How to approach ambiguous multiple-choice questions (MCQs) questions in BCG Case Chatbot?

Hey,

Let me be super-honest with you: I spoke about this with a few MBA friends who now work at BCG and they told me that they don't love the chatbot either, so you are not alone in complaining :)

What I would do is trying to write down the clear logic link between the different risks and the investment. Here:

1. Regulatory risk: if the regulator limits this ex-post (e.g., up to a certain market share, or limiting usage), the value of the investment is likely to be reduced.
   1. Very clear link -- good!
   2. Can we falsify this? If regulators set limitations, our investment is not likely to be reduced. I can't -- this confirm my selection
2. Interdependency risk: if the other operator changes their strategy, exits a specific geographical market, etc., the investment is likely to lose value (or, I might have to cover more OPEX).
   1. Very clear link -- good!
   2. Can we falsify this? If the other operatore changes their strategy, our investment is likely not to be reduced. Well, it technically can, but I really have to find edge cases to falsify the hypothesis (e.g., we have more 5G supply for our clients and we can monetize this) -- I would keep it
3. Risk that one company may not want to continue investing due to lower calculated profitability: this feels very similar to the interdependency risk. Very clear link (as above) and hard to falsify -- same as above
4. Decline in phone sales market: What matters is the evolution of the demand of 5G, that does not logically require phone sales to grow.
   1. Weaker link -- I would not select this!
   2. Can we falsify this? If the phone sales decline (e.g., people do not buy iPhone 18 but stick to 17), the 5G usage is not directly impacted. I say yes -- so I would confirm the 'no selection'.
       

Again: overall, I agree with you on how this exercise can not be treated deterministically (i.e., prompt x always and only leads to y in any case and with any realistic assumption). However, I still think that the idea of writing down the relationship and trying to falsify it might help :)

Best,

Tom

Hi Rada,

In general, the BCG chatbot is intentionally designed to include some ambiguity, so it’s normal to come across questions where more than one answer feels defensible. Also, keep in mind that you don’t need to be perfect to pass; the bar is relatively low, so missing one or two tricky questions shouldn't be a deal-breaker

In your specific example, I think the key is to stay very close to the wording of the question. The question is about the risks of a joint venture, not the broader market. Regulatory risk, misalignment between partners, and interdependency are all directly linked to the JV itself. On the other hand a decline in phone sales is a general market risk that would affect the company regardless of whether the JV happens or not.

So the main recommendation is to be very precise in how you interpret the question. A useful trick is to quickly paraphrase it in your head: “What are the risks specifically tied to entering a joint venture with a partner?” That usually helps filter out broader, less relevant options

Hope this helps.  If you’d liketo practice together or go deeper on this, just let me know.

Best,
Franco

Hi Reda,

The key thing to understand is that you’re not being tested on “what could be true,” but on “what is most directly relevant to the case”.

In your example:

• regulatory risk → clearly linked to 5G infrastructure
• interdependency → directly linked to the JV
• partner misalignment (profitability) → also directly linked to the JV

“Decline in phone sales” is not wrong in real life, but it’s more indirect / second-order. That’s why it’s excluded.

How to approach these MCQs

Think in terms of:

1. Direct vs indirect impact
Pick answers that are immediate and structurally tied to the situation.

• JV → think partner risks
• infrastructure → think regulatory, capex, execution

Avoid broader market effects unless they are central.

2. Stay close to the question
“What are the risks of this decision?”

Not:
“What are all possible risks in the industry?”

3. Look for “testable logic”
The chatbot expects answers that follow a clean, almost textbook logic:

• partnership → misalignment risk
• regulated industry → regulatory risk

4. Don’t overthink / over-expand
In a real interview, you could argue your point.
Here, you need to align with the expected logic, not defend alternatives.

So overall:

• prioritize direct, first-order risks
• stay tightly linked to the prompt
• avoid answers that are “true but less central”

Once you think this way, these questions become much more predictable.

hey Rada :)

you’re right, these questions can feel ambiguous, but the key is that the BCG chatbot is not testing “defensible in discussion” thinking, it’s testing whether you pick the most direct, case-relevant drivers.

in your example, the trick is to stay very tight to the core situation: joint venture for 5G infrastructure. that immediately points to partner-related risks. so regulatory risk fits, interdependency fits, and misaligned incentives or profitability also directly sits in that bucket. decline in phone sales is more indirect and second order, so even if it’s logically possible, the chatbot will usually not count it.

a good mindset is to ask yourself: is this a first order risk that directly affects the decision, or a broader market factor that’s only indirectly linked? the chatbot almost always rewards the most direct, structural answers over wider industry logic.

also helpful is to think in buckets: here it’s something like external, partner, market. then pick the options that best represent the most critical buckets for this exact situation, not all plausible ones.

so in short, optimize for how “close” the answer is to the core problem, not how defensible it is in a real conversation.

best, Alessa :)

Hi there,

Ambiguous MCQs in the BCG chatbot are by design, not by accident. They're testing how you prioritize and decide under uncertainty  which is exactly what consultants do every day.

The approach:

1) Don't get stuck looking for the "perfect" answer. There often isn't one. Look for the dominant answer, the one that is most defensible given what you know.
2) Apply the same logic you'd use in a real case: what is the most important driver here? What does the data actually support (vs. what sounds plausible but isn't directly evidenced)?
3) When two answers feel equally valid, ask yourself: which one would a consultant default to given limited information? Usually it's the one that is most conservative and most supported by the data in front of you.
4) Don't overthink it. These tests are timed. Make a call and move on.

For what it's worth I have 10 of the past real BCG chatbot tests in my prep pack. If you grab the pack, I'm happy to review 1-2 of the questions you found most ambiguous and walk through the reasoning. Feel free to message me.

Good luck — fingers crossed!

Fair frustration. The BCG chatbot MCQs are a known weak spot. In a real interview you could argue your choice, here you cannot.

A few things that help:

• Stay close to the direct question, not side risks. Ask, is this risk specific to the decision, or general to the industry. The chatbot wants the specific one.
• Think about what would make the deal fail, not what makes the business hard. MCQs lean toward risks inside the structure, like partner alignment, interdependency, governance. Not outside market shifts.
• When in doubt, pick what a deal team would discuss in a pre-signing meeting.

In your 5G JV example, "decline in phone sales" hits both companies equally, so not a JV risk. Profitability misalignment between partners is, because it can break the deal.

Treat the chatbot as a pattern tool, not a real interviewer. Learn what it rewards, move on, and save real reasoning for human practice.

Good luck.