Synthesize
See the pattern.
Question the explanation.
Turn scattered signals into findings you can inspect. Follow the evidence, investigate what might explain a change, and keep the unanswered questions in view.
A finding you can question.
Illustrative data · Expand the citations to inspect them
FINDING / CHECKOUT
Completion is down.
Price visibility needs a closer look.
The funnel shows a decline [1]. Customer feedback raises a question about when the full price appears [2]. The relationship needs investigation.
Inspect the evidence
[1] PostHog · Checkout funnel
Completion fell from 64% to 52%. Previous week: 320 of 500 sessions; current week: 260 of 500. Both use the same funnel definition.
[2] Intercom · Customer conversation
“I got to the payment step before I could see the full price.” One customer describes going back to check their plan. This account does not represent every abandoned checkout.
What can we conclude?
Observed
Checkout completion declined in the measured window [1].
Possible explanation
The total appears too late for some customers [2]. Whether this explains the decline is inconclusive.
Still unanswered
How many customers experienced this? Did anything else change in checkout?
A focused next question
What does customer feedback tell us about price visibility at checkout?
Findings with a foundation
Understand what matters.
See what backs it up.
Findings bring related evidence together so you can examine a change in context. The supporting sources stay within reach.
Follow the citations.
Inspect the records behind a finding. A meaningful issue can emerge from one source; additional sources can add context without proving a shared cause.
Read confidence alongside impact.
Confidence reflects how well the evidence supports the finding. Severity describes the importance of the issue. A serious problem can still need more investigation.
Keep the question open.
An observation can be clear while its explanation remains uncertain. Use that uncertainty to decide what to investigate next.
Focused investigations
Go from “what changed?”
to “what could explain it?”
Give the investigation a question worth answering. It tests explanations against available evidence and tells you where that evidence falls short.
- 01
Start with a precise question.
Choose the change you want to understand. Open an investigation, review the context, and send your question to begin.
- 02
Test possible explanations.
Investigations search project evidence to examine competing hypotheses. See which are supported, refuted, or still inconclusive, with citations to inspect.
- 03
Keep the answer and its limits.
Read the answer alongside its confidence and supporting evidence. Ask to save the investigation when you want to return to it or use it to develop a brief.
Ongoing monitoring
Keep watching
what matters.
Scheduled monitoring checks configured metrics against rules and expected ranges. Review a warning’s source, time window, and actual versus expected result before deciding what to do.
Checks run daily. Pause monitoring with the project’s Watcher control, or start a focused investigation when a warning needs a closer look.
Know what the status means
- A warning to review
- A configured rule has been met. Inspect the evidence and context; the warning itself does not explain the cause.
- Stale or insufficient data
- There is not enough fresh evidence to assess the metric. Missing data is shown separately from a healthy result.
- Acknowledged, still visible
- Acknowledging a warning records that it has been seen. It does not mean the underlying risk has resolved.
Next: Act
Give the finding
a useful next step.
Bring the evidence and open questions into a brief your team can review. Keep the reasoning attached as you decide what happens next.