Tech Support Phone Payment PCI: Safer Ways to Take Cards
Industry Payment Guides

Taking Cards by Phone for Tech Support: PCI and Safer Options

Tech support phone payment PCI rules: how pause-and-resume recording, keypad masking and payment links keep card numbers off agents and call recordings.

Concerned woman on a mobile phone call beside a smiling support agent wearing a headset, on a white background

Photo: Pixabay

On this page 6 sections
  1. Why Recorded Calls and Card Numbers Collide Under PCI DSS
  2. Pause-and-Resume Recording: Manual vs. Automated
  3. IVR and DTMF Masking: Keeping Card Digits off Agents' Screens
  4. Texting a Payment Link Mid-Call
  5. Tech Support Phone Payment PCI Scope for Remote Agents
  6. Which SAQ Applies to Your Setup

Key takeaways

  • A recording, transcript or screen capture that holds a spoken card number is stored cardholder data. A captured security code is the more serious problem because PCI DSS limits its storage more tightly than the card number.
  • Pause-and-resume recording cleans up the audio but leaves agents and their desktops in scope. Keypad masking and payment links take the digits away from your team entirely.
  • In tech support, a customer who types card details on a computer your agent is remotely controlling exposes the number. Move payment to the customer's phone or end the session first.
  • Your acquirer decides which self-assessment questionnaire you file, and keeping card data away from agents is usually the biggest lever for a shorter one.

If your agents hear card numbers on recorded calls, every system that touches those calls falls inside your PCI DSS scope. That scope covers the recordings themselves and the call platform that stores them. It also covers any transcripts and the agents' computers. The most reliable fix for tech support phone payment PCI problems is to make sure agents never hear the digits at all.

You have three practical ways to get there. Pausing the recording while the customer reads the card is the cheapest option and the weakest one. Having customers key their digits into an IVR or a masked keypad line removes far more risk, and texting them a link to a hosted payment page does the same.

It helps to know who sets these rules. PCI DSS is written by the PCI Security Standards Council, an industry body the major card brands created, and it is not federal law. You're bound to it by contract: the card networks require compliance through the merchant agreement you signed with your acquirer, and your acquirer is the party that asks you for proof. If a breach happens, state breach-notification laws can add obligations on top of the contract fallout.

This article is general information. Your acquirer and a qualified security assessor (QSA) have the final word on your setup.

Why Recorded Calls and Card Numbers Collide Under PCI DSS

Support companies record calls to train agents and, more importantly, to prove what a customer agreed to. Take an illustrative case in which a customer later claims they never agreed to a $249 annual plan. A recording of them saying yes is the evidence you send to the issuer, which is why our guide to winning "I was scammed" tech support chargebacks leans so heavily on recorded consent.

Trouble starts when that same audio file also holds a customer reading a card number aloud. At that point the file is stored cardholder data. PCI DSS expects stored card data to be protected and kept no longer than you need it. Meeting that bar gets hard once copies of one call exist in the call platform, its backups, your QA scoring tool and perhaps a vendor's analytics service.

A captured security code is worse. PCI DSS treats the three- or four-digit code far more strictly than the card number, and encrypting the recording doesn't fix that. Any recording that captures the code should be handled as a serious finding. Read the storage requirements in the current standard on the PCI Security Standards Council's website, or ask your QSA how they apply to your call archive.

Two newer habits make this worse, and both are common in support shops. Platforms that transcribe calls or write AI summaries can turn a spoken card number into a searchable 16-digit string, and that string tends to travel into CRM notes and tickets. Remote sessions add a second leak. If the customer types card details into a payment page while your agent is watching or controlling their screen, the agent sees the number, and so does any session recording.

Pause-and-Resume Recording: Manual vs. Automated

Pause-and-resume is where most companies start, because it works with the phone system you already have. There are two versions.

With a manual pause, the agent clicks a button to stop recording before asking for the card, then clicks again afterward. It fails in predictable ways. Agents forget late in a busy shift. Customers sometimes start reciting the number before the agent has clicked, or an agent resumes while the customer is still giving the expiration date. Then there's the caller who circles back five minutes later ("wait, was that 4-4-1?") after recording is on again. If you rely on manual pause, spot-check recordings on a schedule so you can show it works.

An automated pause ties recording to what the agent does on screen. When the agent opens the payment form in your virtual terminal or CRM, recording stops, and it resumes when the form closes. That removes most human error. The catch is that it needs an integration between your call platform and your payment screen, and it only protects you if screen recording pauses at the same moment as the audio.

Both versions share one limitation. Your agent still hears the card number, and the agent's desktop and phone still handle it. Recordings get cleaner while your people and their equipment stay in scope.

Pausing also carries a dispute cost, since the gap in the recording falls exactly where the payment happens. Close it with a script step. Once recording resumes, the agent restates the amount and the renewal terms, and the customer confirms on tape.

IVR and DTMF Masking: Keeping Card Digits off Agents' Screens

An IVR (interactive voice response) payment setup has customers enter card details on the phone keypad instead of saying them out loud. Tech support teams generally use one of two versions.

In a transfer to an IVR, the agent hands the call to an automated line. Once the customer keys in the card, the call either returns to the agent or ends, and the agent never hears a digit. For tech support, the weak point is the handoff. People who called because technology confuses them can get lost in an automated menu and hang up halfway through. If many of your callers are older, test this carefully, because payment friction and family disputes often go together, as our article on refund policies for senior customers explains.

DTMF masking keeps the customer on the line with the agent. DTMF (dual-tone multi-frequency) is the set of tones a phone keypad makes. A masking service sits in the call path and replaces each tone with a flat, identical one, so neither the agent nor the recording can tell which keys were pressed. On screen, the agent watches asterisks fill in and can coach the customer ("now the expiration date"), while the real digits go straight to the payment provider.

How much masking shrinks your scope depends on where the tones get captured. If the masking provider intercepts them before they reach your phone system, your recordings and agent desktops may stay out of the card data flow. If raw tones pass through your systems first, those systems remain in scope. Ask every vendor exactly where capture happens and get the answer in writing.

Pricing models differ. Some masking services charge by the minute or by the transaction, and others bill a platform fee, so get quotes before you commit. Plan for one practical snag as well: customers on mobile phones have to bring up the keypad mid-call, and some will need to be walked through it.

Pay by link over the phone is simple in practice. Your agent sends a text or email with a link to a hosted payment page, usually with the amount and description filled in. The customer enters the card on their own phone, and the agent sees a confirmation when it goes through. At no point does the card number pass through your phone system or your agents' desktops. Not every gateway offers this, so ask your provider whether it supports hosted payment links and how they're delivered.

Links bring a couple of side benefits. Because the customer completes a web checkout, the page can run 3D Secure authentication, which may help with certain fraud-related disputes; ask your processor how that applies to your account. Customers also get a written receipt, which helps when the cardholder and the person on the phone turn out to be different family members.

This method has a particular problem in tech support. Scammers in this space use texted links constantly, and many of your callers have been warned never to click one. You can lower that suspicion. Send from a number or sender name the customer will recognize, with a link domain that matches your business name. Before anything goes out, have the agent say exactly what's coming: the amount, and that it's a payment page for your company. Set links to expire, and never ask a customer to read back a code from a text.

One more rule matters in this industry. Don't let the customer open the link on the computer your agent is controlling. Ask them to pay on their phone, or end the remote session before they pay on the computer. Otherwise the agent sees the card number on screen, and the link protects nothing. If you're unsure whether to send it before the session or after the fix, see charging before or after a remote support session.

Tech Support Phone Payment PCI Scope for Remote Agents

Many support teams work from home, and that changes the math. If remote agents hear card numbers and key them into a virtual terminal, each home setup joins your card environment. That reaches past the laptop and softphone to the home network, the notepad by the keyboard and the personal phone that could photograph a screen. A PCI compliance program for a call center with agents in twelve living rooms is a much bigger job than one for a single locked office.

If agents must handle card numbers, the usual controls apply. Agents should work only on company-managed devices, with desktops locked down so payment screens can't be screenshotted or copied. Paper has to go, because a clean-desk rule is the only reliable way to keep card numbers off sticky notes. You also need logs showing who accessed which payment. Insider risk is real: an agent who hears forty card numbers a day is in a position to misuse them.

A simpler answer is to take card data out of home offices entirely. With DTMF masking handled in the provider's network, or with payment links, a remote agent's setup mostly drops out of the card data flow. You still need good security for customer accounts and remote-access tools, but you're no longer protecting card numbers in kitchens.

Which SAQ Applies to Your Setup

Most merchants prove compliance with a self-assessment questionnaire (SAQ). Your acquirer tells you which one to complete, or whether your volume requires a formal report from an assessor.

Eligibility for each SAQ type depends on how card data enters your business and which systems touch it, and the criteria change between versions of the standard. The table below shows only the general direction. Confirm the specifics against the current SAQ documents on the PCI Security Standards Council's website and with your acquirer.

Your setupWhat your team touchesGeneral direction
Agents hear the card and key it into a virtual terminal on a dedicated, isolated computerSpoken digits, one computerA shorter questionnaire may be available if strict isolation criteria are met
Agents key cards on networked PCs, calls recorded, softphones in useDigits, desktops, network, recordingsUsually the longest questionnaire
DTMF masking where the provider captures digits before your systemsMasked tones onlyPotentially much smaller, depending on architecture
Payment links to a fully hosted pageNothingOften among the shortest, if card handling is fully outsourced

Here's an illustrative example; the company and its setup are assumed, not a real merchant. A support company has 12 agents, three of them remote. Its agents key cards into a virtual terminal on general-purpose laptops, and every call is recorded with AI transcription. Its card environment therefore includes all 12 laptops, three home networks, the call platform, every recording and transcript, and the QA tool.

Now suppose that company switches to payment links and keeps recording on for the whole call. Recordings and transcripts become clean, and the laptops stop handling card numbers. The questionnaire its acquirer asks for could shrink a lot. Dispute evidence improves too, because nothing has to be paused.

Tech support is already treated as a high-risk merchant account category, and underwriters look closely at how you handle customers on the phone. A clean payment process works in your favor, and our guide to tech support merchant account approvals covers the rest of that review.

Start with an audit you can run this week:

  1. Pull five recent calls that included a payment and listen to the payment portion.
  2. Search transcripts and ticket notes for long digit strings.
  3. Check whether your QA tool or any analytics vendor keeps its own copies of calls.
  4. If card numbers turn up anywhere, ask a QSA how to find and purge the archive before you change your process going forward.

After that, map how card data should enter your business from now on. If agents will keep keying cards, you can request a free quote from MannyPay to review virtual terminal setups and payment gateways such as Authorize.net and NMI that fit your call flow.

About the author

Marcus Bennett

Marcus Bennett is Payments Risk & Compliance Analyst for Beyond the Swipe, covering high-risk processing, chargebacks, rolling reserves, frozen merchant accounts and fraud prevention.

This article is for general educational purposes and is not legal, tax or financial advice. Card network rules and state laws change; confirm current requirements for your business.

Related reading