DEV Community

yutianle
yutianle

Posted on

Putting the phishing ecosystem numbers in context

Putting the phishing ecosystem numbers in context

Large security stories often arrive with several impressive numbers. The Microsoft disruption report says the phishing ecosystem facilitated more than one billion emails and SMS messages, targeted more than 75,000 organizations, and reached more than 150 countries. It also cites thousands of templates and roughly 17,000 unique domains. Each number answers a different question.

The figures can help leaders understand scale, but they can mislead when combined without units. A message is not a victim. An organization that received a message is not necessarily an organization that clicked. A domain is not a campaign, and a template is not a successful compromise. The analysis should keep the metric, unit, source, and time scope together.

The Microsoft and Health-ISAC reporting provides a useful starting point for a measurement table. Microsoft describes campaign volume and organization reach. Resecurity describes 911 published builds and 2,360 phishing templates. Proofpoint describes over 100 million messages per month blocked for organizations using CoGUI in its visibility. These are different collection systems with different coverage.

Separate the denominators

The one-billion-message figure is a reported ecosystem volume. The 75,000-organization figure is a reported reach measure. The 17,000-domain figure is a count of domains used by the kits. None of them is a confirmed victim count.

Public scan data adds another denominator. ZoomEye returned 163,176 observations for app="Microsoft SharePoint" at 2026-11-16T14:49:42Z. That is a product population, not a list of compromised tenants. A report can use it to describe external visibility and to explain why a product-level count cannot answer an organization-level question.

The separation is not pedantry. It changes the response. A large message count supports broad awareness and email-control review. A large product count supports asset inventory. A confirmed tenant event supports session revocation, permission review, and data assessment. Each response needs a different evidence base.

Explain collection bias

Security vendors see only the telemetry they collect. Proofpoint's message volume reflects its customer visibility. Microsoft's organization count reflects its investigation and partner data. A public scanner sees internet-visible services and may miss private tenants, geofenced pages, and blocked requests. The numbers are useful because they are large, but they are not a random sample of the internet.

The analyst should also note when a source measures a trend rather than a population. A monthly block count can include many attempts against the same organization. A domain list can include domains that were never used in a successful campaign. A template catalog can include lure pages that buyers never deployed.

Set the right confidence language

Use "reported" for figures that come from a vendor or investigation. Use "observed" for a collection with a timestamp. Use "confirmed" only when the evidence identifies the event and the entity. Use "affected" only when the source supports that relationship.

This vocabulary is more accurate than a single risk score. It also helps a reader compare reports without assuming that every large number belongs to the same denominator. A security leader can ask which number supports the decision at hand.

A practical dashboard

A useful dashboard might show:

  • reported ecosystem message volume, with source and period;
  • reported organization reach, with source and definition;
  • documented domains and hosting roles, with observation dates;
  • public product observations, with exact query and scope;
  • confirmed internal user clicks, sign-ins, and data access.

The first four rows help with situational awareness. The last row drives local response. A dashboard that mixes them without labels may look complete while hiding the most important gap.

ZoomEye fits into the public observation row. It offers repeatable queries and a collection timestamp. It does not replace the private evidence needed for containment. That is the right division of labor for a large phishing ecosystem and for the organizations that need to decide what to do next.

References

Top comments (0)