This interviewee really has a lot of knowledge in product and business development, however, he did not get Jay's hint to quantify those impacts 😂 My approach: 1. let it run some time for a small portion of people to see if some key metrics change (ie. spent hours as key indicator for retention measurement) 2. a/b test, and select some key metrics as comparisons, it provides more accurate results. Not sure if I am on the right track?
I think the one place where this may fall short is that Jay specifically talked about the fact that they don't have a podcast portion of their platform, and would need to allocate many engineers for a good portion of time to build it out. Maybe the answer then is just add a Step 0 on your approach and build a lightweight podcast interface?
I feel this is more like an awesome product job interview, but ds product interview. However it provides a different perspective for me as a DS, thanks!
Where is the data science in the answer? Personally I would have tried to come up with ideas to estimate cross-selling from current subscribers to podcast propositions. Using somehow the customer data available plus some external data possibly (similar interest/etc...). Or design a marketing survey on current customers, giving an idea how much data would be needed to infer something on the whole customer population.
Can the interviewer summarize the ideal answer at the end. Honestly speaking, the whole dialogue is not organized. The logic is random. I think you can answer the question by some assumptions
I am yet to see a company where a data scientist is doing this. This is a classic product manager case…
PMs are sometimes hired under the guise of “data scientist”
Sounds like the FB model
This interviewee really has a lot of knowledge in product and business development, however, he did not get Jay's hint to quantify those impacts 😂 My approach: 1. let it run some time for a small portion of people to see if some key metrics change (ie. spent hours as key indicator for retention measurement) 2. a/b test, and select some key metrics as comparisons, it provides more accurate results. Not sure if I am on the right track?
I think the one place where this may fall short is that Jay specifically talked about the fact that they don't have a podcast portion of their platform, and would need to allocate many engineers for a good portion of time to build it out. Maybe the answer then is just add a Step 0 on your approach and build a lightweight podcast interface?
I feel this is more like an awesome product job interview, but ds product interview. However it provides a different perspective for me as a DS, thanks!
amazing content, please post more!
Where is the data science in the answer? Personally I would have tried to come up with ideas to estimate cross-selling from current subscribers to podcast propositions. Using somehow the customer data available plus some external data possibly (similar interest/etc...). Or design a marketing survey on current customers, giving an idea how much data would be needed to infer something on the whole customer population.
It's a data-driven answer which works for most data science positions
Can the interviewer summarize the ideal answer at the end. Honestly speaking, the whole dialogue is not organized. The logic is random. I think you can answer the question by some assumptions
He would have failed if this was an actual interview.
Is this even a data science interview ?