アメリカ就活｜SoFiのシニアプロダクトマネジャーの面接での会話を書き起こしました 【全て英語】

前置き

前日「アメリカでのプロダクトマネジャー面接攻略」と言ったテーマの記事をブログにて書きました。

本noteはその記事の補足ですので、もしプロダクトマネジャーの面接攻略に興味がある方はぜひ「アメリカでのプロダクトマネジャー面接攻略」を読んでからこちらを読んでください。

アメリカでのプロダクトマネジャー面接攻略【実際の英語での面接のやり取りの文字起こしも公開】

PMには興味ないけど、英語での面接を見てみたいという方ももちろん大歓迎です。

また、前提としてこの記事は本当にただただ少し前に受けたSoFiのシニアプロダクトマネジャーのポジションの面接を文字に起こしたもので解説等はブログで書いていますので、そこはご了承ください。

個別で質問や相談がある方は僕の無料のニュースレターを購読してそのメールに返信してください。ニュースレターへの返信という形で送られてきた質問には必ず返信します！

Toma Letter | tomablog

また、僕のTwitterのDMできた質問にも答える可能性はあるので、フォローお願いします。

では、以下面接の書き起こしです。

I: Interviewer (面接官）
T: Toma （僕）
となってます。

では、どうぞ！

---

本編

Topic:
Our team needs to build a chat functionality for the app to get and answer personal finance questions. How would you design this?

I -> Interviewer
T -> Toma (トーマ)

I: Ok. So my team is assigned to this project to build an in-app chat functionality for the users to ask and get answers to any type of personal finance questions. And I want you to lead this initiative. How would you design the solution to this?

T: Ok, cool. I am excited to take a lead on this. Before diving into actually coming up with the solution, I would like to structure my approach to this. Could you give me some time?

I: Sure. Take your time.

T: Thank you! It should be quick.

2 minutes passed

T: Alright, thank you for being patient Nicole.
So first, I would like to gather more insights into users and their pain points as I believe understanding users and their problems is critical for coming up and designing a good product.

Then from there I define the product goal like what we essentially want to achieve through this product. Also, I would like to know some constraints if there is any before brainstorming the solutions.
Once those are defined. I will then spend some time coming up with solutions and making priority for them.

At the end, if we have time, I plan to think about success metrics so that we can properly measure the success of the product and keep improving it if necessary.

Does this sound good to you?

I: Yeah. That sounds good to me!

T: Awesome! Then, let me ask some questions about the users and their pain points. I know that SoFi’s main target users are relatively young generations who are starting their personal financial journey. So is it safe to assume that our target user is Gen Z people?

I: Yeah. That assumption is accurate.

T: Cool. And you mentioned that we need to design this chat functionality to help the users ask personal finance questions and get answers to them. How did we find out there is a need for that? I am assuming that they are starting out their personal finance journey and have a lot of questions, but they could google those information, right? Was there anything that got us to think we should have the chat functionality to replace that in app?

I: That’s a great question. So those who are starting out in their personal finance journey had a problem that there is too much information about personal finance online which makes it actually harder to know which source they can trust.

T: I understand. Thank you for the context. When I was starting out on my personal finance journey, I also struggled finding the right information on search engine like Google as I knew little and didn’t know where to start. So now I am thinking that we could use this chat functionality to guide them to find the right information handy.

I: Yeah that sounds good.

T: Ok. So now I think I have user persona and problems defined here. The target user for this feature is Gen Z people who are just starting their personal finance journey using our app, and their biggest pain points are:

1. They don’t know which source they can trust

2. They want to get the answer to their exact question without having to browsing lots of websites and apps.

Did I get them right?

I: Yeah. That sounds spot on.

T: Great. Before thinking about the potential solutions, I have one question. Do we have any resource constraints like money, time or engineering resource?

I: No. You can assume there is no such constraints at all for this project.

T: Sounds good. Ok, so now I would like to spend some time thinking about potential solutions and pros and cons for those solutions so I can decide which one I would like to go with. Could you give me some time for that?

I: Sure! Take your time.

T: Thank you Nicole.

3~5 minutes have passed

T: Ok. Thank you for your patience again Nicole! My solution is AI-powered multi-language personal finance assistant in-app chatbot, and I would focus on these two things when we implement this:

1. Interactivity — This AI model is fully interactive, meaning that users can ask generic questions and AI model will followup more questions to identify what they are asking exactly so that it can give the users the best answers.

2. Internationalization — Users can ask any questions in any commonly used languages, and AI model will answer in the language they asked the questions in.

I also thought of

1. Personalization — This AI model will grow in house, meaning that the more users use this app the more data it gets, the more personalized the assistant could be.

However, this is an enhancement and not required for MVP as I believe the first two should be enough to solve users problems.

I: Cool. Those sound like good ideas, and I agree that personalization would be an enhancement and not required for the initial release.
And yeah, everyone is talking about the AI, and I am curious how this chatbot is different from a bunch that exist out there.

T: Good questions. Let me explain those solutions a bit more in detail to give you better ideas on how this solution is unique and will help users solve their problems.

1. This AI model is fully interactive, meaning that users can ask generic questions and AI model will followup more questions to identify what they are asking exactly so that it can give the users the best answers. This should make the users feel like they are talking to people on the phone or teller at the bank which could help them trust us. Also, it will eventually introduces them to the appropriate educational resource like blog post that we publish. This way, we can promote our resources and other products, while ensuring that we are providing the accurate information to the users.

I: I like your idea of using our educational resources to help them get the right answer while introducing to more of our products. And I remember you mentioned internationalization.

This is really interesting and I never thought of that. How did you come up with this idea? And why you think it’s important to implement?

T: I am happy to answer that. This actually comes from my personal experience. I have a lot of international friends that work in the US, and I realized that they try to search anything related to personal finance in their own language.

Of course there is not a lot of information on Google in their language about American money system and if there is any, often times it’s not that accurate or detailed.

Considering a lot of people in the US speak English as their second or third language, I thought it would be great to make this feature even more accessible by allowing them to use any language so that they know we can give them right information right away by asking questions in their own language.

Also, recent AI models can handle text-based internationalization really easily. So I thought the value we can get vs. implement cost is really high here.

I: Interesting. Yeah, we still need to a bit more research into how many people actually would like to use another language, but that’s definitely an interesting and good thoughts. So now I know your solution, how would you measure the success of this product?

T: Ok. To measure the success of this product, I would use two metrics:

1. Ratings or any comments on App Store

2. Accuracy Score through quick survey while chatting with customers

#1 is easy and intuitive. It’s simply observing the change in ratings of our app on app store before and after the release of this feature.
#2 is a bit more complex, but this should give us better insights into this feature’s performance.

So, just like I explained earlier, users will interact with the chatbot. I would like to implement a quick survey that will popup right when we give them the answer to their questions. Not for every single question, but when the chatbot gave the user the educational resource which it thought was the most helpful for the users. Essentially, it would ask “Did you get the help you needed?” from 1 to 10 they would score it. This data should help us identify if it’s actually giving them the need they needed. If this is equal to or above 7 I would say this feature is successful.
Also, this data will help us keep improving our chatbot as it would allow us to know where it’s failing and what type of questions it sucks at answering.

Do you think those are good metrics? Do you have any question?

I: Yeah, that sounds good. And no I don’t have any question. I think your answer was pretty clear.

T: Awesome!