100 Payment Processing Questions: What ChatGPT Recommends

The payment processor ChatGPT recommended most often was Stripe, in 68 of 100 answers, and Stripe was its own pick in 54 of the 100. In short: one default, then a rate card.

Stripe is the default, and it is the only one

Stripe stands alone as the default: it appears in 70 of 100 answers, is recommended in 68 and picked in 54, 77 percent of its appearances. Square is next at 55 appearances, 53 recommendations and 34 picks, then PayPal at 41, 39 and 22. PayPal is named in 51 answers, but 10 of those mention it only in passing, and when it is offered as an option it is picked in about half (22 of 41). Behind the three sit Helcim (29 appears, 28 recommended, 16 picked), Adyen (25, 25, 19), Braintree (17, 17, 9), Shopify Payments (16, 16, 8) and Paddle (12, 12, 10). Venmo is named in 28 answers and Apple Pay in 8, and every one of those mentions is coded passing: a way to pay, not a processor to choose. They are in the dictionary because they were listed before collection, and they show how much of the model's payment vocabulary is wallets rather than processors.

When the buyer asks outright, the default holds. Nineteen of 20 recommendation-intent queries recommended a dictionary brand and 18 picked one. Stripe was among the picks on 15 of the 20, Square on 8, PayPal on 7, Paddle on 4, Helcim and Adyen on 3 each. Three answers picked a single brand: Stripe for the marketing agency and for the solo consultant, Square for selling at farmers markets. The two with no pick are the law firm, which left the set entirely (below), and "best value for money", which recommends Helcim, Square and Stripe for different cases and then asks for monthly card sales and average transaction before choosing.

Eighty-four answers ask for your numbers

Eighty-four of 100 answers ask the buyer for their own figures, typically monthly card volume, average transaction size or typical invoice, and 69 of those offer to work out what the options would cost. This is the mechanism the earlier categories did not have. Sixty-three answers quote at least one published rate in the form "2.9% + $0.30", and 54 cite a vendor's own pricing or fees page. Category answers recommend 4.2 brands on average but pick only 2.7, and 22 of the 100 answers pick nothing at all, twice as many as any other category so far (CRM, with 11, is next). In a fifth of the answers the model is not choosing a processor. It quotes the rate card and holds the decision until it has the inputs.

Head-to-heads: 28 forks when it describes, 19 when it chooses

The model forks on most head-to-heads when it describes the options, and on fewer when it has to choose. Twenty-eight of 30 comparison queries recommended every brand named in the query; the other two, Square versus Clover and Payment Depot versus Helcim, list both brands without assigning either a fit. The picked level is where this category splits from the others. Only 19 of the 30 picked every named brand. Two picked one side: Stripe over Braintree for a SaaS company, and Square and Toast over Clover. Nine picked nobody, among them Stripe versus PayPal, Square versus PayPal, Stripe versus Checkout.com and "Stripe or PayPal for an online store": each describes who each side suits without telling the buyer which to choose. All 16 of Stripe's comparison recommendations are on queries that name Stripe, and 12 of them are picks; like Mailchimp in email marketing, it is chosen as a side of the fork, never imported into somebody else's comparison.

The lane is stable, and it is stated in the model's words

Every brand behind Stripe gets a stable lane, and the model states it in one repeatable sentence. Square is "in-person, POS, the free plan": it is picked for farmers markets, retail stores, restaurants and the medical practice. Helcim is "interchange-plus as volume grows": 28 of the 29 answers that name Helcim discuss interchange pricing, and it is recommended in 28 of those 29. Paddle and Lemon Squeezy are "merchant of record for software that does not want a tax team". Adyen and Checkout.com are "enterprise, global, high volume". Toast is restaurants. The lane is the reason a brand is named at all, and in a category where the model defers the final choice to a cost calculation, owning the sentence that puts you in the calculation is the position that matters.

Vertical prompts re-rank the field, and two of them replace it

Most vertical prompts keep the general brands and re-rank them: restaurants get Toast, Square, Clover and Lightspeed; medical practices get Square, Stax and Chase Payment Solutions, with InstaMed named for EHR integration; nonprofits get Stripe and PayPal; real estate agents get Square, Stripe, Helcim and QuickBooks Payments. High-risk businesses stay in the dictionary but in a corner of it, with Durango Merchant Services, PaymentCloud and Worldpay recommended and none picked. Two prompts swapped the field out. Both law firm queries recommended LawPay and Clio Payments, neither in the dictionary; the category answer shortlists them "rather than generic processors like Stripe or Square", on trust-account grounds. The open-source query recommended Kill Bill, Lago and Hyperswitch and pointed out that a fully self-hosted Stripe would still need a processor behind it. Five answers in total recommended no dictionary brand: the two law firm answers, the open-source answer, and the two head-to-heads that list both brands without a fit.

Ranking in Google is a different game, and Adyen is the proof

A brand's Google ranking differs from its standing in the answer, and Adyen shows how much. Of the 360 recommended brand mentions, 294 were for brands whose own website does not rank in Google's top ten organic results for that query, 82 percent. Of the 222 picked mentions, 170 were, 77 percent. Google's top ten for these queries is 79 percent third-party pages (787 of 1,000 results). Adyen is recommended 25 times and picked 19, and its site is in the top ten for none of those queries. PayPal is recommended 39 times and ranks for 2. Stripe is recommended 68 times and ranks for 22.

Vendors are the sources, and NerdWallet is the referee

Vendor pages make up most of the sources, and the main outside voice belongs to NerdWallet. Across 100 answers there were 301 citation events to 98 domains, one per cited domain per answer. Fifty-three percent (161) went to vendors' own sites, and 47 answers cited nothing but vendor pages; only project management (60 percent, 50 answers) leaned harder on vendors. The most-cited domains are stripe.com (42), squareup.com (28), nerdwallet.com (25), paypal.com (18), adyen.com (15) and helcim.com (13). NerdWallet is the only third-party site cited more than nine times; TechnologyAdvice (9), Fit Small Business (6) and myPayAdvisor (6) follow. The ten most-cited domains cover 58 percent of citation events. Ninety-eight of the 301 citation events (33 percent) involved a domain that also sat in Google's top ten for that query. Reddit and Wikipedia have zero citations in this sample, for the sixth teardown running.

The table sets ChatGPT's payment processing answers beside the other categories tested earlier in the series, side by side under the same rulebook.

What ChatGPT's recommendations mean for payment processing brands comes down to four things, in order.

Publish the rate card where the model can read it. Fifty-four answers cite a vendor's own pricing or fees page, and the model quotes the number it finds there. A processor whose pricing sits behind a quote form is priced from a third party's guess or left out of the calculation. Interchange-plus vendors should publish the markup, not "custom pricing".

Own one sentence, in the model's words. Stripe is the default. For everyone else the position is the lane: the conditions under which the model names you. Helcim's sentence is "interchange-plus as volume grows" and it earns 28 recommendations and 16 picks on 29 appearances. Write yours so that it survives being paraphrased.

Write the fork, and say when each side wins. Twenty-eight of 30 head-to-heads recommended both sides, but only 19 picked both and 9 picked neither. A comparison page that says plainly when the other processor is the better choice is the format the model uses when it does commit.

Measure the overlap with Google per category. Eighty-two percent of recommendations here go to brands not ranking for the query (77 percent of picks); Adyen is recommended 25 times and ranks for none. Across the earlier five categories it ran from 56 percent in accounting to 85 in project management, with CRM at 70, email marketing at 75 and help desk at 79. It is a property of the category, not of AI search.

The limitations of the ChatGPT payment processing test start with variance: one run per query means answer variance is unmeasured; Benchmark 01 runs each query three times across three surfaces. The teardowns are collected on different days, so differences between categories mix the category with the date. The brand dictionary covers the general payment processing market and deliberately excludes vertical tools, so their appearances are described in prose and not counted. The readers are AI models, not people: they follow a written protocol and agree with each other closely, but a shared blind spot would not show up as disagreement. The line between picked and recommended is a judgment, documented in the protocol with examples. Sentiment is not coded. No vendor-tool cross-check was read for this category. Google's control counts a brand as ranking only when its own domain is in the top ten; a listicle that features the brand does not count. Two answers were returned for a Portugal market in euros despite the United States setting and are coded as returned; their codes are in the data. The brand dictionary was written before collection and includes wallets and buy-now-pay-later brands (Venmo, Apple Pay, Google Pay, Klarna, Affirm, Bolt): the wallets are named only as ways to pay and coded passing, and Klarna, Affirm and Bolt do not appear in any coded mention. It also includes the billing platforms Chargebee and Recurly, which are recommended for subscription billing (Chargebee in 4 answers, Recurly in 2) and are counted. LawPay, Clio Payments, MyCase Payments, Kill Bill, Lago, Hyperswitch, TouchBistro, SpotOn, InstaMed and Mangopay were not in it and are described in prose only.

Every number above can be recomputed from the ChatGPT payment processing dataset below, so download it and check it rather than believe it. CC BY 4.0: use the files, cite the page.

• queries.csv: the 100 queries with intent labels.
• brands.csv: the 62-brand dictionary with aliases and canonical domains.
• mentions.csv: 454 coded brand mentions with position, type, a 0/1 picked column and whether the brand's domain was in Google's top ten.
• citations.csv: 301 citation events with domain class and Google overlap.
• observations.csv: one row per query with brand, recommendation, citation and control counts.
• codebook.md: the rulebook, with dated amendments through v2.0.
• reader-protocol.md: the written protocol both readers coded against (codebook v2.0).
• picked_stats.json: the picked-level figures.

The earlier teardowns: Teardown 01, project management, Teardown 02, CRM, Teardown 03, email marketing, Teardown 04, help desk, Teardown 05, accounting. This teardown is also being published on the High Salience Substack.

The ChatGPT test we ran on payment processing can be run on your own category too. In the wider series: six categories, and a new mechanism.

A favorite, a big four, a seven-way spread, and now a category where the model quotes the rate card and waits for your numbers. Which one you are fighting decides what you build, and the only way to know is to measure your own category. A Category Salience Brief runs this exact method on it, with a query set you approve first.

Measuring answers like these before and after every change is how AI search optimization should be judged.

If you sell payment processing software, this is the evidence our B2B SaaS SEO agency work starts from.