Related Posts
Additional Posts in Tech Strategy & Product
What’s a good CISA program?
Product Management at a large bank vs Business Analytics/S&O for FAANG? Recently started in the former role, but have interview calls for the latter just come up in my mailbox. Similar comp when adjusted for the different job locations. Can anyone help me with the Pros and Cons please. I know the roles are different, and so are the industries, need to understand difference career paths and difference in corporate cultures. JPMorgan Chase Google LinkedIn Citi
New to Fishbowl?
unlock all discussions on Fishbowl.






I recently stepped into a startup with no product function, so I know the pain. There are a lot of directions you can go, but as a rule of thumb, start by making feedback capture easy and consistent. I use Airtable to log things like customer, product area, request type, priority, etc. Even if the feedback isn’t immediately relevant or tied to the roadmap, it’s way easier to spot themes later when you can filter for things like frequency or customer segment. The key here is something that easy to sort/review.
Feedback is only one side of the equation though; user behavior is another. Personally, I’m a big fan of PostHog to see how people are actually using the app, where they get stuck, and what’s breaking. Even pulling basic product stats in SQL can surface behavior insights.
Sometimes it’s feedback validated by behavior, other times it’s the data that sends me digging into feedback or researching behavior.
Happy to connect and share more if that’d be helpful.
As a PM who originally comes from the research world, I’m used to performing open ended coding and advanced analysis on user feedback. However few in the product world spend this level of effort- often times just looking for the right validations for their assumptions.
If you’re in the position where you are looking at lots of raw feedback with limited resources or knowledge to evaluate you could look at research tools on the market or start simply with loading the data to LLMs and prompting the models to carefully analyze and evaluate the results. Depending on the amount of records - ask the models to validate their every step of the process. This will take the burden off of you to evaluate results at a high level and for limited expense you could get some detailed direction on the data and ideas for what to build from there.
Coach
I'll have to check what our data/LLM policy is, but I think Procurement is working on an Enterprise plan with OpenAI, so hopefully that get's finalized soon (and I get a license lol)