When Optimize surfaces a recommendation, it also tells you how sure it is that the recommendation is accurate. This article explains what that "confidence" means and how to use it.
What is a confidence level?
Every recommendation Optimize surfaces answers two separate questions:
- How much does this matter? (called severity) — based on how much time is being lost, how many people are affected, and whether it creates risk or blocks important work.
- How sure are we this is real? (called confidence) — based on how much evidence Optimize has for the pattern.
This article is about #2: confidence.
How Optimize decides its confidence
Confidence is based entirely on what Optimize observed in your team's actual work, not a guess. Optimize looks at:
- How many times the behavior showed up across recordings
- How many different people did it the same way
- How clear the underlying activity was (a clean, obvious pattern vs. a fuzzy, one-off event)
A recommendation earns high confidence when Optimize saw the same behavior repeatedly, across multiple people or sessions — not just once. A recommendation gets lower confidence when the evidence is thinner (e.g., only observed a couple of times, or only from one person).
Why this matters to you
Confidence helps you decide where to focus first:
- High severity + high confidence = the findings we're most sure are real and worth fixing. Start here.
- High severity + lower confidence = could be a big win, but worth a quick gut-check with your team before acting.
- Lower severity findings are usually fine to skip unless they're quick, easy fixes.
What to do if a recommendation seems off
Optimize's recommendations are based on real activity, but they're not perfect — for example, it may occasionally miss that a tool is already integrated a different way than expected. If a recommendation doesn't match reality:
- Open the recommendation and select Hide issue.
- Choose a reason (e.g., "Not relevant to our team," "Known constraint," or "Already addressed") and add a short note.
- Your feedback helps Optimize avoid surfacing the same inaccurate recommendation again.
Bottom line
Confidence tells you how much evidence backs a recommendation — not whether it's important. Use severity to see what matters most, and confidence to see how sure Optimize is about it. The findings with both high severity and high confidence are the safest bets to act on first.