← Writing

Book teardown · Payments

The Anatomy of the Swipe

The one mental model every fintech PM should be able to draw from memory.

Author · Ahmed Siddiqui 2020 · 207 pp Read · ~6 min teardown Verdict · keep on the shelf

If you read nothing else

A "swipe" isn't one event. It's two separate journeys across five players, running at completely different speeds. Once you can see that split, every payments product decision (fees, fraud, rewards, chargebacks) stops being trivia and starts being a system you can reason about.

Siddiqui spent his early payments career at Marqeta, drawing the card-transaction flow on whiteboards "probably at least one hundred times" until it clicked. The book is basically that whiteboard, made legible. It opens, tellingly, not with technology but with a person: someone whose car dies with $5 in the bank, who gets emergency funds onto a debit card and swipes their way back to work. Payments, the book insists, are never really about rails. They're about what the rails let people do.

01The model worth stealing

When you tap a card, two things happen, but not at the same time, and not even on the same day.

First, an authorization: a question races from you → the merchant → the merchant's bank → the card network → your bank, which checks your balance and fraud signals and fires back a yes or no. Seconds. No money moves. Then, hours or a day later, clearing and settlement: the money actually moves, and it arrives at the merchant minus a set of fees. The biggest of those fees, interchange, flows to your bank. That's the quiet reason your bank, not the merchant, pays for your points.

Information first, money later. Get that one split into your bones and most of payments stops being a black box.

① AUTHORIZE seconds · a question, not a payment Cardholder you Merchant the store Acquirer merchant's bank Network Visa · MC rails Issuer your bank The issuer checks funds + fraud, approves or declines, and the answer returns along the same path. no money has moved yet. ② SETTLE ~1-2 days later · now the money moves Who actually gets paid on a $100 swipe: MERCHANT NETS $97.50 $2.50 fees The $2.50, unpacked: Interchange $1.80 → issuer Network $0.15 Acquirer $0.55 Interchange is the biggest slice, and it flows to your bank, not the store. That's who funds your points. (Figures illustrative; they vary by card + merchant type.)
One swipe, two journeys: a yes/no question that moves in seconds, and money that moves a day later, minus interchange.

02Three ideas I took from it

03How I read it

My take

What stuck with me is that the book refuses to start with the rails. It starts with a person who can't cover a $500 emergency, and only then explains the machinery that either serves or fails them. That's the right order. The mechanics of interchange and settlement only matter because of who's on the other end of the swipe, and what the system's design costs or saves them.

That framing is close to how I think about money products: the plumbing is the means, the person in a hard moment is the point. The interchange flow in particular reframed something for me. The whole issuer model runs on cards that stay active and keep swiping; interchange only earns when money is moving. I work at the opposite end of that lifecycle, the part where the swipes have stopped and someone can't pay, and seeing the economics laid out this plainly made the stakes obvious. The work of keeping someone in the system when they fall behind isn't a cost center bolted on after the fun part. It's how you protect the active, swiping relationship the whole interchange engine depends on. Same system, opposite end.

The one critique: it's deliberately a primer. If you already live in payments, Parts 1 and 2 are revision, so skim to "Digging Deeper" and "Payments in Action," where interchange and the build-it examples earn the page count.

04Verdict

The best on-ramp I've found for anyone who needs the card-payments model in their head and learns visually. Founders, PMs entering fintech, and engineers shipping their first payment flow get the most from it. If payments is already your day job, keep it as the thing you hand to a new teammate, and steal the diagrams.