The Clock Has Started: Turning CRA Reporting into a Product Operating System

The Clock Has Started: Turning CRA Reporting into a Product Operating System

Since September 11, 2026, manufacturers of products with digital elements that fall within the Cyber Resilience Act cannot treat exploited-vulnerability and severe-incident reporting as a future compliance project. The European clock is already running: an early warning within 24 hours and a fuller notification within 72 hours. The real challenge is not completing a form. It is building an operating system that connects security, product, engineering, support and legal teams before an event begins.

1. Start with scope—not with every SaaS product

The CRA covers hardware and software made available on the EU market as products with digital elements. It can also cover a remote data-processing solution developed by or under the responsibility of the manufacturer when the product cannot perform one of its functions without that solution. SaaS is not automatically in scope merely because it is digital. The first operational task is therefore a documented scope map: which products are placed on the market, who is the manufacturer, which components are supplied separately, which back-end services are essential, and whether the organisation also acts as importer or distributor. This is not a marketing label and should be validated by qualified legal counsel. The useful output is a table connecting each product and version to its responsible entity, markets, distribution model and accountable contact. Without it, an incident team can lose the first hours debating whether any duty exists. This article offers a product and operating framework, not legal advice.

2. Turn awareness into a system event

The reporting clock starts when the manufacturer becomes aware of an actively exploited vulnerability or a severe incident affecting the security of the product. That wording makes organisational awareness a design problem. A researcher report, supplier advisory, telemetry anomaly and support ticket are not identical, yet they cannot remain isolated in separate inboxes. Create one intake path that records an immutable timestamp, source, suspected product, confidence and owner. Then define decision authority: who can determine that exploitation is active, who classifies severity, and who can start the regulatory workflow. Distinguish technical detection time, escalation time and decision time so the process does not reward delayed acknowledgement. A mature team does not wait for perfect certainty before preserving the clock. It records what is known, what remains unknown and why the current classification is reasonable. That creates both speed and an auditable basis for revision.

3. Design a state machine, not an email chain

An event should move through visible states: signal, triage, credible suspicion, potentially reportable event, 24-hour early warning, 72-hour notification, remediation, final report and closure. Every transition needs a condition, owner, timestamp and evidence. That lets the technical investigation and regulatory path run in parallel instead of waiting for each other. A state machine also prevents two common failures: closing too early because an issue first looked small, and reporting twice because separate teams opened incidents with the same root cause. Link the internal event identifier to the reporting-platform identifier, affected product versions, code changes and user communications. When new information arrives, update the state without rewriting history. Reconstructability matters as much as speed: a month later, the organisation should be able to explain who knew what, at what time, why a decision was made and what changed afterward.

4. Pre-build the 24-hour packet

An early warning is not expected to be a completed forensic investigation. Its purpose is to communicate the information already available quickly, including initial indications of affected markets or infrastructure where relevant. Build a template that can pull product identifiers, versions, legal entity, contacts, awareness time and a concise description from authoritative systems. Give every field an explicit status: verified, estimated, unknown or not applicable. Stating uncertainty is safer than presenting false precision that must later be corrected. The approval path must be short and include delegates for nights, weekends and holidays; a completed packet is useless if only one person can release it. Quarterly exercises should measure time to open the event, identify the product, reach a classification and submit the warning. Measuring only writing time hides the real bottlenecks.

5. The 72-hour notification joins evidence, impact and action

The fuller notification adds information about the nature of the event, an initial assessment and severity, and mitigation or corrective measures already taken. This is where organisational architecture becomes visible. Security holds indicators and attack paths; engineering understands components and versions; product knows functions and user groups; support sees external symptoms; and legal teams evaluate parallel duties. Replace the unstructured shared document with an evidence register that records source, time, confidence and owner. The regulatory narrative can be generated from that register, while sensitive exploit details remain access-controlled instead of being copied into every system. Separate technical facts, classification decisions and external wording. The objective is not to make the event look tidy. It is to produce a consistent, updateable account that distinguishes observed fact, inference and hypothesis.

6. Version inventory is a reporting prerequisite

An organisation cannot describe impact if it does not know which versions customers are running. A product catalogue should connect each SKU or package to software and firmware versions, third-party components, support period, update channel, distribution regions and remote-processing dependencies. An SBOM helps identify a component, but it does not replace deployment knowledge. Teams also need adoption data, active-device information, configuration context and a privacy-respecting way to reach the accountable product owner. When a library is vulnerable, the product question is not simply whether it appears in the build. The team must know which versions include it, whether the vulnerable path is reachable, which customers are exposed and how an update will reach them. The same inventory improves quality, support and deprecation work. Compliance investment therefore becomes product-lifecycle capability.

7. User communication is part of incident response

Where the CRA requires affected users to be informed, the communication must help them understand the event and apply available corrective or mitigating measures. A useful notice explains what happened, which versions are affected, what the user should do, what the manufacturer has already done and where verified updates will appear. Generic legal language is not enough. Product teams should pre-design the channels: in-product notification, email to the accountable administrator, status page, API response or device-management message. Keep those channels consistent so support does not tell a different story from the public notice. When remediation requires a firmware update, permission change or key rotation, the update experience itself is a security control. Measure more than messages sent: track the remaining vulnerable population and the percentage of users who complete the required action.

8. Reliable reporting begins in supplier contracts

Digital products depend on libraries, cloud services, connectivity modules, chipmakers and security providers. If a supplier waits two days to disclose a vulnerability, the manufacturer may have lost most of the 72-hour window before its own investigation begins. Contracts and procurement workflows should define security contacts, notification timing, minimum information, evidence preservation, version sharing and remediation support. Not every supplier needs the same controls; classify them by criticality and dependence to a product function. Free and open-source software also requires careful distinctions among individual developers, stewards and commercial activity. Reporting obligations for open-source software stewards begin later, on December 11, 2027, but product dependency exists today. A responsible manufacturer needs its own way to receive, validate and operationalise advisories rather than expecting a community to manage customer impact on its behalf.

9. Rehearse the clock, not only the attack

Many security exercises focus on detection and containment, then stop before a regulatory decision. A CRA exercise should continue through a mock submission, user communication and corrective update. Inject incomplete information intentionally: an unknown version, an unavailable supplier, a customer in an additional Member State or an indicator that later proves false. This tests whether the process can work with uncertainty. Run at least one exercise outside normal business hours and confirm that delegates can act. The retrospective should improve systems, not merely the runbook: add a missing catalogue field, fix an access control, introduce an SLA alert or create a build check. Useful measures include time to documented awareness, classification time, completeness of the 24-hour packet, consistency between reporting and user communication, time to remediation and the percentage of customers that complete an update.

10. Use 2026 to build the 2027 operating model

Reporting duties are already active, while the CRA's broader main obligations apply from December 11, 2027. This is a build window, not a waiting period. In the first 30 days, create the scope map, authority model, intake channel and reporting templates. Within 90 days, connect the version catalogue, evidence register, user-communication path and first full exercise. Then embed the same capabilities in product development: threat modelling, SBOMs, coordinated-vulnerability disclosure, secure update delivery, support periods and measurable security outcomes. ENISA's Single Reporting Platform is already live and was upgraded to version 1.3 on November 4, 2026, a reminder that the compliance interface is itself a changing system that needs an owner and tests. Dor Arad's product perspective is to treat this intersection as an operating decision: sound regulation is not a layer added after launch, but a capability that helps an organisation respond quickly, communicate truthfully and preserve trust under pressure.

Back to all articles →