Every company I've worked for either didn't know how changes affected things like uptime or …

In the cases of uptime and conversion rate, might you only implement the change after looking at the desired metrics and verifying that they've improved? re GP comment: it's more about the tone -- one should seem confident and well-acquainted with what they choose to show -- than the actual numbers. If you told me that you improved conversion rates by 2%, or 20%, I would barely know the difference, but I would see both of those very differently than just "improved conversion rates". If you don't have numbers, I would try to be specific in some other area instead (e.g. technologies used, names of big clients). Similarly, phrases like "had creative input across the full stack" might give me pause -- what does that mean? It implies a low amount of impact; why not say something more attractive, like "contributed and assisted others across the full stack"?

> where exactly are people getting these stats? You can basically make them up - just make up something plausible for your field. It's not like everybody else isn't rounding up anyway.

Most are making up these stats. Don't say 100% improvement, though, unless you can back it up but 30% with a process improvement should be easy to justify in an interview. The point here is to make your resume "skimmable" since no one reads text anymore.

The product management team, or whoever was pushing for the project in the first place, should have this data, and will often be very happy to share. They might not themselves know 100%, but they should have some estimate of benefits. It's not always widely shared, but you could certainly ask.

The modern CV art is purely about how confident and fluent you are in answering anything on it. The actual roles and projects may be anything; you just base it as a biopic adaptation based on your life.