Findings are facts OmniCommand established from your data: wasted spend, anomalies, trend breaks, fatigue, gaps, and more. Every finding carries its evidence, and many carry a severity based on the money or outcomes at stake. When a finding has no measurable stake, it carries no severity badge, honestly, rather than an invented one.
Recommendations are the actions those findings add up to. Each one states:
- the suggested action, in one sentence;
- the estimated money at stake;
- the findings behind it, with their evidence;
- the why, one click away.
The workflow. Recommendations move through bands your whole team shares:
- Needs triage: new recommendations, ranked with the most at stake first. Start what you accept, dismiss what does not apply (your reason is kept), snooze what can wait.
- In flight: what the team is working on. Add progress notes, and watch the review date OmniCommand sets from the recommendation's own measurement window.
- Completed: what you finished, with before-and-after measurement.
- Snoozed, Dismissed, Retired: sleeping items with wake dates, declined items with their reasons, and items the system retired because the data no longer supports them.
Before-and-after measurement. When you complete a recommendation, OmniCommand measures the metric it named over equal windows before and after the completion date. While the window is still filling, it says "Measuring until" the date, with no premature numbers. The result is shown with the count of other account changes recorded in the same period, and it is labeled "not attributed," because honest measurement does not claim causation.
When OmniCommand says nothing. No data is never dressed up as data. If a comparison lacks enough volume, or a finding cannot name what it would measure, the screen says so plainly.