To me what's helpful is that I have all events in a database. So I can ask Claude Code to do something like this:
1. Review all my code (and take screenshots if needed)
2. Check all existing logged events and see how often they are triggered
=> Based on the *actual user behavior*, what are things users might be doing that we're not tracking?
A basic example: I had lots of pages with good traffic where I forgot to track the key CTA. Are there other ways of finding out that I forgot this? Yes, but there's so many things I might forget that I want some way to review this in natural language.
Of course, that might already exist, and please let me know if there are tools that do this well.
Ok interesting — I had a similar issue and solved it an entirely different way.
I had Opus 4.8 with Playwright MCP step through a web flow, taking note when analytics events didn't fire, or fired with useless payloads. I then audited this by comparing those results against the codebase. It worked pretty well!
I like that with the DB approach you can query by user behavior, that is useful. In my case I was trying to just reach 100% coverage of all non-trivial user behavior.
Interesting! I'd say those are very complementary: This type of integration test (I 'll try to write post talking about what tests I run, mainly to think through it more clearly) tell you that stuff you have is working and then (for me) it's useful to think about how it can be better, and how I can quickly measure features. We're all learning how to build in this new world.
“Given logging we have, what are valuable interactions we don’t track?”
How does using deebeelog help you answer this?
To me what's helpful is that I have all events in a database. So I can ask Claude Code to do something like this:
1. Review all my code (and take screenshots if needed)
2. Check all existing logged events and see how often they are triggered
=> Based on the *actual user behavior*, what are things users might be doing that we're not tracking?
A basic example: I had lots of pages with good traffic where I forgot to track the key CTA. Are there other ways of finding out that I forgot this? Yes, but there's so many things I might forget that I want some way to review this in natural language.
Of course, that might already exist, and please let me know if there are tools that do this well.
Ok interesting — I had a similar issue and solved it an entirely different way.
I had Opus 4.8 with Playwright MCP step through a web flow, taking note when analytics events didn't fire, or fired with useless payloads. I then audited this by comparing those results against the codebase. It worked pretty well!
I like that with the DB approach you can query by user behavior, that is useful. In my case I was trying to just reach 100% coverage of all non-trivial user behavior.
Interesting! I'd say those are very complementary: This type of integration test (I 'll try to write post talking about what tests I run, mainly to think through it more clearly) tell you that stuff you have is working and then (for me) it's useful to think about how it can be better, and how I can quickly measure features. We're all learning how to build in this new world.
By the way I DMed you on Twitter - do you want to chat / meet up?