PCI DSS 11.6.1 explained in a sentence: it asks for one thing, a mechanism that spots unauthorised change to the HTTP headers and script content of your payment page, as that page arrives in the shopper’s browser, and tells a human about it. That is the whole of it. It is a detection requirement, not a prevention one. It does not ask you to block anything, and it is not satisfied by file integrity monitoring on your web server.
Most coverage of this requirement either paraphrases the standard’s own wording back at you or ends in a vendor shortlist. Useful for a procurement exercise, wrong audience for a merchant with a checkout, a developer and no budget line for a monitoring platform. What follows is PCI DSS 11.6.1 explained from the browser’s point of view: what it is designed to catch, why headers are in there at all, where the wording is deliberately loose, and what an acquirer or a QSA will accept as evidence.
What PCI DSS 11.6.1 actually says, in plain English
The requirement calls for a change and tamper detection mechanism that alerts personnel to unauthorised modification, including additions and deletions, of the HTTP headers and the script contents of payment pages “as received by the consumer browser”. The mechanism has to evaluate those received headers and pages at least weekly, or at a frequency you set through a targeted risk analysis performed to the elements in Requirement 12.3.1.
It became mandatory on 31 March 2025. Before that date it sat as a future-dated best practice through the PCI DSS v4.0 transition, which is why a lot of merchants met it for the first time at their 2025 assessment and were surprised. The current text lives in PCI DSS v4.0.1, published in June 2024 and available from the PCI Security Standards Council document library.
Why “as received by the consumer browser”is the clause that does the work
Server-side file integrity monitoring watches your filesystem. A skimmer injected through a third-party script, a compromised supplier CDN, or a tag inserted in a tag manager container never touches your filesystem at all. Your FIM report stays clean. The shopper’s browser still executes the skimmer, still posts a copy of the PAN, expiry and CVV to an attacker’s host, and still renders a perfectly normal order confirmation.
So a clean FIM report proves something about your server and nothing about the page. The assessment question is simple: can you show what the browser actually received, and can you show how you would know if it changed? If the answer only describes your origin, you have documented the wrong layer.
Weekly, or a frequency you can defend
Weekly is the default. Anything less frequent needs a targeted risk analysis under 12.3.1, which means naming the asset you are protecting, the threat the requirement addresses, the factors that make an attack more or less likely on your estate, the frequency those factors justify, documented management approval and a review at least every twelve months. A paragraph saying “we are low risk”is not that. Most merchants find it cheaper to run the check weekly than to write and maintain a defensible analysis for running it monthly.
Who 11.6.1 applies to, including SAQ A merchants
This is where the advice in circulation has gone stale, so read the dates carefully.
The SAQ A position changed in January 2025
SAQ A under PCI DSS v4.0 included both 6.4.3 and 11.6.1. In January 2025 the Council published a revised SAQ A for v4.0.1 that removed those two requirements from the form and added an eligibility criterion instead: the merchant confirms that its site is not susceptible to attacks from scripts that could affect the merchant’s ecommerce systems. The obligation did not evaporate. It moved from a control you evidence to an eligibility statement you sign.
Read plainly, that is the death of the easy answer. You still have to know what runs on the page in order to confirm that sentence honestly, and your acquirer can ask you how you reached it. Several UK acquirers are asking exactly that. If you are working out which form applies to you, start with the SAQ A eligibility criteria for UK merchants and confirm the version your acquirer is accepting at the time you submit, because the forms have moved twice inside two years.
Hosted fields and redirects change the picture, not always the obligation
If card data is entered inside an iframe served by your payment service provider, or on the PSP’s own page after a redirect, the card details never touch your DOM. That removes most of the skimming risk and is the single cheapest control available. It does not remove your page from the attack path. An attacker with write access to your checkout can swap the iframe’s source, overlay a fake field on top of it, or change the redirect target. Nothing in the hosted model prevents that, which is why your containing page still deserves the check even when the form is not yours.
Your PSP’s compliance is not your compliance
A PSP’s AOC covers the PSP’s systems. It says nothing about the eight other scripts you load on the same page: analytics, chat widget, consent banner, A/B testing, affiliate pixel, review widget, fraud fingerprinting, a font. Those are yours. If you merge the two in your documentation, an assessor will separate them for you, and not gently.
How 11.6.1 differs from 6.4.3
Merchants conflate these constantly and then document the wrong thing.
6.4.3 is about authorisation and inventory. Every script loaded and executed in the shopper’s browser on a payment page must be authorised, its integrity assured, and held in an inventory with written business justification. It is the question: should this be here?
11.6.1 is about detection. It asks whether you would notice something changing without permission, scripts and headers alike, within a week. It is the question: did something change?
One artefact can feed both, if you build it properly. A checkout script allowlist that records the script URL, what it does, the business justification, the owner, the date it was authorised and the date it was last reviewed satisfies the inventory half of 6.4.3 and gives your weekly diff something to compare against. Merchants who keep a bare list of domains end up rebuilding the whole thing under assessment pressure, usually in the week before the deadline. The detail of how the two requirements interlock is set out in this breakdown of what PCI DSS 4.0 requirements 6.4.3 and 11.6.1 ask of your payment page.
Document only one and the gap shows immediately. An inventory with no detection mechanism means you know what should be there and would never learn when it was not. A detection mechanism with no inventory means your alerts have no baseline, so every change looks equally plausible and nobody can say whether the new domain was approved last March.
What the requirement is really trying to catch
Three routes put malicious code on a payment page, and 11.6.1 is aimed at all three.
- A compromised third-party supplier. Ticketmaster UK, June 2018: a chatbot script from a supplier was altered at source and served onto payment pages. The ICO fined Ticketmaster UK £1.25m in November 2020. The domain was allowed. The content was not what it had been. Domain allowlisting alone would not have raised a flicker.
- First-party code altered after access. British Airways, 2018: a modified script on the payment path, with card data skimmed from 21 August to 5 September 2018 before it was found. The ICO issued a £20m penalty in October 2020. A weekly browser-side diff is built precisely for this shape of change.
- An unpatched or end-of-life platform. Old Magento installations and unmaintained extensions remain a standing route in, because the injection is cheap once a known exploit exists.
Why HTTP headers are in the requirement
This half gets ignored almost universally, and it is the half that protects everything else. An attacker who can weaken your Content-Security-Policy header, or strip it, or add a header-level redirect, has silently removed your main control. Nothing visibly breaks. The page loads, the order goes through, and the policy that was supposed to stop a new exfiltration host no longer exists. Snapshot your response headers in the same weekly check as your script diff and compare them line by line, including the policy string itself, not just whether the header is present.
The things a file check never sees
Modern skimmers are selective. Plenty only fire on the payment step, or for a fraction of sessions, or not when developer tools are open, or not for visitors from certain networks. A loader may pull its payload from somewhere that is not a file at all: tag manager containers, or data read from a public blockchain, as seen in EtherHiding-style skimmer activity, where the malicious code sits in a smart contract and the page just fetches it.
Google Tag Manager is the documented blind spot nobody puts in their scope diagram. Anyone with publish rights in the container can push a tag that executes inside an origin you already allow. CSP will not stop it, because the origin is permitted. SRI will not stop it, because the injected tag was never hashed. A container permissions review belongs in your 11.6.1 routine, every bit as much as the script diff.
Mechanisms that satisfy 11.6.1, and where each goes blind
Content Security Policy with a reporting endpoint
Roll out in report-only mode, collect violations at an endpoint, then enforce. CSP is strong at one thing: telling you when the page tries to load from or send to a host you never approved. That is exactly the signal a new exfiltration domain produces. It is blind to a change inside a script you already allow, which covers both the Ticketmaster and British Airways shapes above. Treat it as half the answer. There is a tested configuration in this guide to Content Security Policy for a checkout page.
Subresource Integrity hash pinning
The most precise control available for pinned third-party scripts, and the one with the real maintenance cost. When a supplier edits their own file without telling you, the hash mismatch blocks the script, and if that script is in the payment flow your checkout breaks in public. Agree a change notification process with the supplier before you enforce SRI on anything in the payment path. The mechanics, including which scripts are safe candidates, are covered in this piece on pinning the scripts you cannot control.
A scheduled synthetic checkout run
The cheapest mechanism that genuinely meets the “as received by the consumer browser”test. A headless browser loads your checkout the way a shopper does, you store every script src value and the full set of response headers, and you diff this week’s file against last week’s. Any difference raises a ticket. Budget roughly 15 to 30 minutes of someone’s week once it is running, plus a morning to build it.
Outside signals
Your acquirer’s common point of purchase reporting, a CPP letter, is a detection channel too, though a late one. By the time card issuers have triangulated your checkout as the common point, the skimmer has been collecting for a while. Keep it in the picture as corroboration, never as your mechanism.
“Alerts personnel”is the clause people under-read
A CSP report endpoint quietly filling up a bucket nobody opens is not a detection mechanism. The requirement says personnel are alerted. Name the person, name the deputy, define where the alert lands, define what triage looks like and what gets escalated. That trail is not administrative overhead on top of the control. Under assessment, that trail is the control.
Building evidence an assessor will accept
Four artefacts carry almost all the weight, and none of them needs a purchase order.
- A dated script inventory. Script URL, what it does, business justification, owner, authorisation date, last-reviewed date. Dates in the future or all identical are a tell, and assessors read dates first.
- Alert samples with triage notes. Two or three real examples beat a policy document. Show the alert, the ticket, who looked at it, what they concluded, when it closed. A false positive handled properly is excellent evidence.
- Your targeted risk analysis, if you check less often than weekly. Asset, threat, likelihood and impact factors, the frequency those justify, approval, annual review date.
- The routine itself, written down. Who runs the check, on what day, what they compare, where the output is stored, how long it is retained. One page is enough.
Store the weekly output somewhere immutable enough to prove it was not written the night before the assessment. A version-controlled repository with commit timestamps does this for free.
The short version
11.6.1 is detection from the browser’s side, covering scripts and headers, at least weekly, with a named human receiving the alert. 6.4.3 is authorisation and inventory, and one well-structured allowlist feeds both. Server-side file integrity monitoring on its own does not satisfy it. CSP catches new domains but not changed content; SRI catches changed content but breaks when suppliers edit files; a headless-browser diff of script tags and response headers catches both and costs you half an hour a week. If you take one thing from this explanation of PCI DSS 11.6.1, make it the headers: that is the control an attacker removes first, quietly, and the one nobody checks.
Frequently Asked Questions
What does PCI DSS 11.6.1 require in simple terms?
It requires a mechanism that detects unauthorised changes to the HTTP headers and script content of your payment pages as they are received by the shopper’s browser, and alerts a person when something changes. Checks run at least weekly, or at a frequency justified by a targeted risk analysis under Requirement 12.3.1. It has been mandatory since 31 March 2025.
Does PCI DSS 11.6.1 apply to SAQ A merchants?
It did under the PCI DSS v4.0 version of SAQ A. In January 2025 the Council published a revised SAQ A for v4.0.1 that removed 6.4.3 and 11.6.1 from the form and added an eligibility criterion requiring the merchant to confirm its site is not susceptible to script-based attacks. Confirm with your acquirer which version they are accepting, because some still ask how you reached that confirmation.
How often do payment page scripts and headers need checking?
At least weekly by default. A different frequency is permitted only where a documented targeted risk analysis supports it, including the asset, the threat, the contributing factors, management approval and an annual review. In practice, running a weekly automated diff is usually less work than maintaining that analysis.
What is the difference between PCI DSS 6.4.3 and 11.6.1?
6.4.3 is about authorising and inventorying every script that executes on a payment page, with written business justification. 11.6.1 is about noticing when scripts or headers change without authorisation. One records what should be there; the other tells you when reality diverged from that record.
Can Content Security Policy alone satisfy PCI DSS 11.6.1?
Not on its own. CSP is a strong signal when the page contacts a host you never approved, but it cannot see a malicious change inside a script served from a domain you already allow, and it will not stop a tag published through an allowed tag manager container. Pair it with a content-level check, such as a weekly rendered-page diff, and include the policy header itself in what you compare.
