmark]. Since the benchmark implies [Y], my estimate is plausible / too high / too low, so I would present a range of [A-B]."
For the EV-2W case, the spoken version would be: "My estimate is around 38 lakh EV two-wheelers in FY28. As a sanity check, Ola alone targets 10 lakh per year by FY27, and TVS+Bajaj+Ather together add 15-20 lakh, so around 30 lakh organised volume plus tail brands makes the 32-45 lakh range plausible."
Where Candidates Lose Marks
The most common issue is doing a sanity check at the wrong level. If the estimate is annual revenue, a unit-volume benchmark alone is incomplete unless you also convert units to rupees. If the estimate is daily passengers, an annual traffic benchmark must be divided by 365 before comparison.
The source examples show this conversion discipline repeatedly. Indian airport traffic of 37-40 crore annually becomes around 1-1.1 crore passengers per day. JNPT plus Mundra throughput of 160-170 lakh TEUs annually becomes 4,400-4,600 TEUs per day before adding laden, empty, feeder, and transshipment moves to reach around 8,000-9,000 TEU-movements per day.
Structuring a Sanity Interview Answer
"Estimate the FY28 electric two-wheeler market in India and sanity-check your answer against something real."
The number one error is stopping after the spreadsheet-style answer. Interviewers reward the candidate who adds, "Here is the real-world benchmark that makes me comfortable with this estimate."
Conclusion
Sanity-checking is the move that turns a neat estimate into a credible business answer. Use a benchmark the interviewer recognises, compare it at the same unit level, and present a range that reflects both your math and your judgement.
The most frequent mistake is saying "this seems reasonable" without naming the benchmark. That costs points because the interviewer cannot tell whether your confidence comes from market knowledge or from wishful thinking!