Read the Insights page

The dashboard tells you how your account is doing. Insights tells you how one tenant is doing, and — more usefully — what to change. It is where cache effectiveness, per-rule performance, and the ranked list of caching opportunities live.

Dashboard or Insights?

They answer different questions, and reaching for the wrong one is the usual reason a number looks strange.

Insights follows the tenant switcher in the App header. If the numbers look unfamiliar, check which tenant is selected before checking anything else.

Cache effectiveness and response time

The top band is the tenant's hit rate and its response-time profile. Both are measured — they are counts and timings the Gateway observed on real traffic, not estimates.

Read the response-time numbers as the Gateway's own round-trip wall clock: the time from the Gateway receiving a statement to it returning a result. On a miss that includes network, your warehouse, and downloading the body. It is not your engine's internal execution time, and Airbrx has no way to see that number.

Rule performance

Per-rule effectiveness answers the question that actually drives tuning: which of your rules are earning their keep?

Two shapes are worth acting on:

• A rule that matches a lot and hits rarely. The rule is catching statements but the cached entries are not being reused. Usually the cache key is too specific — keyed per user on a query that is not per-user, or including a parameter that varies on every call — or the TTL is shorter than the interval at which the query is actually run.
• A rule that matches nothing. Either its conditions are wrong, or a rule above it is matching first. Cache rules are first-match-wins in priority order, so a broad rule with a higher precedence will shadow the specific one you wrote afterwards. The rules analyzer flags this shape directly.

Opportunities

The opportunities list is the most actionable thing on the page. Airbrx analyzes statements that went to your warehouse and ranks the ones that would have cached well. Each one arrives with a suggested action, and the list is grouped by that action rather than by how it was detected:

The same list appears in the rules workshop as a Recommendations panel, so you can act on an opportunity in the place where you would write the rule. Insights is where you decide what to do; the workshop is where you do it.

Treat "Cache for longer" with the freshness question in mind. Airbrx will happily extend a TTL past the point where your users would notice stale data — it does not know your business's tolerance. Where the answer depends on a clock boundary rather than a duration, reach for a TTL strategy instead of a raw number, so the entry expires when its answer actually changes. See Invalidation strategies.

When the page is empty or stale

Insights reads pre-aggregated summaries rather than raw logs, and those summaries are built by a background process. On a tenant with steady traffic, new activity appears within seconds. On a quiet tenant, a summary may not be rebuilt until the daily pass — so a tenant that ran three queries this morning can legitimately show nothing until tomorrow.

An empty Insights page on a busy tenant is a different signal. Check the dashboard for volume: if it is also zero, traffic is not reaching the Gateway at all and the problem is a connection pointing around it, not a reporting delay.

Where to go next

• Read the dashboard — the account-wide view and what its numbers do and do not measure.
• Cache rule cookbook — recipes for the rules an opportunity is asking you to write.
• Build custom reports — the same data through the analytics API.