The problem is the "bias" runs in both directions.
If you pass a bunch of laws after 9/11 and then another 9/11 does not happen, does that mean you actually need all of those laws, or that it's all overreach and bear-repelling rocks and all you really needed was to tell the passengers on planes to resist hijackers?
In other words, it's not bias, it's uncertainty.
Interesting problem.
Just thinking about this particular issue, maybe some options are:
1) counting time and then expressing success as the longest period of time during which the bad events didn't happen,
2) letting it happen initially to have a base "bad rate", then taking action to bring that rate down, and continuously make an effort to keep the occurring rate down
People preventing accidents at industrial sites have good experience doing this.
Consider gun related crimes in Australia before gun buy backs ("bad rate") and gun related crimes in post. You would have a solid number of averted deaths and injuries.
Now let's wait for thoughtful opinions from US users regarding the topic of gun regulation :)
Because overwhelmingly it does each and every day.
[edit] This bias comes down to the same issue as folks in infrastructure roles have in the tech world demonstrating their impact. How do you demonstrate impact avoiding SEVs when, if you're successful, the SEV never happened? It's very hard to count the number of things that don't happen.