The DXP Scorecard logo
Independent Platform Evaluation
Scored on implementation experience
Not vendor briefings
Analyst EditorialSecurity

The Optimizely Script Hijack: What Every Buyer Should Ask

An attacker put malware on some of the world's best-known websites without breaching one of them. The way in was a testing script those sites had chosen to trust.

DXP Principal Analyst GroupPublished 8 min read

The short answer

On 16 November 2026 an attacker used a flaw in the Optimizely Web Experimentation signup flow to insert malicious JavaScript into customer projects, and researchers at Report URI found it served to visitors of 178 organizations. Optimizely's content management products were not affected. The lesson is wider than one vendor: any script a supplier hosts and your pages load can be changed without you, so ask every vendor how that code is controlled.

Key takeaways

  • The product hit was Optimizely Web Experimentation, the A/B testing tool. Optimizely says no other product was affected, which includes Optimizely SaaS CMS and Optimizely PaaS DXP, and that fewer than 5 percent of its customers were involved.
  • No affected website was breached directly. The attacker changed code inside Optimizely, and each site's own pages loaded it from Optimizely's content delivery network, a host those sites already trusted.
  • Optimizely says it detected the activity the same day, blocked the malicious code and closed the flaw within three hours of detection. That is a fast response. The first public account still came from an outside researcher, about two hours before Optimizely's own status page notice.
  • This risk is not specific to Optimizely. Testing, personalization, analytics, chat and consent tools from every large vendor work the same way: one line of script on every page, with the content of that script decided somewhere else.
  • Security reviews tend to stop at the content management system. The scripts that run in your visitors' browsers deserve the same questions, and after this week a vendor should expect to be asked them.

Our analysts’ take

Most marketing leaders can tell you which content management system runs their website and who signed the contract. Far fewer can list every third-party script that loads on the checkout page. Those scripts arrive one at a time, each for a good reason: a testing tool for the growth team, a personalization tag, an analytics library, a chat widget. Each one is a standing permission for another company to run code in your customers' browsers.

On Monday 16 November 2026 that permission was abused at scale. An attacker found a flaw in the signup flow of Optimizely Web Experimentation, used it to act as users on targeted customer accounts, and wrote malicious code into those customers' projects. The sites that loaded the Optimizely script then delivered that code to their own visitors. Researchers at Report URI counted 178 organizations, among them Zoom, the BBC, CNN, Salesforce, IKEA and HSBC.

We are not writing this to tell anyone to leave Optimizely. Our position is narrower and, we think, more useful. This was an attack on a way of working that nearly every marketing team depends on, and the buyers who learn from it will be the ones who change what they ask a vendor before signing, whichever vendor it is.

What to look for

  1. 1

    Who can change the script

    Ask which roles can edit code that runs on every page, whether that needs single sign-on and a second factor, and whether a second person must approve the change. A field that puts code on your whole site should not be editable by one account on its own.

  2. 2

    A record of every change

    Ask for an audit log of code and experiment changes that you can export and that the vendor cannot quietly lose. After an incident, the first question is what changed and when, and you want to answer it yourself.

  3. 3

    Limits your own site enforces

    A Content Security Policy is a set of rules that tells the browser which scripts may run and where data may be sent. Ask your agency whether your site has one, whether it is strict, and when it was last tested.

  4. 4

    How fast you would be told

    Ask what the contract says about incident notification: how many hours, through which channel, and to whom. A status page you have to remember to check is not a notification.

  5. 5

    Whether the script needs to be there

    Every tag was added for a reason that may have expired. Ask which pages each script loads on, and whether payment and login pages need any of them at all.

What happened on 16 November, in plain terms

Optimizely Web Experimentation works by having each customer add one script to its pages. That script is served from Optimizely and carries the customer's experiments, plus an optional field called Project JavaScript where the customer can write custom code that runs on every page. The attacker wrote into that field.

The way in was a signup flaw

Optimizely's statement of 18 November says an unauthorized third party exploited a flaw in the Web Experimentation account signup flow, impersonated users on targeted customer accounts and inserted malicious JavaScript into their projects. The company says this was not a credentials breach, meaning customer passwords were not stolen to do it.

The code arrived from a trusted host

The altered script was still served from Optimizely's own delivery network with Optimizely's certificate. Report URI checked 1,611 Optimizely scripts and found the malicious code in 188 of them, across 178 organizations, and said the real number is likely higher. The researchers reported that the code was built to steal payment card details and cookies and to swap cryptocurrency payment addresses.

The response was quick, the notice less so

Optimizely says it detected the activity the same day, blocked delivery of the code and closed the flaw within three hours of detection, then removed the injected code from every affected project and contacted every Web Experimentation customer. Report URI published its findings at 5:47 p.m. Eastern on the Monday. Optimizely's status page first acknowledged the incident at 8:02 p.m., and a fuller statement followed on the Wednesday.

Why this is a buying question, not just an IT one

It would be easy to file this under security and move on. We think that misses who made the decision that mattered. The script was on those pages because a marketing team bought a testing tool, and the terms on which it runs were set, or not set, at purchase.

The experience layer is sold as scripts

Experimentation and personalization are the features that separate a suite from a plain content system, and much of that layer reaches the browser as vendor-hosted JavaScript. Adobe Target, Sitecore Personalize and the HubSpot tracking code are all delivered that way in common setups, as is Optimizely's. The more of a suite you switch on, the more of this code your pages carry.

Certifications cover the vendor, not your page

A SOC 2 report or an ISO 27001 certificate tells you how a vendor runs its own systems. Neither tells you what a script from that vendor is allowed to do once it is on your site. Report URI found that 68 of the organizations it tested afterwards had no Content Security Policy at all, so nothing on their side could have limited the damage.

You can outsource the tool, not the risk

An Optimizely partner quoted by CMSWire put it this way: you can outsource the technology but not the risk. The visitors whose card details were exposed were customers of the websites, not of Optimizely, and the duty to tell them falls on the brand whose page they were on.

Questions to put to any vendor whose code runs on your site

These apply to Optimizely customers this week and to everyone else at the next renewal. None of them needs a security background to ask, and a vendor that cannot answer them clearly has told you something.

Can one account put code on every page?

Ask whether custom code fields can be locked, restricted to named people or put behind an approval step. If the answer is that any editor can publish JavaScript site-wide, ask what it would take to change that.

How would we know code had changed?

Ask whether the product can alert you when project-level code is edited, and whether you can pin or review the version that is live. Customers hit on 16 November were told to review their Project JavaScript by hand, and to rotate the credentials of every user, API token and integration with write access.

Can we host it or serve it another way?

Some testing and personalization products can run on the server or at the edge, or let you host the script yourself, so the browser never fetches live code from a third party. That costs convenience. Ask what the options are and what you lose with each.

What does the contract say about incidents?

Ask for the notification window in hours, a named contact, and a commitment to a written root-cause report. Then check who in your own company receives that notice, because the person who signed the order form has often moved on.

Which pages truly need it?

Ask your agency for a list of every third-party script by page. Take testing and marketing tags off payment and login pages unless there is a reason you can name, and have a strict Content Security Policy written for the rest.

Platforms in this piece

Frequently asked questions

What happened in the Optimizely security incident?

On 16 November 2026 an attacker exploited a flaw in the Optimizely Web Experimentation signup flow and inserted malicious JavaScript into customer projects. Websites that loaded the Optimizely script then served that code to their visitors. Report URI found it on sites belonging to 178 organizations. Optimizely says it closed the flaw within three hours of detecting it.

Was Optimizely CMS affected by the hack?

No. Optimizely says the incident was limited to Web Experimentation, its A/B testing product, and that no other Optimizely product was affected. That means Optimizely SaaS CMS and Optimizely PaaS DXP were not involved. A site using both the CMS and Web Experimentation should still check its experimentation projects.

Which companies were affected by the Optimizely malware?

Report URI named Zoom, the BBC, CNN, Salesforce, GitLab, IKEA, HSBC, Equifax, Air France-KLM, Ledger and Peacock among 178 organizations whose visitors were served the code. Optimizely has not published a list. It says fewer than 5 percent of its customers were affected and that all of them have been contacted.

What should Optimizely Web Experimentation customers do now?

Review the Project JavaScript and any custom experiment code in every project for anything you did not write. Review which users, API tokens and integrations have write access and rotate their credentials. If your site takes payments, treat card entries made on 16 November as potentially exposed and involve your security and legal teams.

Would a Content Security Policy have stopped the Optimizely attack?

A strict one would have limited it. A Content Security Policy tells the browser which hosts scripts may load from and where data may be sent, and the malicious code fetched its payload from a domain that was not Optimizely's. Report URI found that 68 of the organizations it tested had no policy at all.

About this editorial

This is a position our principal analysts took on November 20, 2026. It is not updated: when the view changes, we publish a new editorial. The scorecards linked above carry the current ratings for every platform named here.

Scorecards and platform inclusion are unsponsored: no vendor pays to be named in an editorial or to change a rating. Disagree with us? Tell us.

Still deciding?

Answer seven questions for a shortlist built around your team, or put two of the platforms named here side by side.