Reading a payment market in one day. 6 chapters and a final quiz.

Chapter 1. Framing the day: five questions, five sources.

Starting with the players is a poor way into an unfamiliar market. The names come quickly, the logos are reassuring, and you think you have understood something when all you have done is memorize a directory. The structure stays invisible. One day is enough to rebuild it, provided you ask the questions in a fixed order and stick to that order even when a find tempts you to jump ahead. Every question has an answer published somewhere, by an identifiable institution, on a known date. The job is knowing where, and stopping there.

The order is not decorative. Pricing does not come before rails, because an interchange cap barely matters in a market where cards carry 15% of payments. Deadlines come last, once you know the usage, the infrastructure, the regulator, and the pricing. Without those, a regulatory date is just a date with no identifiable consequence. A partial answer to all five questions is always worth more than an exhaustive answer to three of them. The picture of the market warps as soon as one layer is missing.

The five questions, and who publishes the answers

Chapter 2. The hierarchy of sources, and the lag each one carries.

Sources are not equal, and the most recent is almost never the best. A central bank statistic is slow because it is collected from every reporting participant, checked, then aggregated under a classification that stays stable from year to year. An operator’s press release is fast because it goes through none of those three steps. So choose by intended use, never by apparent freshness. To compare two countries, use the slow source. To date an event in the past few months, use the fast one, and state in the memo that the source has changed.

Six families of sources, from most to least reliable

All three dates say the same thing. A rigorous international comparison always covers a completed year, and often the year before that. A fast-changing market appears frozen in a state it left long ago, which yields memos that are accurate about the past and wrong about the present. The fix is not to discard the slow source. Use it to compare countries, then cover the last 12 months with the operator’s own publications, and flag each change of source in the body of the text.

Chapter 3. Screening a figure: six checks before you cite it.

A marketing figure is rarely false; it is framed. The publisher chooses the unit, the period, and the scope that show it in the best light, then publishes it without having to lie once. A hurried reader takes the figure, leaves the framing behind, and carries into the memo a claim the publisher never actually made. Six checks are enough to restore the framing, and they take a minute per figure.

The six checks, in order

• Unit. Transaction count, value, users, accounts, merchants? Two different units often rank two players in reverse order.
• Period. A calendar year, a fiscal year, a quarter, a single month? Multiplying a record month by 12 gives a false annual figure.
• Cumulative or flow. Does the figure cover a period, or the entire time since launch? A running total since launch is the most common framing for services that have stalled.
• Scope. The whole country, only the members of one consortium, only the publisher’s own customers? An operator’s total is not a national total.
• Active or registered. Opened accounts or used accounts, and over what window: 30 days, 90 days? The gap is structural, not marginal.
• Publisher and vintage. Who publishes it, with what interest, and when was it measured? A correct figure ages: it becomes false without anyone correcting it.

Then there is the figure you cannot find. It comes up often in poorly documented markets, and there is nothing unusual about it. The memo then records the absence in a specific sentence, such as “no public card fraud statistics are published in this country.” That statement is usable information: it steers the next step toward an interview with a local practitioner. It is worth far more than an estimate borrowed from a neighboring country because it shares the same language or currency.

Chapter 4. Mapping the players in four layers.

A list of players is hard to remember and harder to analyze. A stack, by contrast, can be read at a glance and compared across countries, because the functions are the same everywhere even when the names are completely different. Four layers are enough. Every player you meet during the day goes into exactly one layer, and that sorting tests your understanding. A player you cannot place is one whose business you have not understood.

Comparing countries is less about names than about how functions are distributed. The question is who sits in which layer, and under what kind of ownership. Three models recur everywhere: the central bank as operator, a consortium owned by the country’s banks, and a supervised private company. The model chosen for the clearing layer partly explains price levels, the pace of innovation, and how easily nonbank entrants get access. You find it in the operator’s articles of association, not in its marketing.

Layers 1 and 2 merge in several countries, and that merger tells you something. In Bangladesh, the National Payment Switch Bangladesh is owned and run by Bangladesh Bank. In Czechia, Česká národní banka, the Czech central bank, operates CERTIS, the country’s only interbank system, which handles both gross settlement and bulk clearing. A central bank that operates the system sets the pace of technical change itself.

• Which layer? A player that appears in two layers is either an integrated group or a mistake on your part. Decide which.
• Who owns it? Banks, the central bank, private capital, a telecom operator: ownership predicts behavior better than public statements do.
• How long has it had that name? Mergers, acquisitions, rebrandings: three causes of outdated citations, and very common in this industry.
• Is it still operating? A discontinued system keeps getting cited for years, and the Paypedia registry keeps 80 discontinued systems for exactly this check (Paypedia registry, 1,163 systems in 146 countries, 2026).

Chapter 5. Price, access, and deadlines: the three answers that decide.

Three answers turn a description into a view someone can act on. Price, because it sets the margin available to everyone and determines whether a model imported from another country still works. Access, because a dominant rail can remain legally closed to nonbank entrants. Deadlines, because a dated rule can make an analysis written six months earlier obsolete without warning. These answers come from the regulator and the operator, never from the market’s sales pitch.

Four pricing regimes, four philosophies

These four regimes cannot be read in isolation. India’s zero cap wipes out the business model of the RuPay debit issuer, which survives only through regulatory mandate and government incentives. Brazil’s debit cap, combined with Pix being free for individuals, explains why instant payments are replacing debit cards. A cap is never just a cost line: it shifts volume to the rail that is not capped.

Access works the same way. A rail can carry half of a country’s payments and still admit only locally licensed banks as participants. Identifying the dominant rail is not enough. The description has to specify which authorized intermediary provides access, and under what regulatory status. The answer is in the participation rules the operator publishes, in a technical document nobody promotes, and never on its marketing pages.

Deadlines: what is changing, and when

• Euro area: receiving instant credit transfers mandatory since January 9, 2025, and sending them since October 9, 2025, at the same price as a standard credit transfer (EU Regulation 2024/886).
• United States: the Fedwire Funds Service migrated to ISO 20022 in a single cutover on July 14, 2025; the previous proprietary format is no longer accepted.
• Australia: surcharging banned on designated network cards from October 1, 2026; cap on foreign-issued cards from April 1, 2027 (RBA, 2026).
• Europe (cards): no new Maestro cards issued since July 1, 2023, with knock-on effects on co-badging with domestic schemes (Mastercard).

Chapter 6. Writing the one-page memo.

The one-page memo is the deliverable, and its length limit does part of the analysis by itself. On one page, you do not copy; you prioritize. Whatever does not fit goes into an appendix or disappears, and that sorting immediately exposes the sections where the research was thin. Your reader will decide without you, without time to follow up with you, and often in a meeting where your memo circulates without its author. After reading it, they must be able to act, or know exactly what is still missing.

# {Country} — market read
Author · Date written · Sources as of {MM/DD/YYYY}

## In three lines
{How people pay · who owns the dominant rail · what changes in 18 months}

## 1. Usage
{Mix by instrument, WITH unit (count / value) and period}
Source: {institution, publication, year}

## 2. Infrastructure
Layer 1 settlement: {system, operator}
Layer 2 clearing: {system(s), operator}
Layer 3 schemes and rails: {domestic card scheme, instant rail, QR}
Layer 4 distribution: {acquirers, aggregators, wallets}

## 3. Regulation and access
Regulator: {institution}   License required to hold funds: {license}
Access to the dominant rail: {direct / via a bank / via a licensed aggregator}
Source: {legal text, date}

## 4. Pricing
{Interchange or MDR capped or uncapped · surcharging allowed or banned}
Source: {legal text, effective date}

## 5. Dated deadlines
{Date} — {event} — {legal text} — {scope}

## Confidence and caveats
Solid: {points verified against a primary source}
Fragile: {points resting on a single or secondary source}
Undocumented: {what is not published in this country}

## What would invalidate this memo
{The event that would force a rewrite}

Three sections set a memo apart from a country profile. The three-line summary, always written last, once the author knows what has actually been established. The confidence block, which rates each claim by the strength of the source behind it. The closing line, which names the event that could make the memo wrong overnight. Experienced readers go straight to the last two.

Then there is what you leave out. No five-year forecast without a published model, no market share lifted from a vendor overview without saying what it is, no recommendation to enter the market, since the brief was to describe. A memo that drifts into recommendations gets judged on that ground, and the descriptive work disappears.