Finding vulnerabilities is the part of security that gets the attention. Getting them fixed, and keeping them fixed, is where most programs quietly lose.
Raj Handa, Senior Cybersecurity Engineer in offensive security at Workday, leads its pentest program and runs its bug bounty. Early in our Scale To Zero conversation, he described the one outcome his team works hardest to avoid:
“Paying to find a vulnerability once only to let it reoccur in the next release.”
Later, Scott Weston sent in a question that let us dig into exactly that: when you assess internal and external cloud assets at scale, how do you prioritize? Do you rely purely on CVSS or EPSS?
Two Numbers That Cut Through the Noise
Raj didn’t argue against severity scores. He pointed to two metrics that tell you whether your program is actually getting healthier.
1. Vulnerability Recurrence Rate
His example: an external vendor finds an IDOR or broken access control flaw in your API. Six months later, your internal team finds the exact same class of flaw in the same service.
That is a signal to look deeper. Point it out to the engineering team, highlight the recurrence and the risk, and push for a fix at the root rather than at the symptom.
“You should never pay or be able to discover the same architectural mistake twice.”
2. Shift in Severity Complexity
In a healthy program, the nature of findings should evolve. In year one you might see basic configuration issues or missing headers. By year two or three, Raj said, your automated pipeline should catch that kind of noise, and human testers should be finding business logic bypasses, complex tenant boundary hops, or tool-use escalation in AI agents. If they aren’t, your testing needs a second look.
If you want to pair these with exploit signals, our post on prioritizing with CISA KEV and EPSS covers that side, and which vulnerabilities are actually running in production covers reachability. Recurrence and complexity tell you something different: whether your fixes are sticking.
Why Findings Rot in the Backlog
Before getting to fixes, it helps to know where the time goes. Raj sees lag at two points.
The ticket that just sits. The typical habit is to log a Jira ticket and consider the job done. Raj’s view is that security is a partnership: make sure the ticket is assigned to an individual, follow up, highlight the SLA for that severity and the risk. If there is still no traction, escalate to their manager, then the VP.
The roadmap that always wins. Once a team acknowledges a finding, they work at their own pace against an already defined product roadmap. Raj’s fix is to get the remediation into the very next sprint, or the one after, depending on the work, so it isn’t “lying there in their sprint forever.”
Building Structural Friction Against Inaction
“You cannot just rely on the good intentions of your engineers. You have to build a structural friction against inaction.”
Ship Tickets, Not Reports
Raj’s team doesn’t deliver generic reports. They deliver engineering tickets with reproducible code.
“Don’t just say sanitize user input, otherwise developers are going to ignore it. Provide the exact curl command.”
He added that with AI, it is now easy to pinpoint the exact code line where the exploit lands. Include the exploit script and a threat-model-based remediation recommendation, so engineers know how to fix it the first time. Raj said that in his experience, providing that deeper context can eliminate almost 80% of findings.
Tie SLAs to Release Gates, With Executive Ownership
Raj’s second structural control is a hard line:
“If an SLA is breached, you should automatically be able to block the next release train.”
Highs and criticals get non-negotiable SLAs. If open highs or criticals have breached their SLAs, it’s a no-go from security. The important part is alignment: agree this approach with your VP and CISO first.
“The moment an executive puts down their name on a security risk, I’ve seen the backlog tickets get prioritized overnight.”
Turn Every Finding Into a Regression Test
On the bug bounty side, Raj tracks what he calls bounty regression rate: did a bug you already paid for quietly slip back into the codebase in a later sprint? If so, the program is “operating like an open drain.”
“A mature program uses every valid payout as a mandate to build a permanent pipeline integration test or some sort of a rule.”
He also measures the velocity “from validated finding to deployed regression test,” which he sees as the test of whether engineering treats security as a partner or just a checklist. For where those tests can live, see our GitLab CI/CD security checklist.
Never Pay Twice Checklist
-
Track recurrence by vulnerability class and service, not just by individual finding.
-
Treat a repeat class in the same service as a root-cause issue, not a new ticket.
-
Compare findings year over year: basic issues should move to automation.
-
Every ticket includes repro steps, the exact command, and a remediation recommendation.
-
Every ticket has a named owner and a target sprint.
-
Highs and criticals have non-negotiable SLAs tied to release gates, agreed with the VP and CISO.
-
Every valid finding becomes a pipeline test or rule before it is closed.
-
Recurrence and regression rates are reported alongside mean time to remediate (my addition: put them on the same dashboard leadership already sees).
Context Builds the Relationship
One thing I took away from Raj’s answer: the extra context in each ticket isn’t only about speed. It changes the relationship. When engineers learn that security always shows up with a repro and a fix path, they prioritize without needing an escalation to a VP.
Listen to the Full Episode
Thanks to Scott Weston for the question. Raj also covers extension confusion in AI editors, bug bounty in the age of AI-generated reports, and how to split pentesting across automation, internal teams, and vendors. See also AI Slop in Bug Bounty for the metrics Raj uses on the bounty side.
-
Watch on YouTube: Pentesting, Bug Bounties & The IDE Supply Chain Blind Spot ft. Raj Handa
-
Listen on Apple Podcasts or Spotify
-
Join the conversation on LinkedIn
Scale To Zero is the security podcast we run at Cloudanix. If your team has a trick for keeping fixed bugs fixed, I would love to hear it.
People Also Read
- Extension Confusion: The IDE Supply Chain Risk Hiding in Your AI Editor
- AI Slop in Bug Bounty: How to Keep Your Program From Becoming a Spam Folder
- How Often Should You Pentest? Treat Penetration Testing as a Portfolio
- Bug Bounty and Pentesting Through M&A: How to Bring an Acquisition Into Scope
- 52 PostgreSQL Fixes, One UnrealIRCd Flaw Under Attack: How to Decide What to Patch First With KEV and EPSS