What Is NDR? Network Detection and Response Explained

Network detection and response, or NDR, is a security approach that watches network traffic for signs of an attack already inside. It learns what normal traffic looks like, flags behavior that breaks the pattern, and responds by isolating hosts or blocking connections. NDR matters because most attacks keep moving inside the network long after the first breach.

By Abdullah Shafique, AI Engineer

What is network detection and response?

Network detection and response is a cybersecurity technology that applies behavioral analytics, machine learning, and artificial intelligence to network traffic to find suspicious activity, then acts on it. Gartner renamed the category from network traffic analysis in 2020, when automated response joined passive monitoring. In May 2025, Gartner published its first Magic Quadrant for NDR, treating it as a distinct security category. MarketsandMarkets projects the NDR market will reach USD 5.82 billion by 2030, growing at a 9.6% compound annual growth rate.

NDR watches two traffic flows. North-south traffic moves between internal networks and the outside world, carrying initial access attempts and data leaving the company. East-west traffic moves between systems inside the network, showing an attacker who already has a foothold and is widening access. That inside view is what sets NDR apart from tools built for the perimeter. NDR also covers devices that cannot run a security agent: internet-connected sensors, operational technology, and legacy systems. Our CDR guide covers the cloud side of detection, where the signals come from control planes and workloads instead of packets.

Why is the network the best place to catch an attack?

Most attacks now hide in encrypted traffic and move between systems with stolen credentials, and the network is where both show up. Zscaler's ThreatLabz 2024 research found over 87% of cyberthreats leverage encrypted TLS/SSL channels, which makes old-style deep packet inspection useless unless a team decrypts traffic first.

Vectra and Enterprise Strategy Group reported in November 2022 that 70% of organizations had fallen victim to an attack that used encrypted traffic to avoid detection, and 45% had fallen victim more than once. Two thirds of those surveyed still lacked visibility into all of their encrypted traffic.

Lateral movement is the other reason the network matters: Vectra cites Elisity research finding that over 70% of successful breaches involve lateral movement, undetected for 95 days on average without network visibility. A workstation that suddenly opens file servers it has never touched, or a login at 3 a.m. from a service account, reads as normal to a firewall and as an attack to an NDR platform that knows the baseline.

How does NDR actually work?

An NDR platform runs three stages: collection, detection, and response. Collection copies traffic through SPAN port mirroring, dedicated network TAPs, or packet brokers that filter and balance traffic across sensors. In cloud environments, the same telemetry comes from native sources such as AWS VPC Flow Logs, Azure Network Watcher, and GCP audit logs.

Detection starts with a baseline: behavioral analytics learns what normal looks like for each network segment, device type, and user group, and machine learning models sort deviations into benign anomalies and genuine threats. For encrypted traffic, the platform works from metadata instead of content: packet sizes, connection timing, and TLS fingerprints such as JA3 and JA3S identify malware families and command-and-control tools without breaking encryption. DNS correlation maps encrypted connections to the domains they resolved, which exposes regular check-in patterns, domain generation algorithms, and data leaving through DNS tunnels.

Response is what separates NDR from older network monitoring. Detections arrive ranked by severity and confidence, so analysts start with the incidents most likely to be real. Automated actions can end a connection to a malicious destination, isolate an infected host, or trigger a playbook in a SOAR platform. Full packet capture keeps the raw evidence for forensics and threat hunting after the incident is contained.

How is NDR different from EDR, XDR, SIEM, and IDS?

ApproachWhat it coversWhere it falls shortBest fit
NDRNetwork traffic, north-south and east-westCannot see endpoint processes or file activityTeams needing visibility across IoT, OT, and cloud networks
EDREndpoints: laptops, servers, and workstationsNeeds an agent on every device; no network-wide viewTeams starting with device security
XDREndpoint, identity, cloud, network, and email in one viewBreadth can cost depth in one areaTeams correlating attacks across domains
SIEMEstate-wide logs, stored for search and reportingDetection depends on log quality and availabilityTeams needing retention and compliance reports
IDSKnown threat signatures on network trafficMisses novel threats and encrypted trafficTeams keeping a legacy signature layer

The practical test is a lateral movement scenario: log in with a stolen credential to a server the account has never touched, move to a second system, and start a large outbound transfer. Check whether the platform builds one incident with a timeline or a pile of separate alerts, and whether it can isolate the host without touching the rest of the segment. A tool that only matches known signatures will miss the whole sequence. Our EDR guide covers the endpoint side, and our XDR guide covers how these signals combine across domains.

What does a breach cost when network threats go unanswered?

IBM's 2025 Cost of a Data Breach Report, released July 30, 2025, put the global average cost of a data breach at $4.44 million, the first decline in five years. The average breach lifecycle, the mean time to identify and contain a breach, fell to 241 days. Organizations that detected the breach themselves saved an average of $900,000 compared with those told about it by an attacker.

The time to act keeps shrinking. CrowdStrike's 2025 Global Threat Report, released February 27, 2025, found the average eCrime breakout time, the time from initial access to lateral movement, had dropped to 48 minutes, with the fastest recorded breakout at 51 seconds. The same report found 79% of attacks to gain initial access were malware-free, so the intrusion arrives as a legitimate login instead of a file an antivirus product can catch.

Chris Kozup, Chief Marketing Officer at Darktrace, said in June 2025: "As digital environments become more complex and traditional network perimeters dissolve, organizations rely on Darktrace's differentiated AI-based approach to detect and autonomously respond to known and novel threats." Gartner's first Magic Quadrant for NDR, published May 29, 2025, named Darktrace a Leader.

What should a team check before buying NDR?

Start with coverage across every network the team runs: offices, data centers, cloud VPCs, and the operational technology and internet-connected devices that cannot take an agent. Ask how the platform handles encrypted traffic without decryption, and what share of detections come from metadata analysis instead of signatures.

Test detection on a real attack during the trial: move laterally with a stolen credential, open an unusual outbound connection, and check how fast the alert arrives and whether it arrives as one incident or a pile of separate events. Ask how the platform ranks alerts, what false positive rate the team should expect in the first month, and which response actions it can take on its own: host isolation, connection blocking, and SOAR playbook triggers.

Check pricing against the team's actual traffic: some vendors price by bandwidth, others by sensor or protected device, and the number changes fast when the team adds a site or a cloud region. Ask what it costs at twice the current traffic before signing.

Frequently asked questions

What does NDR stand for?

Network detection and response: a security technology that applies behavioral analytics and machine learning to network traffic to find suspicious activity, then responds by isolating hosts, blocking connections, or triggering automated playbooks.

How is NDR different from IDS?

A traditional intrusion detection system matches traffic against known signatures, so it catches known threats fast and misses novel attacks and encrypted traffic. NDR adds behavioral analytics and machine learning, which flag unknown threats, and response actions that a legacy IDS lacks.

Can NDR see encrypted traffic?

Yes, without decrypting it. NDR analyzes metadata such as packet sizes, connection timing, TLS fingerprints, and DNS lookups to identify malicious behavior inside encrypted sessions. Vectra reported in 2022 that 66% of organizations still lacked visibility into all their encrypted traffic.

Does NDR replace EDR or SIEM?

No. NDR watches network traffic, EDR watches processes and files on endpoints, and SIEM stores logs from across the whole estate for search and compliance reporting. Most teams run NDR alongside EDR and SIEM, and some combine them under an XDR platform.

How fast do attackers move after breaking in?

CrowdStrike's 2025 Global Threat Report found the average eCrime breakout time was 48 minutes, with the fastest recorded breakout at 51 seconds.

Which traffic worries your team more: the encrypted traffic you cannot inspect, or the east-west traffic moving between your own servers? Share in the comments how your team watches for lateral movement today. If this helped, use the share buttons to pass it to your team.

Comments

Popular posts from this blog

SaaS Backup: Protecting the Data Your Cloud Apps Hold

Passkeys for Business: Why Passwordless Login Wins

MCP Server Security: Securing the Model Context Protocol Integration Layer for Enterprise AI Agents