On this pageWhat EASM CoversUnknown Assets and Dangling RecordsDiscovery From Seed DomainsA Worked Discovery ExampleEvidence By Discovery SourceWorking With Each SourceInside Records That Extend DiscoveryAttribution of Discovered AssetsMonitoring the Surface for ChangeFrom Inventory to ActionSmall, Mid-Sized and Large EstatesStarting and Measuring a ProgramTools, Products and Managed ServicesEASM With ThreatDefenceFrequently Asked Questions It works from the outside, from the same public records and network responses that an attacker can read. In most organizations the result is larger than the asset register, because projects, suppliers and acquired businesses publish systems as well as the IT team. This guide explains how discovery works from seed domains outward, what each data source reveals and how to pull it, how an asset is attributed to its owner, and how the inventory becomes monitoring and action. What EASM CoversEASM covers the systems that answer from the public internet and the records that describe them. The UK National Cyber Security Centre (NCSC), in its EASM buyer’s guide, describes it as the process of identifying, monitoring and reducing vulnerabilities in assets that are accessible from the internet. Domains and hostnames: EASM covers every root domain the organization has registered and every name beneath it. Addresses: It covers the public IP addresses the organization holds or rents, on premises and in cloud accounts. Services and applications: Each open port is recorded with the product behind it, from a VPN gateway to a web application or an API. Certificates and third-party platforms: TLS certificates issued for the organization’s names are included, along with hosted services published under them. An internal asset inventory starts from what the organization manages. EASM starts from what the internet can see and works back to an owner, which is why the lists rarely match on the first pass. The findings are wider than software vulnerabilities. The NCSC guide gives exposed services that should be unreachable from the internet, DNS misconfigurations and email security weaknesses as examples, and notes that products often call them issues or risks. Lookalike domains, impersonation and leaked credentials concern the organization’s name and people more than its own systems, and are covered under attack surface management and brand protection. Unknown Assets and Dangling RecordsUnknown assets come from ordinary work done outside the central change process, such as project systems left online, supplier-hosted sites, self-service cloud accounts and acquisitions. Shadow IT discovery is often the first result an EASM program delivers. Records left behind are the other half. A DNS record that still points at a deprovisioned cloud resource is a dangling DNS entry, in the term Microsoft’s documentation uses, and another party may be able to claim the released resource name and serve content under the organization’s subdomain. The NCSC buyer’s guide counts domains vulnerable to takeover, such as dangling CNAMEs, among the DNS security checks an EASM product may perform. Each discovery run therefore resolves every alias in the zone and matches the live targets against the organization’s cloud inventory. Our guide to subdomain takeover covers how a takeover works and how a dangling record is fixed. Discovery From Seed DomainsDiscovery starts from a few facts the organization confirms, called seeds, and expands outward from them. A seed is typically a root domain, a registered address range or a legal entity name. Confirm the seeds. The asset owner lists the root domains and address ranges the organization controls, including those of subsidiaries. Enumerate names. Each root domain is expanded into hostnames from DNS records, Certificate Transparency logs and name guessing. Resolve and locate. Each name is resolved to an address, and each address is placed with a network owner and a hosting provider. Pivot to related assets. Shared certificates, name servers and registration details point to further root domains, which become candidate seeds. Find live services. Open ports and products are read from internet scanner records and confirmed with a direct scan, using a tool such as nmap. Fingerprint each service. The product, version where shown, sign-in page and certificate are recorded. Attribute and record. Each asset is marked confirmed, probable or unrelated, with the evidence for the decision. The process is recursive. A candidate seed is attributed before it is expanded, since expanding another organization’s domain fills the inventory with its hosts. Passive work, which reads public records, can run before scope is agreed, while active work that sends requests to the systems waits until an asset is attributed. Passive discovery: Certificate log searches, registration queries and scanner index searches read records held by other parties, so the organization’s own systems see none of it. Active discovery: Port scans, HTTP requests and TLS handshakes go to the asset itself, appear in its logs and may raise alerts. NIST SP 800-115 notes that active discovery reveals its point of origin and can trigger intrusion detection alerts, and that the information from passive and active techniques alike is seldom completely accurate. A service reported by one source is therefore confirmed by a second source or a direct request before it is recorded as live. A Worked Discovery ExampleA single run shows how the steps connect. acme-corp confirms one root domain, acmecorp.com, and one address block registered in its name. Zone export. The DNS provider’s export lists the names the IT team manages, and shows that dev.acmecorp.com is delegated to name servers in a cloud account run by the development team. Certificate logs. A search for the root domain returns api.dev.acmecorp.com, which sits in the delegated zone, and one certificate that covers both http://www.acmecorp.com and shop.acme.com.au. Resolution. The API name resolves to an address in a cloud provider’s published range. Another name from the export, old-campaign.acmecorp.com, is an alias to a hosting platform hostname that no longer resolves. Pivot. The second root domain, acme.com.au, shares a registrar and name servers with the seed, and stays a candidate until the marketing team confirms that it registered the domain. Services. A scanner index shows a firewall administration sign-in page on one address in the registered block, and a direct scan confirms that the port still answers. Attribution. The API host is confirmed when the development team’s cloud inventory lists its address. The shop is recorded as managed by a third party, since a commerce platform runs it. Decisions. The firewall interface is restricted to the internal network, the campaign record is deleted, the API host gets an owner and enters the testing scope, and acme.com.au joins the seed list. Evidence By Discovery SourceEach source answers a different question, and attack surface discovery is reliable only when several are combined. Source What it reveals How to pull it DNS records Hostnames, addresses, aliases to third-party platforms and mail providers. Export the zone from the DNS provider. From outside, query A, AAAA, CNAME, MX, NS and TXT records and test candidate names from a wordlist. Certificate Transparency logs Names on publicly trusted certificates under the domain, including hosts that were never announced. Search a log search tool such as crt.sh for the root domain, or follow the logs and filter for the organization’s domains. Registration records The registrar, dates and name servers of a domain, and the holder of an address block. Run an RDAP query for the domain and query the regional internet registry for the address. RDAP has been the definitive source of registration data for generic top-level domains since January 28, 2025. Internet scanner records Open ports, service banners and certificates, as last seen by a public scanner. Search an internet scanner platform, such as Shodan, by address range, hostname or certificate name. Cloud provider ranges The provider, service and region behind an address. Match addresses against the range files published by AWS (ip-ranges.json) and Microsoft (Azure IP ranges and service tags). Web fingerprints What a host is: product, version, sign-in page, redirects and certificate details. Request each name over HTTP and HTTPS and record the status, headers, title, favicon hash and certificate. Working With Each SourceEvery source has a blind side, which matters as much as what the source reveals. DNS Records A zone export is the most complete list of names for a domain the organization manages. From outside, a zone cannot normally be listed, because name servers refuse zone transfers to unknown clients, so discovery depends on guessing names and on sources that leak them. CNAME records: An alias shows which third party serves the name, so each alias target is checked to confirm it still exists. MX and TXT records: Mail routing and verification tokens show which mail and software services the organization uses. Wildcard records: A wildcard answers for every name, so guessed names are compared with the answer for a random one. Certificate Transparency Logs Certificate Transparency is a protocol for publicly logging TLS server certificates as they are issued, specified in RFC 6962 and its successor, RFC 9162. Requesting a certificate for a new hostname therefore publishes the hostname, before any page links to it. Names only: A logged name may no longer resolve, so each one is checked before it enters the inventory. Wildcard certificates: A wildcard covers every host under a domain and lists none of them. Private authorities: Certificates from an internal certificate authority are absent from the public logs. Registration Records Registration records connect assets to a legal owner. Domain records: Registrant contact details are often withheld, so shared name servers and a shared registrar are the more dependable hints. Address registry records: A block registered to the organization is strong evidence of ownership. Internet Scanner Records Internet scanner platforms probe the public address space continuously and index what answers. CISA’s Internet Exposure Reduction Guidance lists Shodan among its examples, and notes that threat actors use such platforms to identify publicly accessible systems. The records are a view of the past, so a port may have closed since the last visit. Their value is that the defender sends nothing to the target. Cloud Provider Ranges Range files identify the provider behind an address and say nothing about which customer holds it. Microsoft publishes the Azure file weekly, and AWS publishes its current ranges as a JSON file. Account inventory: Ownership is confirmed from inside the account, where a read-only query through the provider’s API lists public addresses, load balancers and storage endpoints. Unmanaged accounts: Accounts opened outside central control are the usual gap. Web Fingerprints A web fingerprint turns a hostname into a description of what runs there. The redirect chain, page title, headers, favicon and certificate distinguish a VPN portal from a default installation page or a copy of the production site. A version read from a banner is a lead, and penetration testing and vulnerability scanning differ in how far each goes to confirm it. Inside Records That Extend DiscoveryOutside discovery finds what public sources reveal, and records held inside the organization supply the rest. The NCSC buyer’s guide lists input integrations, such as loading information from asset databases, among the supporting functions of an EASM product. Record Gap it closes Where to pull it Registrar domain list Root domains the organization registered that share nothing public with the seeds. The domain export in each registrar account, including accounts opened by marketing teams and agencies. Delegated zones Names under a subdomain that is served by another DNS provider. The NS records in the parent zone, then a zone export from the provider each one points to. Cloud account list Accounts opened outside central control, whose public endpoints carry no name under the organization’s domains. The organization or management group listing at each cloud provider, checked against billing records. Certificate issuance records Hosts covered by a wildcard certificate or by a private certificate authority. The issued certificate list in the certificate authority account or the certificate management system. Sign-on and purchasing records Hosted software services that hold the organization’s data under the provider’s own domain. The application list in the identity provider and supplier invoices held by finance. Each record is loaded as candidate seeds or as attribution evidence. An asset that an inside record lists and outside discovery missed points to a source or a seed that needs adding. Attribution of Discovered AssetsAttribution is the decision that a discovered asset belongs to the organization, made from evidence and recorded with it. It keeps other organizations’ systems out of the inventory and sets the boundary for active testing. Evidence Weight Reason The name sits in a DNS zone the organization administers. Strong Only the zone administrator can create the record. The address is in a block registered to the organization. Strong The registry record names the holder. The resource appears in the organization’s cloud account. Strong The provider’s API confirms the customer. The host shares name servers, a favicon or a tag identifier with confirmed assets. Moderate Suppliers reuse the same configuration across customers. The page shows the organization’s name or logo. Weak Anyone can copy a name or a logo. The host shares an address with a confirmed asset. Weak Shared hosting places many customers on one address. Each asset then carries a status: confirmed, probable, managed by a third party, or unrelated. The NCSC guide asks the same of products, which should validate or confidence score discoveries, show their provenance, and let users exclude what is out of scope. For a third-party platform, the organization controls the name and the provider controls the system, so active testing may need the provider’s permission too. Our guide to penetration testing explains authorization and rules of engagement. The inventory holds one record for each asset, with enough detail to review the decision later. Identity: The record holds the name, the address, the port and the product that answered. Provenance: It names the source that first reported the asset and each source that has confirmed it since. Dates: First seen and last seen dates show how long an exposure has existed and whether the asset still answers. Attribution: The status is stored with the evidence behind it and the person who decided. Ownership: A named owner and the purpose of the exposure are recorded, and a blank in either field is treated as a finding. Decision: The record states whether the asset stays exposed, is restricted or is removed, with the date. Monitoring the Surface for ChangeAttack surface monitoring repeats discovery on a schedule and reports the difference from the last run. The inventory is the baseline, and the changes are where new exposure appears. New names and certificates: A hostname or certificate appears that was absent in the previous run. Changed targets: A DNS record now points to a different address or platform. New services: A port has opened, or a sign-in page or console has appeared on a known host. Removed systems: A host has stopped answering while its DNS record remains. Published guidance gives a floor for how often discovery should run. Reference What it sets Where to find it CISA Binding Operational Directive 23-01 US federal civilian agencies run automated asset discovery every 7 days. BOD 23-01 on cisa.gov ASD Essential Eight maturity model, Maturity Level One Automated asset discovery at least every two weeks, to support vulnerability scanning. Essential Eight maturity model on cyber.gov.au NIST Cybersecurity Framework 2.0, ID.AM-01, ID.AM-02 and ID.AM-04 Inventories of hardware, of software, services and systems, and of supplier services are maintained. NIST CSWP 29 on nist.gov NIST SP 800-53 Rev. 5, control enhancement RA-5(4) The organization determines what information about a system is discoverable, meaning what an adversary could obtain without breaching it, and takes corrective action. SP 800-53 Rev. 5 on nist.gov The NCSC guide notes that EASM products often refresh daily. Certificate logs and DNS can be followed close to real time, while full port sweeps typically run daily or weekly. A change report is acted on according to the kind of change. Expected change: A change that matches a ticket or a release is recorded against it and closed. Unexplained change: A new name or service with no matching request goes to the owner of the parent domain or address block for attribution. Routine churn: Addresses behind content delivery networks and autoscaled cloud services change constantly, so those assets are tracked by name and service. An unexplained change that exposes a new system is also a reason to test it, as the guide to continuous penetration testing explains. From Inventory to ActionAn inventory reduces risk only when each exposed asset leads to a decision. The steps below take an asset from discovery to a confirmed fix, and the guide to the attack surface sets out the wider order for reducing exposure. Assign an owner. Each asset gets a named person or team who can change it. Question the exposure. Where the owner has no reason for a system to be reachable from the internet, access is removed or restricted. Start with management interfaces. An administration interface for a firewall, VPN device or server rarely needs to answer from the internet, so it is the first exposure to close. Prioritize what remains. CISA advises using its Known Exploited Vulnerabilities catalog as an input to prioritization. Confirm by testing. A test shows whether a suspected weakness on the asset can be used. Recheck. The next discovery run confirms that the port is closed or the record is gone. Small, Mid-Sized and Large EstatesThe size of the organization changes where the effort goes. The NCSC buyer’s guide uses a small business with no in-house technical support, an organization with subdomains created over many years and a large enterprise with a large address space as its examples. Small organization: A short list of domains makes discovery quick, so the value lies in plain advice on what to close and who can close it. Mid-sized organization: Years of projects leave old subdomains behind, so subdomain discovery and the dangling record check carry most of the weight. Large enterprise: Large address blocks and subsidiaries make service identification and attribution between business units the main effort. MSP or MSSP: A provider keeps separate seeds, a separate inventory and a separate authorization for each client. The NCSC guide lists hierarchical access control, a parent and child arrangement in which one organization can view data across several others. Starting and Measuring a ProgramAn EASM program can start from a list of domains and reach a monitored baseline within a few weeks in most environments. Name an owner for the inventory. One role keeps the seed list and settles attribution disputes. Collect the inside records. Registrar, DNS and cloud account lists are gathered, with read-only access arranged so that they refresh. Agree the conditions for active scanning. System owners approve the scan windows, request rates and source addresses, and name any fragile hosts to exclude. NIST SP 800-115 advises caution with older systems, since some scans can cause system failures. Run the first discovery and review it with system owners. Assets without an owner, exposed management interfaces and dangling records are handled first. Set the schedule and the route for changes. Each source gets an interval, and each kind of change gets a person who answers for it. Progress is then read from the dates and statuses in the inventory. Unattributed assets: The count of assets still marked probable shows how much of the surface has an unconfirmed owner. Assets outside the register: The count of confirmed assets missing from the internal asset register shows how far that register lags. Time to close: The interval from a decision to restrict or remove an exposure to the run that confirms it shows how quickly owners act. A rising asset count after a new source is added usually reflects better coverage, so each measure is read beside the list of sources in use. A program run this way also supplies the discovery stage of continuous threat exposure management. Tools, Products and Managed ServicesEASM can be run with open tools, bought as a product or taken as a managed service. Discovery is the easier half, so the choice depends mostly on who will attribute assets and act on changes. For any route, the NCSC buyer’s guide raises points worth settling first. Discovery method: Establish how the attack surface is discovered and which sources are combined. Freshness: Decide how current the data needs to be, since refresh rates differ between checks. Scan traffic: Confirm how the SOC can identify the scanner’s traffic, for example by source address, so that it is told apart from an attack. Confirmation of fixes: Check whether a scan can be run on demand, which some products offer so that a resolved issue is confirmed immediately. Vulnerability assessment: Establish whether the product only warns that a system is likely affected by a known vulnerability or also checks that a live system is definitely vulnerable, which the guide notes often needs additional configuration and permissions. Our guide to vulnerability scanning explains how those checks work. EASM With ThreatDefenceExternal discovery at ThreatDefence is the work of the ASM agent, which starts from the domains a customer has confirmed as its own. Discovery record: The agent maps the internet-facing domains, hosts, services and certificates that sit behind those domains. Change record: It records every change, and a new or changed host can start a test by the testing agent inside confirmed and authorized scope. Action items: Each problem has one tracked action item, and SOC analysts review significant findings. Partner delivery: MSPs and MSSPs can deliver the service under their own brand, with a separate scope for each client. The agent runs on the ThreatDefence AI SOC platform, where a 24x7 SOC works beside automated security monitoring, compliance and cyber risk management. A host found beyond the confirmed scope waits for its owner’s authorization before any test, and the Brand agent covers lookalike domains and leaked credentials separately. How discovery feeds testing is set out on the agentic penetration testing page. Frequently Asked QuestionsWhat is the difference between EASM and attack surface management?Attack surface management is the broader term and typically includes internal assets, identities and cloud configuration as well as internet-facing systems. EASM is the external part: it is limited to what can be reached from the internet and is discovered from outside, without access to the network. Many organizations start with EASM because it needs only a list of domains.What is a seed in attack surface discovery?A seed is a starting point that the organization has confirmed as its own, typically a root domain, a registered address range, an autonomous system number or a legal entity name. Discovery expands each seed into hostnames, addresses and services. New root domains found along the way become seeds only after they have been attributed to the organization.Does EASM need agents or credentials?No. Discovery reads public records and the responses that exposed systems give to any visitor, so nothing is installed. Coverage improves when the organization also supplies DNS zone exports and read-only access to its cloud accounts, because those confirm ownership and reveal endpoints that outside sources have missed.Does EASM replace an internal asset inventory?No. The two are built from opposite directions and are most useful when compared. The inventory lists what the organization knows it manages, and EASM lists what the internet can reach under its names and addresses. Assets that appear only in the EASM result are the unmanaged ones, and inventory entries that EASM cannot see confirm that a system is internal.Can EASM find shadow IT?Yes, where the system is reachable from the internet and tied to the organization by a name, a certificate or an address. That covers test servers, supplier-hosted sites under a subdomain and cloud resources with public endpoints. Software services that staff sign up to without any link to the organization’s domains leave nothing for outside discovery to read.How often should external discovery run?Daily is a reasonable baseline in most environments. The published minimums listed in this guide are slower, at every 7 days for US federal civilian agencies and every two weeks under the ASD Essential Eight maturity model, and they suit an organization whose exposed systems change rarely. Where teams publish systems often, name and certificate sources are worth watching continuously.Does discovery send traffic to the organization's systems?Yes, in part. Reading DNS, certificate logs, registration records and scanner indexes sends nothing to the organization’s own servers beyond ordinary name lookups. Fingerprinting and port scanning do send requests, so those checks are kept to attributed assets, run at an agreed rate, and come from source addresses the security team knows.Does EASM cover IPv6 addresses?Yes, mostly through names. An IPv6 range is too large to sweep address by address, so IPv6 hosts are found from AAAA records, certificate names and cloud account inventories, and are then scanned individually. CISA’s BOD 23-01 covers assets reachable over IPv4 or IPv6, while its weekly discovery requirement is set for the IPv4 space an agency uses.What is the difference between EASM and vulnerability scanning?EASM finds and attributes the assets, and a vulnerability scanner checks a given list of targets for known weaknesses. The NCSC buyer’s guide separates them the same way: an EASM product can warn that a system is likely affected by a known vulnerability, while vulnerability scanning checks whether a live system is definitely vulnerable. The EASM inventory supplies the target list a scanner needs.