The top MITRE ATT&CK™ behaviors to monitor for on your endpoints and servers

When the OWASP Top 20 Vulnerabilities was first published it revolutionized our industry’s approach to vulnerability management. Instead of playing wack-a-mole with thousands of individual vulnerabilities every time a new one was discovered, we approached vulnerability management by primarily addressing these Top 20 Techniques. (Yes, I know we’ve gotten away from that a bit with all the flashy vulnerability names we starting using…but that’s another post for another time.)

Still considered “advanced”, behavioral detection has just begun to hit the mainstream. But, as the incident response (IR) cases we support continually confirm, adoption is still lagging for 90% of the mid- and SMB market. It’s in no way controversial anymore to state that, in order to detect and stop modern attacks, organizations need to have behavioral monitoring capabilities, especially on the endpoint.

A problem is that we’ve been led to believe that if we adopt behavioral detection, we need to spend a lot to maximize coverage of all the various attacker behaviors. This is a disservice.

Our intention in detection among most organizations (aka, the ones that don’t have a full time threat intel team) should be to stop focusing on individual, novel attack techniques and concentrate defenses against the Top 20 most commonly observed ATT&CK techniques that are also achievable to monitor. These are the ones that actually matter, and the ones that will catch more bad guys, more often.

The following list is consolidated data and cross referenced with various forensic reports on observed attacks over the last 36 months:

TOP 20 ADVERSARY TECHNIQUES:

Rank	Tactic	Id	Technique
1	Execution	T1059	Command Line Interface / Powershell
2	Initial Access	T1078	Valid Account Misuse
3	Discovery	T1082	System Information Discovery
4	Persistence	T1060	Registry Run Keys
5	Credential Access	T1003	Credential Dumping
6	Lateral Movement	T1021	Remote Services
7	Execution	T1055	Process Injection
8	Persistence	T1053	Scheduled Tasks
9	Defensive Evasion	T1218	Signed Binary Proxy Execution
10	Persistence	T1547	Boot/Logon Autostart Execution (esp. Shortcut Modification)
11	Execution	T1047	Windows Management Instrumentation (WMI)
12	Defense Evasion	T1036	Masquerading
13	Privilege Escalation	T1574	Hijack Execution Flow
14	Defense Evasion	T1027	Obfuscated Files or Information
15	Defense Evasion	T1497	Virtualization/Sandbox Evasion
16	Lateral Movement	T1544	Remote File Copy
17	Defense Evasion	T1089	Disabling Security Tools
18	Initial Access	T1190	Exploit Public Facing Application
19	C2	T1219	Remote Access Software (e.g. RDP)
20	C2	T1505	Webshells

Making your detection capabilities robust against these 20 techniques will deliver more bang for the buck than any other approach while saving you time and money from hunting “Bluebird” techniques and behaviors you’re more than likely not going to see in your one network.

Is Top 20 enough?

Yes!

We respond to a lot of attacks and have been doing threat hunting and response in organizations large and small. There have been very few attacks that don’t exhibit behaviors that overlap with the above list of 20 that you could be monitoring for today.

When SolarWinds Solarigate a.k.a. SUNBURST hit in December 2020, everyone said this was novel; and the entry vector certainly was. Once you dug in though, the same top 20 behaviors could be observed: The novel supply chain vulnerability was used to spawn malicious Powershell (T1059), scripts (T1059), memory injections (T1055), lateral movement (T1544) techniques and credential dumping (T1003).

When Hafnium hit Exchange Servers using the latest Exchange zero-days we saw the same things: new novel entry vectors leading to many of the same top 20 common behaviors like WebShells (T1505) spawning PowerShell commands (T1059) and injecting Cobalt Strike into memory (T1055).

Everyone effectively monitoring for the top 20 attacker behaviors had the visibility to see these attacks unfold and my prediction is the next big vulnerability will be found by monitoring for them as well.

Conclusion

Ultimately, the Top 20 approach is an acknowledgement that not all techniques are necessary to alert or monitor to detect attacks. Defense in Depth still works: every tactic and technique you have visibility on is a detection opportunity in the attack chain, and the top 20 is broad enough to cover you against even some of the most advanced attackers. We are all strapped for resources; don’t chase the highest coverage and focus on the top 20. With these 20, there are exceedingly few attacks that could ever get past your notice.