Automating triage and incident response of phishing alerts

Security orchestration and automation is an undeniably hot topic. Forrester named it one of the top 10 technology trends to watch in 2018-2020. So, it’s clear there are lots of eyes on the space. But as SOC managers start to look at implementing security automation, they often find themselves asking, “where do I start?”



Welcome to the first of our four-part blog series where we will take a look at the steps to automate some of the most common SOC processes. Over the next few weeks, we will provide practical guides to automating steps that are part of managing, investigating and responding to alerts related to:

Finding opportunities for automation in a SOC isn’t hard; multiple areas can benefit substantially from it. Anything that involves a lot of manual work, requires fast response,  has low alert fidelity and/or requires involving an end user is a prime candidate for security automation. Automating tasks that fit these categories can save precious security analyst time that can be refocused on higher value activities like threat hunting and…gasp!…actual analysis.

Security automation triggers



Without further ado, let’s start with phishing. But first, why phishing?

Why Phishing?
Why phishing?

A seemingly simple but malicious email can be the prelude to a more sinister threat. In fact, 91% of cyberattacks start with a phishing email. Also, phishing is arguably the top attack vector, accounting for 90-95% of all successful cyber attacks world wide. Because a phishing email can lead to a much larger attack, phishing alerts demand the utmost attention.


Phishing checks three of the boxes as far as attributes that make it ripe for automation – a significant degree of manual work, low alert fidelity and the need to involve one or more users.

Phishing is extremely noisy for most organizations. Why? In addition to alerts that may appear through a SIEM or mail proxy, most companies have a mailbox where suspicious emails are sent both through automation and by end users sending what they think are questionable emails. Those mailboxes are extremely noisy and leave security teams with a lot to comb through.


But there’s a problem. The majority of phishing alerts can also turn out to be false positives. Thus, the entire process of handling phishing alerts – from data gathering and enrichment to feedback and remediation – can be painstaking and may take hours to complete. And you may end up having spent all that time on what turns out to be a false positive, robbing your attention from more pressing alerts and cases.


Let’s go over a typical process for handling a phishing alert and identify areas that you can speed up through security automation.


Step 1: Data Gathering/Enrichment Step 1
The first step in handling phishing alerts is to gather information to make sure you have full context before making a decision based on those alerts. For instance, you might want to know the particular end user who was the source of the alert and what role he/she plays in your organization. So, if that user plays a key role – maybe in HR or Finance – then you might have a spear phishing attack on your hands.


You might also want to source web intelligence from tools like XForce or AlienVault OTX to know if what you have is already considered a “known bad” by intelligence communities and global repositories. In addition, you might want to look into threat intelligence to determine whether the URL, hash, or file found in the phishing email has been associated with a known threat.

The information you gather will help you frame the right context so you can better defend your organization or even just decide whether the email is in fact malicious or just a regular spam. This early in the process, you can – and should – employ security automation to gather the information you need from various sources, as this step alone can be extremely time-consuming as the number of information sources grows.

Step 2Step 2: Deeper Analysis
While some phishing emails use links that direct users to malicious sites, others have file attachments that contain malware themselves. Thus, if a suspicious email is accompanied by an attachment, it would warrant deeper analysis.


But just grabbing those attachments, sending them to a mailbox, getting a report out, analyzing the data, making assumptions based on the report, and determining whether certain files are malicious or not, can take hours. Worse, it needs to be done on practically every email.

That’s a lot of work for an analyst to do, especially since he/she usually needs to collect data from 5,6,7, or 8 different sources to get the best possible context of a particular alert. A lot of this is just basic analysis that can certainly be automated.

Step 3Step 3: First-level determination

Once you’ve collected all the information you need and developed context, the next step would typically be to conduct first-level determination or triage. This step may consist of several detailed, but otherwise mundane, triage of alerts where decision making can be automated by simply basing those decisions on context. The rest can be escalated to an analyst for deeper investigation.

For example, let’s say you have this suspicious email you haven’t encountered before. Now, after Loginchecking its file attachment against a sandbox, seeing that the URL and hash all seem safe, and finding nothing suspicious about it, it would be logical to simply close the case automatically.

In this scenario, if you have a hundred phishing emails a day, maybe only three will end up needing an analyst’s intervention. This will certainly save the analyst from wasting a lot of time and energy on alerts that don’t need their expertise and keep them focused on the alerts that really matter.

Step 4Step 4: Deeper Investigation

But what if say, after checking the email against a sandbox, looking at the URL, and checking it against intelligence sources, the email is found to be potentially malicious? Well, this is where you want your analysts to step in to conduct a deeper investigation on the case.

Security automation can be used to bring together all relevant information sets (e.g. from mail-related log-source querying, endpoint-related querying, etc.) and put them in front of an analyst, who can then draw from his/her experience and expertise to make decisions on how to best move forward.

Step 5Step 5: Escalation/Response Path

While this step may not be automated, as they require people to handle escalation and incident response, the next step can certainly be.

Person  Step 6: Feedback/Remediation

After the phishing attack has been fully investigated and analyzed, automation can be used to carry out remediation tasks that would help your organization see to it that a similar attack will not be able to slip through if it ever happens again in the future. So, for example, automation can be used to blacklist the associated URLs and hashes, add associated IPs to firewall rules, carry out threat intelligence updates, and so on.

As we look back at the steps we mentioned, you’ll notice that a significant number of steps can be automated. That way, the analyst can focus on what he/she does best (analyze and make critical decisions), and do so in record time.

Security automation doesn’t replace the need for experienced analysts to apply critical thinking in the investigation and remediation of security events. However, when correctly applied, security automation can enable faster, better decision making for analysts and allow them to work more cases in a given day.

Rules of the Road
A few best practices to take with you as you embark upon your security automation journey:

  1. Always automate data collection and enrichment
  2. Automate triage activities when possible
  3. Automation empowers (not replaces) human decision making
  4. Sensitive actions should be analyst assisted
  5. Embrace consistency…it’s value to security operations can’t be overstated

Stay tuned for next week’s post in this series, where we will take a look at applying these same security automation principles to addressing malware alerts.

Can’t wait for the next installment? Get the whole story by checking out our recent webinar on Security Automation Quick Wins.