Managed bug bounty: we build the programme and triage every report

Two jobs almost nobody wants in house: designing a reward programme that attracts good researchers, and reading everything that comes through it. We do both, on HackerOne, Bugcrowd, Intigriti or your own channel. And if the programme is already running and it is only the inbox you cannot keep up with, we do just the triage.

Last updated:

  • CVSS 4.0 + EPSS — Every valid report scored metric by metric, with the real likelihood of exploitation
  • 24 h / 4 h — Acknowledgment to the researcher within 24 business hours; critical escalation within 4
  • 5-10 days — Service onboarding: technical briefing, access, integrations and agreed workflow
  • €0 — What a duplicate or an out-of-scope report costs your team: we close them ourselves

Programmes don't fail for lack of reports. They fail at the inbox

A well-designed bug bounty attracts reports. That is exactly the problem nobody mentions before you open one: most of what arrives is duplicates, out-of-scope submissions and scanner output with no demonstrated impact — and all of it has to be read anyway to find the three findings that matter. At 2-4 hours per report and a normal volume of fifty a month, that is over a hundred hours a month of expensive people doing sorting.

When that sorting backs up, the damage runs two ways. Inwards, your development team stops trusting what reaches it and real findings get treated as noise. Outwards, the researcher who waited three weeks for a reply does not come back, and says so: a programme's reputation is built in the forums where those people read each other. Meanwhile, the critical vulnerability that arrived on a Friday in August is still in the queue.

That is why the programme and the triage are one service, not two. We design the rules and the reward table with the volume we will have to absorb in mind, and we absorb that volume ourselves: your team only sees what has been reproduced, scored and is ready to fix.

What it covers, from programme design to a closed report

  • Programme design: scope, rules of engagement, accepted vulnerability types and disclosure policy
  • Reward table by severity, agreed and published before opening: without one, every bounty is a negotiation
  • Platform setup (HackerOne, Bugcrowd, Intigriti) or your own intake channel
  • Private or public launch, with the invitation strategy and the initial communications
  • Classification of every incoming report: in or out of scope, and vulnerability type
  • Technical reproduction with real offensive judgment, not a reading of the report text
  • Impact validation: what an attacker can actually do with the flaw, and how far it reaches
  • CVSS 4.0 scoring justified metric by metric, with EPSS for exploitation context
  • Reasoned dismissal of false positives, duplicates and out-of-scope reports, closed with the researcher
  • Researcher communication on your behalf, in English and in Spanish
  • Handoff into the tools your team already uses: Jira, GitHub Issues or the platform itself
  • Standalone triage on demand: isolated reports, audit findings or scanner output that need confirming before escalation
  • Monthly report: volume, validation rate, response times and recommendations on the programme

How we work

  1. Programme design: We define scope, rules, accepted vulnerability types, disclosure policy and reward table with you. A well-defined programme gets better reports and less noise.
  2. Launch: We set up the channel — platform or your own form — prepare the communications and coordinate a private or public start according to your strategy.
  3. Triage of every report: Classification, reproduction in a controlled environment, impact validation, CVSS 4.0 and EPSS, researcher response and handoff to the remediation team. This is the bulk of the service and what absorbs the volume.
  4. Supported remediation: Validated findings arrive with PoC, impact and concrete steps. We answer your developers' questions and verify the fix before closing the case with the researcher.
  5. Monthly optimization: A report with metrics and recommendations: widen the scope, adjust bounties or change rules. A programme nobody retunes stops attracting the good researchers within six months.

From design to a closed report

The four phases, with the owner of each step and the committed deadline where there is one. The third — triage — is where the real work goes, and the reason this is a single service.

We design

The rules decide what you get afterwards. A vague scope is paid for in noise for months.

  1. Scope and rules (Your team · Week 1): Which domains, applications and versions are in, what stays out, and which tests are never allowed: denial of service, social engineering against staff, third-party data.
  2. Accepted types and exclusions (QuantumSec · Week 1): Which classes of vulnerability are rewarded and which are closed on sight. This is where half of the future disputes with researchers are avoided.
  3. Reward table (QuantumSec · Week 2): Ranges by severity, agreed and published before opening. Negotiating each bounty case by case stretches response times and eventually drives the good researchers away.
  4. Disclosure policy (QuantumSec · Week 2): What the researcher may publish and when, with a safe harbor clause reviewed by your legal team.

By the end of this phase: The programme defined in writing: scope, rules, exclusions, reward table and disclosure policy.

We launch

Opening to the public on day one is the most expensive mistake. You start by inviting.

  1. Platform or own channel (QuantumSec · Week 3): We configure HackerOne, Bugcrowd or Intigriti, or build your own channel, and wire it into our triage flow.
  2. Integration with your tools (QuantumSec · Week 3): Jira, GitHub Issues or whatever you already use, so validated findings land where your team already works.
  3. Private start (QuantumSec): We open with a limited group of invited researchers. It measures the real volume and your remediation capacity before you are exposed.
  4. Phased public opening (Your team): The programme widens when remediation times can keep up, not because the calendar says so.

By the end of this phase: A live programme with controlled volume and findings already landing in your tools.

We triage

We read what comes in. Your team only sees what has been reproduced and scored.

  1. The report arrives (Researcher): Through the platform or your own channel. It lands in our queue, not in a shared mailbox nobody watches.
  2. Initial classification (QuantumSec · 24 business hours): In or out of scope, and vulnerability type: IDOR, XSS, SQLi, business logic. With an acknowledgment to the researcher and a case number.
  3. Technical reproduction (QuantumSec · 72 business hours): We attempt the attack in a controlled environment: that the payload really works, that the impact exists and that it is not a duplicate of something already known.
  4. Impact and scoring (QuantumSec): What a real attacker gets — data, escalation, money, reputation — with CVSS 4.0 justified metric by metric and EPSS to place the likelihood of exploitation.
  5. Critical escalation (QuantumSec · 4 business hours): If the impact could be critical we alert your security lead through the agreed emergency route, whatever the time.
  6. Response and closure (QuantumSec): We write the reply on your behalf — valid, duplicate, out of scope — with its technical reasoning. A badly answered researcher stops reporting, and tells others.

By the end of this phase: An empty inbox and, in your tools, only reproduced, scored, actionable findings.

We tune it

A programme nobody retunes stops attracting the good researchers within six months.

  1. Monthly report (QuantumSec · Monthly): Volume, validation rate, duplicates, time to first response and to validation, and the status of every open case.
  2. Scope review (QuantumSec · Quarterly): Whatever you have shipped since the last review goes into the published scope, so the programme covers what is really exposed.
  3. Reward and rule tuning (QuantumSec · Quarterly): Using the programme's own data: where you underpay, which class of report is surplus and which rule keeps causing arguments.
  4. Widen or hold (Your team): If remediation cannot keep up with triage, the bottleneck is not the inbox: the programme widens in phases instead of opening all at once.

By the end of this phase: A programme tuned to what you actually receive and to how fast you can fix it.

What you get

  • A programme designed, configured and live in 2-3 weeks
  • A reward table by severity, agreed and published
  • A processed inbox: every report with its status — valid, duplicate, out of scope, false positive — and its technical reasoning
  • Every valid finding with PoC, impact, CVSS 4.0 justified metric by metric and an EPSS score
  • Validated findings in the tool your team already works in
  • A complete record of researcher communications, case by case
  • A monthly report with metrics, trends and recommendations on the programme
  • A metrics dashboard available continuously

When it makes sense

  • Companies launching their first reward programme with nobody to hand the inbox to
  • Programmes already live on HackerOne, Bugcrowd or Intigriti where the in-house team cannot keep up with triage
  • AppSec teams with a backlog of unprocessed reports and no way to catch up
  • Open but stalled programmes, because researchers lost faith in the response times
  • Startups that need a programme to meet investor or enterprise client requirements
  • CISOs who need technical certainty about a one-off finding before escalating it to the board
  • Product teams deciding which vulnerability blocks a release and which does not

Bug bounty, VDP and pentesting: what each one covers

They are bought separately and solve different things. The expensive mistake is hiring one expecting what another gives.

Managed bug bountyManaged VDPPentesting
What it isA programme paying per valid findingA disclosure channel with no rewardsAn assessment bounded in scope and time
Who looksA wide community of researchersWhoever stumbles on the flawA contracted team
Guaranteed coverageNo: nobody assures you anyone looks at your ugly moduleNoYes, whatever the scope agreed
Volume and noiseHighLow and steadyNone: just the report
CostManagement + variable rewardsManagement onlyFixed price per project
Works as evidence for a client or auditorPartiallyYes, for NIS2 and the CRAYes, that is its natural format
When it fitsContinuous watch with budget to rewardFirst formal channel and compliance baselineDemonstrable coverage and a deliverable

Compare with

Frequently asked questions about bug bounty and triage

Can you do just the triage for a programme we already run?

Yes, that is half of our engagements. We plug into your HackerOne, Bugcrowd, Intigriti or in-house channel without changing your platform or your process: we add the validation and response layer you are missing. We also triage one-off reports arriving through other routes — an email to support, an audit finding, scanner output — when you need technical certainty before escalating them.

How is this different from hiring HackerOne or Bugcrowd?

HackerOne and Bugcrowd are platforms: they give you access to a community of researchers and management tools, but your team still does the triage. We are the external team that does that work. We can operate on the platform you already use, or design a programme independent of them.

What is the SLA for a critical report?

Four business hours from the moment triage identifies potentially critical impact: we alert your security lead through the agreed emergency route and coordinate the response from there. Acknowledgment to the researcher is 24 business hours and the first technical assessment 72. The deadlines are agreed in writing at the start.

Why CVSS 4.0 and EPSS, and not the severity the researcher set?

Because the researcher's severity measures what they achieved, not what it is worth in your context. CVSS 4.0 forces a metric-by-metric justification — vector, complexity, privileges, impact — and makes one finding comparable to another. EPSS adds what CVSS does not say: the estimated likelihood that something like it is actually exploited in the next thirty days. A CVSS 9.8 with low EPSS and a CVSS 7.5 with high EPSS are not prioritized the same way.

What happens with duplicates and out-of-scope reports?

We close them ourselves with a reasoned reply, without them ever reaching your team. A duplicate is linked to the original report and an out-of-scope submission is explained by citing the programme's published rules. It is the most thankless part of running a bug bounty and the one that breaks most relationships when handled with curt templates.

How much reward budget do we need to start?

It depends on scope and on the severity you accept, but the usual mistake is not under-budgeting: it is opening the programme with no published reward table. Without one, every report becomes a negotiation, response times stretch and good researchers stop submitting. Set the table before opening, publish it, and revise it after six months with your own programme's data.

What if we receive more reports than we can handle?

That is the normal scenario in the first weeks of a public programme, which is why starting private with a limited group is recommended. External triage absorbs the volume: it validates, discards duplicates and out-of-scope reports, and only confirmed findings with demonstrated impact reach your team. If the bottleneck is remediation rather than triage, the programme is widened in phases instead of opened all at once.

Does a bug bounty replace penetration testing?

No, they cover different things. A pentest is bounded in scope and time, with guaranteed coverage of what was agreed and a report that works in front of a client or an auditor. A bug bounty is continuous and offers no guaranteed coverage: nobody assures you anyone will look at your least attractive module. The sensible combination is pentesting for coverage and evidence, and bug bounty for ongoing watch between engagements.

What if we disagree with your assessment of a report?

The assessment is delivered justified metric by metric precisely so it can be argued. If your context changes the impact — that data is not sensitive, that endpoint is not really exposed — it is adjusted and the reasoning is recorded. The final call on accepting a risk or paying a bounty is always yours.

Can you validate vulnerabilities without production access?

In most cases yes, working on an equivalent pre-production environment or with a test account. When a flaw only reproduces in production we agree it in writing beforehand, with a window, a scope and testing limited to confirming impact without touching real data.

How do you guarantee the confidentiality of the reports?

A confidentiality agreement signed before we start, access limited to the assigned team, and reports and evidence held in the systems we agree on, not in personal tools. If we work on your platform, the data never leaves it.

What do researchers need to know about who runs the programme?

You decide. We can operate on your behalf without researchers knowing there is an external provider, or state that triage is handled by a specialized external security team. Both models are common in the market.

Before you open the programme

Inside the triage

Running it and measuring it

Sources

Talk to a specialist