Check the evidence behind an insight

Before you begin

You need View FleetHQ. Use the finding you opened in Read your fleet's health on FleetHQ. This scenario reads the analysis and records nothing, so you can work through it as often as you like.

What you will accomplish

You'll open a finding's analysis and separate what FleetHQ measured from the conclusion it drew, so you can judge how much weight the recommendation deserves.

Steps

  1. In Insights and recommendations, select your finding.

  2. Select View analysis. Choose a finding that offers it — a finding marked Not researched has no analysis to open yet.

  3. Read Summary, the customer-facing description of what FleetHQ found.

  4. Read the finding's classification. Performance concerns how the fleet is doing, Data reliability concerns whether the metric can be trusted enough to decide on, Good news marks favourable movement, and Opportunity marks something not measured yet.

  5. Read Root cause. This is FleetHQ's supported explanation — a step beyond the measurement, not the measurement itself.

  6. Read Evidence. Each item is labelled by what it is: a recent trend, a metric definition, fleet activity, an earlier finding, or a note.

  7. Read Ruled out to see which alternative explanations FleetHQ tested and rejected.

  8. Note the confidence level — Low, Medium, or High.

  9. Close the analysis and read the recommended steps in the plan beside the finding.

  10. Select View metric to check the underlying scorecard tile and chart for yourself.

What you just did

You looked at the seam in a FleetHQ finding, which is the thing worth knowing about this page. Evidence and root cause are not the same kind of claim. The evidence is what FleetHQ measured; the root cause is what it concluded from that, and the confidence level tells you how firmly. Ruled out is the part people skip and shouldn't — it shows you the alternatives FleetHQ considered, which tells you whether the explanation it settled on was actually tested against anything. Reading in that order, measurement first and conclusion second, is what lets you disagree with a recommendation for a good reason rather than a hunch. And when a finding says Deeper dive needed, read what it found as a lead rather than a conclusion; when it says Analysis unavailable, don't read a root cause into the detector summary that's left.

Next steps

See also

Search