How-to guides
See spend by feature
See AI spend broken down by feature, and whether that spend turns into outcomes.
The provider invoice shows a single number, the total. Nowhere on it does it say how much went to review summaries, support replies, or search assistance. Cutting cost starts with splitting that total by feature.
Where is the money concentrated
Section titled “Where is the money concentrated”Ask Insights “which feature costs the most this month” and you get the features ranked by cost right away. It keeps the context, so follow-ups like “compared to last month?” and “when did it start climbing?” work, which makes it a good place to pick up a lead when you do not yet know which screen to open.
Once you have the ranking, open Overview on the 30-day range and cycle the feature picker to check it against the raw numbers.
Is that cost growing along with outcomes
Section titled “Is that cost growing along with outcomes”Overlay the success metric line on the Overview trends chart. A feature whose cost line and outcome line rise together is growing well. A feature where only the cost line rises is where the work is.
Is the spend coming back as revenue
Section titled “Is the spend coming back as revenue”Set up a success metric like “revenue per AI dollar” for each feature (see Success metrics for how to define one) and you can compare cost efficiency across features on the same scale. A feature that returns ten times its spend and one that does not break even call for different priorities.
Where your finding takes you next
Section titled “Where your finding takes you next”| Finding | Next guide |
|---|---|
| Spike from a specific date | Respond to a cost spike |
| Expensive model on trivial work | Switch to a cheaper model |
| Large prompt tokens | Trim prompt costs |
| No metric for outcomes | Prove the impact |