The PCI Council has stated that it is reexamining its four Self-Assessment Questionnaires (SAQs). Level 2 through Level 4 Merchants use an SAQ to validate their compliance. Although I have no insight into what possible changes or even new versions might come from the Council, I have some suggested changes that reflect both current attack vectors and real-world business practices.
Even though QSAs spend a lot of time preparing Reports on Compliance (ROCs) for their Level 1 Merchants and Service Providers, most of us also work with many merchants who self-assess their compliance. Interestingly, some of these merchants can be quite large organizations but with only limited payment card activity. For these merchants, the PCI Council developed three simplified SAQs that are based on how they process card transactions.
Merchants use SAQ A if they outsource their card processing to a compliant third-party service provider. The requirements to use this SAQ are: You have card-not-present (i.e., MOTO, E-Commerce) transactions exclusively; you do not store, process or transmit any cardholder data yourself; your service provider is PCI compliant; any cardholder data is only on paper; and you store no cardholder data electronically.
The general model of an SAQ A merchant is one with a Web site that links to a hosted order page at a secure third party. Cardholders are transferred to that secure hosted page to enter their card data and then returned to the merchant's site after the transaction is approved or rejected.
In my perfect world, I would like to see two changes to this SAQ. First, I would like the SAQ to stipulate that the service provider is not just PCI compliant (as a service provider, of course; not as a merchant) but that it is a Level 1 Service Provider. My reasoning is that SAQ A merchants depend on their service providers. I want those service providers to have an outside assessment of their compliance.
My second suggestion deals with the merchant's own Web server. I would like it to be scanned for external vulnerabilities, and I want the code redirecting the customer to the hosted order page to be inspected. Each of these tests would be done quarterly. The basis for this suggested change is an excellent Data Security Alert posted by Visa Europe. In that alert, Visa describes attacks on SAQ A merchants where the bad guys have hacked the merchant's server and installed their own code to redirect customers to their criminal site instead of the service provider's site. I would change SAQ A to make this attack less likely to succeed.
I would also like to see a note in the instructions to SAQ A to address mail order/telephone order (MOTO) transactions. It should say merchants cannot do MOTO and still qualify for SAQ A. For example, if the customer can't get to the Web or has some other problem, the merchant cannot use its computer to go on the Web and enter that customer's card number for him or her. In addition, the merchant just turned its workstation into a payment terminal, and heaven knows what has just been done to its PCI scope. I see this situation so often I coined a name for it: SAQ A OMG.
I regret having to make life more complicated for these merchants. But we all need to remember that although you can outsource your processing, you cannot outsource your responsibility.
SAQ B presents a different challenge. Merchants use SAQ B when they either have an imprinter (a.k.a., knuckle-buster or zip-zap machine) or use POS terminals "connected via a phone line" to their acquirer or processor. That terminal cannot be connected to the Internet or to any other system in the merchant's environment. Like SAQ A, the merchant retains only paper records and does not store electronic cardholder data.For SAQ B I would like some clarification for the merchants. For example, terminals that "dial out" with an IP line use the Internet. Therefore, in my opinion, they are ineligible for this SAQ (see SAQ C, below). I would like this fact spelled out explicitly in the SAQ instructions, because many small merchants may not understand this distinction.
My other clarification deals with the "terminal data capture" situation where the terminals retain cardholder data after authorization. At a prescribed time, an employee "batches out" the terminal and transmits a file with the day's (or weekend's) card transactions to the acquirer for processing. To this QSA, the POS device--and, therefore, the merchant--is storing data electronically. Therefore, merchants with terminal capture would not qualify for a simplified SAQ of any kind under the present guidelines.
We have to understand that the decision on which SAQ a merchant should use is not up to the QSA; the acquirer makes that decision. I had an acquirer overrule me on this exact SAQ B situation--where the POS terminals were storing the card data--and I was fine with that decision.
In my heart of hearts, I agreed with the acquirer because the risk was pretty small. But as a QSA, I need to follow the letter of the PCI law and look out for my client (no matter how much they had trouble understanding it at the time). I think the people who developed SAQ B may well have intended to include terminal capture devices. Either way, a brief clarification on the SAQ instructions to merchants would be welcome.
SAQ C may be most in need of an overhaul. It was designed for merchants that have a payment application connected to the Internet, generally to authorize card transactions. The previous prohibition on storing electronic cardholder data applies to SAQ C merchants, but there is one particularly awkward requirement: "The payment application/Internet device is not connected to any other systems within your environment."
The reason I call this "not connected to any other system" requirement awkward is that it almost never exists in the real world. Payment applications need to connect to other systems for antivirus and patch updates; they also need to connect to inventory and pricing systems. I can have a merchant segment the payment application behind a firewall, but how many holes does it have to punch in that firewall to allow for the other business services?
My experience is that almost every call center, and a few zillion retailers, thinks it qualifies for SAQ C. In reality, only a very small number do. Reading the instructions carefully, about the only situation where it applies is for those POS terminal merchants with an IP line for authorizations. I think the people who created SAQ C intended more.
The creators of the three simplified SAQs wanted to help merchants become compliant. To a great extent, they succeeded, and for that we should all thank them. The PCI Council and card brand staff who worked on these SAQs did a tremendous amount of work. Before SAQs A, B and C, there was only SAQ D with its 226 questions covering all of the DSS. And not all of these questions were relevant to a large number of merchants. In fact, that original, overwhelming SAQ probably scared some of these same merchants away from PCI compliance.
My hope is that as the PCI Council takes a fresh look and reflects on the merchant feedback, it will consider these ideas as well as the many others I'm sure it has received. It is neither wise nor realistic to go back to the "one SAQ fits all" approach. A few tweaks of SAQs A and B will go a long way; SAQ C is likely to need a more thorough overhaul.
Are you a Participating Organization on the PCI Council? You should be, by the way. And if you are, have you sent any suggestions for improvements on the SAQs? What were they? I'd like to hear your thoughts. Either leave a comment or E-mail me at [email protected].