Security Intelligence
November 18, 2026 | Security intelligence
What the Optimizely incident reveals about third-party website scripts
On November 17, 2026, Optimizely, a software company that provides tools for building and optimizing websites, disclosed that a threat actor exploited a vulnerability in its Web Experimentation platform and injected malicious JavaScript into customer projects. Optimizely's Web Experimentation product allows organizations to test different versions of web pages, such as alternate checkout layouts, to determine which performs better for visitors.
Initial reporting focused on Optimizely. On November 16, 2026, compromised Optimizely snippets delivered malicious code to visitors of customer websites. Researchers found the injected code in 188 of the 1,611 Optimizely snippets they examined, affecting 178 organizations.
Researchers later identified tampered scripts associated with three additional web services, Kissmetrics, StorifyMe, and Minute Media, that may have used the same infrastructure. The earliest observed activity involving those services dates to November 13, 2026, three days before the Optimizely injections.
The malicious code performed several functions, including replacing cryptocurrency wallet addresses, capturing payment card data entered into checkout forms, collecting cookies and page screenshots, and redirecting iPhone and iPad users to infrastructure associated with iOS exploit kits. Optimizely has since fixed the vulnerability and removed the malicious code from affected projects. How the code reached the other three services has not been publicly disclosed.
Since these reports emerged, Field Effect has received questions from partners and clients about whether Field Effect MDR would detect this activity. Field Effect MDR monitors clients' endpoints, networks, cloud environments, and identities.
In this campaign, the affected organizations' own servers, networks, and website code remained unchanged. Instead, the threat actor altered code hosted on third-party vendor platforms, which was then executed within visitors' web browsers. Because both the compromise and the resulting data theft occurred outside environments typically monitored by MDR services, activity of this type generally falls outside MDR visibility and into the domain of web fraud investigations, payment processors, and incident response activities focused on customer-facing web applications.
This guide explains how the attack worked, who may be affected, what organizations can do to determine whether their websites loaded compromised scripts, and where traditional security monitoring provides visibility.
What is Optimizely Web Experimentation?
Optimizely is a digital experience platform. Its Web Experimentation product enables marketing and e-commerce teams to run A/B tests and personalize website content. Organizations deploy the service by loading a JavaScript snippet, typically from Optimizely's content delivery network (CDN) at cdn.optimizely[.]com, although some organizations serve the same snippet from their own domains.
Optimizely generates each snippet from a customer's project configuration and automatically republishes it whenever project settings change. The script reference embedded on the website remains unchanged, while visitors' browsers retrieve the current version of the snippet each time a page loads. As a result, changes made within the Optimizely platform can immediately affect website visitors without requiring any modification to the website itself.
One configuration option, the Project JavaScript field, allows custom code to execute on every page that loads the Optimizely snippet. Anyone who gains the ability to modify that field can effectively run code in the browser of every visitor to every affected page.
Because the Optimizely snippet is often added by marketing teams or through tag management platforms such as Google Tag Manager, security and IT teams may be unaware that it is present.
How do third-party scripts work?
Website owners add third-party scripts to get features without building them: testing tools such as Optimizely, analytics that count and track visitors, chat windows, video players and advertising.
Someone adds one line of code that points to a file on the vendor's servers, and the line usually stays in place for years.
There are two ways that line can end up on a website:
- Directly in the website's code: A developer adds the script to page templates, and its presence can usually be confirmed by searching the website's code for the vendor's name, such as "optimizely."
- Through a tag manager: A tag manager, such as Google Tag Manager, allows marketing teams to add external services through a web interface. In this case, the website's code contains only the tag manager's script, and a search of the code may not reveal that Optimizely is being used. The Optimizely configuration exists only within the tag manager.
The website itself stores only a reference to the third-party script. Each time a visitor loads a page, their browser retrieves the script from the vendor and executes it within the page. As a result, the script can access the page and its contents in much the same way as the website's own JavaScript. Optimizely, Kissmetrics, StorifyMe and Minute Media scripts all operate in this manner.
Vendors can update these scripts at any time, and visitors automatically receive the latest version the next time they load a page. This allows vendors to deliver new features and fixes quickly. It also means that if a threat actor modifies a script or gains the ability to alter what is delivered, every website that relies on that script can be affected even though the websites' own servers and code remain unchanged.
Hypothetical example
To illustrate how this works, consider the following hypothetical scenario.
Note: This example is for illustration purposes only and does not reflect IKEA's actual website configuration or any confirmed impact on IKEA customers.
Suppose IKEA uses Optimizely to test two versions of a product page. Every page includes a line of code that instructs visitors' browsers to retrieve a script from Optimizely. On the afternoon of November 16, 2026, a shopper visits an IKEA product page. The page itself is delivered by IKEA and contains no malicious code. As part of the normal page load process, however, the shopper's browser retrieves the Optimizely script. On that day, the script has been altered to include malicious code, which now executes in the visitor's browser as though it were part of the website.
The shopper adds a sofa to their cart and proceeds to checkout. If the checkout page also loads the Optimizely script and collects payment card details directly in fields controlled by IKEA, the malicious code could capture the card number, expiry date and card verification value (CVV) as it is entered and transmit it to the threat actor. In this scenario, the malicious code functions as a web skimmer, collecting payment card data directly from the shopper's browser before it is processed by the website or payment provider.
If payment card information is collected through payment-provider-hosted fields embedded in a separate frame, the malicious code typically cannot access that data because browser security controls isolate the payment provider's content from the rest of the page. In either case, the purchase proceeds normally, the shopper receives an order confirmation and IKEA sees what appears to be a legitimate transaction.
If card details are successfully captured, the threat actor obtains payment card information that may be used for fraudulent purchases or sold on criminal marketplaces. The shopper may not become aware of the theft until unauthorized transactions appear on their account. How many payment cards may have been captured during this campaign has not been publicly disclosed.
Exposure would generally be limited to visitors who loaded affected pages while the malicious code was present. For payment card theft, websites that used payment-provider-hosted fields would typically be less exposed because browser security controls prevent page scripts from directly accessing card data entered into those fields.
Who could be affected?
Three groups could be affected.
The first is organizations associated with Optimizely Web Experimentation snippets that researchers observed serving the malicious code. One firm that investigated the incident examined 1,611 Optimizely snippets and identified the malicious code in 188 of them, associated with 178 organizations. Banks and financial services firms represented the largest sector identified, accounting for 32 organizations. (Note: Optimizely CMS and Optimizely Feature Experimentation are separate products. Public reporting to date has focused on Optimizely Web Experimentation.)
The second group is people who visited websites associated with the organizations in the first group while those websites were serving the malicious code. Researchers reported that the first known Optimizely injections appeared on November 16, 2026, and that most affected websites had been cleaned by approximately 21:52 UTC that day, although the exact period of exposure varied by website. These visitors may include customers, partners, or employees who accessed affected websites from corporate or personal devices during the period the code was present. Researchers published a list of 178 identified organizations but noted that the actual number may be higher because only a subset of Optimizely snippets was reviewed.
The third group is organizations that load scripts from three other web services reported to have served tampered code: Kissmetrics, StorifyMe and Minute Media. Researchers traced modified versions of these scripts to infrastructure also used in the Optimizely activity, with the earliest observed activity dating to November 13, 2026. How the code reached these services and when the activity ended have not been publicly disclosed. Organizations using these services should review activity beginning November 13, 2026, and continue that review until their vendors can confirm the period of exposure.
What is the impact?
For organizations in the first group, the primary risk is the exposure of information entered by visitors while the malicious code was present. Researchers reported that, beginning at approximately 13:43 UTC on November 16, 2026, the skimming functionality captured payment card numbers, expiry dates and card verification values (CVVs) entered into website-controlled payment forms. Websites that used payment-provider-hosted fields embedded in separate frames, such as Stripe or Adyen hosted fields, would generally be less exposed because browser security controls prevent page scripts from directly accessing card data entered into those fields. On pages where the malicious code replaced cryptocurrency wallet addresses, it also collected screenshots and cookies that lacked the HttpOnly attribute, potentially exposing active user sessions.
For visitors, the code replaced cryptocurrency wallet addresses with addresses controlled by the threat actor, and those wallets reportedly received approximately $7,040 on November 16, 2026. iPhone and iPad users were redirected to infrastructure associated with two iOS exploit kits, Coruna, which targets iOS 13.0 through 17.2.1, and DarkSword, which targets iOS 18.4 through 18.7. Public reporting has not demonstrated successful compromise of an iPhone or iPad resulting from this activity.
For organizations that accept payment cards, exposure of payment card data may trigger obligations under the Payment Card Industry Data Security Standard (PCI DSS), requirements from payment processors, and potentially customer notification obligations depending on jurisdiction and the findings of subsequent investigations.
What are threat actors leveraging, how, and why?
The threat actor leveraged the trust websites place in third-party JavaScript. According to public reporting, the attacker exploited a vulnerability in Optimizely's platform to gain the ability to modify customer projects and then inserted a small loader into the Project JavaScript field. Because Optimizely automatically republishes customer snippets, any page loading an affected snippet would execute the loader, which then retrieved additional code from the threat actor's infrastructure at 1688360[.]cc.
This approach gave the threat actor reach at scale. Rather than compromising individual websites one by one, the attacker used a trusted third-party service to distribute malicious code across websites associated with 178 organizations identified by researchers. Those organizations included banks, retailers and cryptocurrency platforms where visitors may enter payment information or conduct financial transactions. Researchers also reported that the payload was updated 26 times throughout the day, adding functionality and changing cryptocurrency wallet addresses.
How were threat actors able to bypass protections?
The malicious code was delivered through cdn.optimizely[.]com, a domain already trusted by affected websites. Organizations that use a Content Security Policy (CSP) commonly allow content from that domain, which meant browsers treated the code as coming from an approved source. Researchers reported that 24 affected organizations, approximately 15 percent of those examined, had CSPs strict enough to block the attack, while 68 had no CSP and 50 had policies that allowed the code to execute and exfiltrate data.
Researchers also found that the loader existed in multiple forms: a direct link to the payload, a Base64-encoded version designed to make detection more difficult, and a fully embedded fallback version that could execute even if the external domain was blocked. Stolen information was sent directly from visitors' browsers to the threat actor's infrastructure, meaning the affected websites' own servers generally continued to see normal traffic patterns.
Researchers additionally observed at least one snippet that was cleaned and later found serving a concealed version of the malicious code again. Public reporting has not disclosed whether this reflected a persistent access mechanism, incomplete remediation, or a separate re-compromise event.
Why is this significant?
These findings demonstrate how visitors can be exposed to malicious code even when a website's own servers and codebase have not been compromised. Instead of targeting individual organizations directly, the threat actor leveraged a trusted third-party service that many organizations had already integrated into their websites.
Researchers also identified tampered scripts associated with three additional services, Kissmetrics, StorifyMe and Minute Media, that were linked to the same hosting infrastructure and loader code. While this overlap does not prove that the same threat actor was responsible in every case, it highlights how compromises involving third-party scripts can potentially affect large numbers of websites through a single dependency.
The incident also reinforces the risks associated with unmanaged third-party code. Services such as Optimizely, Kissmetrics, StorifyMe and Minute Media are commonly used to provide analytics, marketing, personalization and other functionality. These tools are often deployed by marketing or business teams, either directly or through tag management platforms such as Google Tag Manager, and may not be included in traditional asset inventories or security reviews. As a result, some organizations lack a complete inventory of the third-party code executing on their websites.
For organizations that process payment card data, the incident is also a reminder of the controls introduced by PCI DSS requirements 6.4.3 and 11.6.1, which call for the management and monitoring of scripts running on payment pages and the detection of unauthorized changes. These controls were designed specifically to reduce the risk posed by web-skimming and other third-party script attacks.
Where does this activity happen, and what can Field Effect MDR see?
Managed Detection and Response (MDR) services monitor an organization's endpoints, networks, cloud environments and identities. In this campaign, the malicious code was introduced through Optimizely's platform, delivered through Optimizely's content delivery network (CDN), and executed within visitors' web browsers on their own devices. Data collected by the malicious code was then transmitted directly from those browsers to infrastructure controlled by the threat actor.
Based on public reporting, affected websites continued serving their normal content throughout the incident. The malicious activity occurred within third-party services and visitors' browsers rather than on the website owners' servers. As a result, the compromise itself largely took place outside the systems that an MDR service typically monitors and protects.
This distinction is important because an organization could have fully monitored endpoints, networks, cloud services and identities yet still have limited visibility into malicious code executing within a customer's browser after being delivered by a trusted third-party script provider.
Determining whether an organization was affected therefore relies on examining the website and the third-party services it uses, including Optimizely, Kissmetrics, StorifyMe and Minute Media. Relevant sources of evidence may include vendor audit logs, project configurations, tag management platforms, content security policy reporting, client-side script monitoring tools and any records of changes made to third-party web assets.
This type of investigation focuses on the website supply chain rather than traditional endpoint or network telemetry, which is why the validation steps outlined in the next section are critical for determining exposure.
How do I know if my organization is affected?
The steps below can help determine whether a website uses scripts from Optimizely Web Experimentation, Kissmetrics, StorifyMe or Minute Media, and whether those scripts may have been associated with the reported activity. Because Optimizely removed the malicious code from affected projects on November 16 and 17, 2026, a script that appears clean today may still have served malicious code during the period of exposure.
- Determine whether the website uses the affected services. Open the website in Chrome or Edge, press F12, select the Network tab, reload the page and search for references to "optimizely", "kissmetrics", "storifyme" or "minute". Review home, product, checkout and login pages, as different sections of a website may load different third-party services. Organizations should also review tag management platforms such as Google Tag Manager, as these services are often deployed through marketing-managed tags rather than directly through website code.
- Request information from the vendors. Organizations using Optimizely should request project audit logs, project identifiers, change histories and any records associated with the reported activity. Organizations using Kissmetrics, StorifyMe or Minute Media should ask those vendors whether their scripts were affected, the period of exposure, and whether any customer environments were impacted.
- Review project configurations and custom code. Within Optimizely projects, review the Project JavaScript field and any experiment-specific custom code. Look for references to known indicators such as 1688360, atob(, .icu, track.js, webshell, or any unrecognized external scripts.
- Assess what data could have been exposed. Determine whether pages loading the affected scripts collected payment card information directly in website-controlled form fields or used payment-provider-hosted fields. Review whether session cookies were configured with the HttpOnly attribute and identify any other sensitive information that may have been accessible through page-level JavaScript.
- Review available logs and monitoring data. Content Security Policy (CSP) reports and client-side monitoring tools may show whether visitors' browsers connected to suspicious domains. Organizations may also review corporate DNS, web proxy or secure web gateway logs to determine whether internal users visited affected sites or connected to known indicators. Public reporting identified domains and infrastructure including 1688360[.]cc, 136.113.106[.]136, 5vobauua3ojpklx[.]icu, eil786npto9u5bq[.]icu, and webshell[.]lol.
- Assess payment card exposure. If a website was serving the malicious code on pages where visitors entered payment card information into website-controlled form fields, the organization should evaluate whether a payment card data incident may have occurred. This evaluation may involve engaging the payment processor, legal counsel, incident response providers and any other parties required by contractual, regulatory or PCI DSS obligations.
Because much of the activity occurred within visitors' browsers rather than on the affected organizations' infrastructure, determining exposure may require combining vendor-provided information, website configuration reviews and client-side monitoring data rather than relying solely on traditional server, endpoint or network logs.
Mitigations
Optimizely has remediated the vulnerability on its platform, so customers do not have software updates to apply. Organizations using Optimizely may wish to review all accounts, integrations and application programming interface (API) keys with permission to modify projects and rotate credentials where exposure is suspected or where organizational policy calls for it. Organizations can also assess whether session cookies lacking the HttpOnly attribute may have been exposed and rotate potentially affected sessions as appropriate. Known malicious indicators can be blocked at Domain Name System (DNS), web proxy and secure web gateway layers. At the same time, organizations using Optimizely may wish to keep cdn.optimizely.com accessible, as it remains a legitimate service used to deliver Optimizely content.
Where payment card data may have been exposed, engaging payment processors, legal counsel and incident response providers can help determine whether card replacement, forensic investigation, regulatory reporting or customer notification activities are warranted.
Over the longer term, organizations can benefit from maintaining an inventory of third-party scripts running on their websites, including those deployed through tag management platforms such as Google Tag Manager, and assigning a business owner to each one. Content Security Policies (CSPs) can help limit the impact of unauthorized scripts when configured appropriately, particularly when unnecessary use of unsafe-eval and overly permissive directives are avoided. Client-side monitoring tools may provide additional visibility into unauthorized script changes and unexpected browser connections. Organizations that process payment cards could consider limiting marketing, analytics and testing scripts on checkout pages where operationally feasible and using payment-provider-hosted fields to reduce exposure of card data to page-level scripts. These practices also align with Payment Card Industry Data Security Standard (PCI DSS) requirements 6.4.3 and 11.6.1, which were introduced to address risks associated with web-skimming and third-party script compromises.
Security takeaway
This incident highlights how different parties control different parts of the website supply chain, and how weaknesses at any point in that chain can expose organizations and their visitors to risk.
Script vendors
According to public reporting, the threat actor exploited a vulnerability in Optimizely's platform to introduce malicious code into customer projects. Because Optimizely automatically republishes customer snippets, those changes were delivered to websites without any action by the website owner. Researchers also observed at least one snippet that was cleaned and later found serving a concealed version of the malicious code again, suggesting the original issue may not have been fully remediated at that point. When the underlying access or mechanism was eliminated has not been publicly disclosed.
Optimizely has since remediated the vulnerability on its platform. The technical details of the flaw, a Common Vulnerabilities and Exposures (CVE) identifier and a severity rating have not been publicly released. For Kissmetrics, StorifyMe and Minute Media, public reporting describes tampered scripts but has not identified how those services were affected or whether a vulnerability was involved.
Website owners
Organizations had nothing to patch on their own infrastructure, but the incident illustrates several common challenges associated with third-party web services. Scripts used for analytics, marketing, personalization and testing are often deployed outside traditional information technology procurement and asset-management processes. As a result, some organizations may not have a complete inventory of the third-party code running on their websites.
Researchers reported that many affected organizations either lacked a Content Security Policy (CSP) or had policies that allowed the malicious code to execute and communicate externally. The incident also demonstrates the limitations of relying solely on domain allow lists, as the code was delivered through trusted services that organizations had already approved.
Pages that collected payment card information directly in website-controlled forms were more exposed to web-skimming activity than those using payment-provider-hosted fields. Cookies lacking the HttpOnly attribute were also accessible to browser-based scripts. Accounts, integrations and application programming interface (API) keys with permission to modify website content or third-party scripts represent additional areas organizations may wish to review.
Website visitors
Visitors' browsers executed the malicious code whenever they loaded affected pages. Cryptocurrency wallet addresses displayed on some pages were replaced with addresses controlled by the threat actor, potentially redirecting payments to attacker-controlled wallets. Some iPhone and iPad users were redirected to infrastructure associated with the Coruna and DarkSword exploit kits, although public reporting has not demonstrated successful compromise of a device resulting from that activity.
MDR providers
Much of the activity occurred on third-party platforms and within visitors' browsers rather than on the affected organizations' servers. As a result, traditional endpoint, network, and cloud monitoring may provide limited visibility into this type of activity. Website-side visibility is more likely to come from controls such as Content Security Policy reporting, client-side script monitoring, vendor audit logs, and third-party risk management practices.
Maintaining an inventory of third-party scripts, assigning ownership for each one and monitoring changes to externally hosted code can provide organizations with a stronger foundation for managing website supply-chain risk.
Stay on top of emerging threats like this.
Sign up to receive a weekly roundup of our security intelligence feed. You'll be the first to know of emerging attack vectors, threats, and vulnerabilities.

