Meta Senior Data Scientist Interview Experience — Onsite Case on a New Chat App, Ending in a Spaghetti-Chart Slide Critique

The interviewer had many years of experience and had a friendly, easygoing demeanor. Overall I felt our conversation flowed pretty smoothly, back and forth, and I was able to quickly follow up on a lot of the points he raised and give balanced, two-sided answers. Structurally I stuck to the standard framework (clarify, goal, pros/cons, hypothesis, data points, ...). Maybe because this was L5, the interviewer's questions were deliberately broad, with a lot of room for me to make my own trade-offs.

Case (since I interviewed mid-year, I don't remember it that clearly anymore, so I'm reconstructing the context as best I can — if anyone else has gotten this case, please feel free to add more. It's a pretty niche question.)

Meta launches a new chat app in Europe, focused on the B2C space. Businesses can use the chat tool to talk to customers for customer service, and Meta monetizes it by charging businesses a monthly subscription. (I spent some time clarifying this setup with the interviewer, because I'd genuinely never seen a case like this geography-wise — I thought it was a pretty unusual question...)

From what I remember, roughly the questions were:

1. How to justify this product (tie it back to Meta's vision and goals).
2. How to measure success (building on the goal we'd just discussed, define success metrics and guardrails. Since it's a service chat product, I talked about metrics related to service resolution quality and subscription revenue: # of businesses with at least one chat, # of chats started by customers per day, # of chats resolved, # of messages per chat, monthly subscription revenue, # of companies subscribed. Guardrails: long-duration chats, % of chats not resolved — he followed up asking what counts as "not resolved," and I gave two approaches: either look at whether the conversation sentiment was positive, or directly survey the customer after the conversation.)
3. How to test it after launch (since this is at the business level, the sample size is small, so I said we could use causal inference, DiD, etc. — this question got moved past pretty quickly).
4. The one that stuck with me most was the last question. He asked me how I'd show results on a one-page PPT for the C-suite, and first showed me a slide with a spaghetti chart on it — I remember it was some metric's trend across different regions — and asked me to critique the slide. (I first explained what the slide was trying to communicate, said a few things I liked about it, and then went for it: I said the spaghetti chart was too messy, the y-axis was compressed because of the different scales, so it was hard to make out the trend for smaller countries; also the metric was too narrow and couldn't show the full picture. I said a few other things I've since forgotten.) Then he asked how I would organize the page instead, and what metrics I'd use. (Since this is a new app and a B2B product, I felt that for a C-suite audience you'd want to emphasize high-level metrics that highlight early-stage growth and profitability. So I picked monthly subscribed companies w/ MoM%, monthly subscription revenue w/ MoM%, and ROI.)

Has anyone else gotten this same case? Could you help check whether my understanding of the prompt was right — did I misunderstand something???