Skip to content
Notifications
Clear all

Thoughts on the new usage analytics dashboard? It's a start.

5 Posts
5 Users
0 Reactions
2 Views
(@gracyj)
Trusted Member
Joined: 1 week ago
Posts: 61
Topic starter   [#15356]

Just logged in and saw the new dashboard! It's definitely a step in the right direction. Being able to see total credit usage at a glance is super helpful for managing our team's budget.

But, I wish it went a bit further. For our customer success workflows, we really need to break down usage *per project* or *per team member* to see where the credits are actually going. Right now it feels a bit high-level. Still, glad they're listening and iterating! What's everyone else hoping to see in future updates? 🤔

xo


Happy customers, happy life.


   
Quote
(@jessicaw)
Trusted Member
Joined: 1 week ago
Posts: 28
 

Yeah, the high-level view is nice for a quick check. But I think you're right about needing more detail. In our small team, I'm already wondering which chatbot conversation used up so many credits last week.

Are there any workarounds right now for seeing per-project usage? Maybe in the raw data exports? I haven't found one yet.



   
ReplyQuote
(@bobw)
Estimable Member
Joined: 1 week ago
Posts: 77
 

Good point on needing that granular data. For the chatbot use case specifically, I've found a temporary workaround by tagging outgoing requests with a custom `X-Project-ID` header and then piping our logs through a simple external analytics tool. It's messy, but it does show us which conversation thread or integration is the heavy hitter.

Haven't seen a per-project breakdown in the raw exports either, which is a real shame. The API billing endpoint seems to only reflect the same aggregated data as the new dashboard.

I'm hoping they open up a more detailed usage API soon. Tracking it all yourself externally feels like we're building the dashboard *for* them.


null


   
ReplyQuote
(@avag2)
Estimable Member
Joined: 1 week ago
Posts: 95
 

The raw data exports are just as aggregated as the dashboard, unfortunately. You're right to be suspicious about a specific chatbot conversation - without granular data, you can't pinpoint a sudden spike to a user, a new feature, or a problematic prompt causing excessive generation.

For a temporary fix, you need to instrument it yourself. Log every request with a unique session or user ID before it hits the API, then correlate that with the timestamps in your credit usage. It's manual, but it's the only way to get that "which conversation" answer right now.

Until the platform provides this natively, you're essentially forced to build your own shadow billing system.


Show me the benchmarks


   
ReplyQuote
(@annaw)
Estimable Member
Joined: 1 week ago
Posts: 96
 

Totally agree about the feeling of building the dashboard for them. Your workaround with the custom header is clever, but that's exactly the kind of lift that gets frustrating over time - maintaining external logs and tools shouldn't be the solution for basic cost attribution.

> hoping they open up a more detailed usage API soon

This is the real unlock. For our implementation clients, a detailed API would let them pipe usage directly into their own cost-center dashboards or project management tools. The dashboard itself is a nice start, but the data behind it is what enables actual workflows.

Until then, we're all just stitching systems together.



   
ReplyQuote