Payments at Pix speed: the stress test for your close

Illustrative scenario, with no client and with assumed figures. A merchant with 4,000 collections a day reconciles every morning, in batch, against the previous day's bank statement. Today 800 of those collections are digital, and 3% arrive without a reference that identifies them: 24 items that a person resolves by hand before noon. A year later, with the same merchant and the same share without a reference, digital collections are 2,000. The manual items are 60 and arrive at any hour, including Saturday night and the two minutes before midnight on the last day of the month.

Nobody changed the collection system. Collecting worked better than before. What broke was what sits behind it: reconciliation, the accounting close and the handling of payments that are later disputed.

That is what this article wants to make clear if you are a CEO, a CFO, or a head of operations, sales or technology at a bank, a fintech, an insurer, a retail chain or a regulated company: with instant payments, the point of failure can move from the collection to everything that happens afterward, especially if reconciliation, the close and the handling of claims were designed for a different collection pace. Brazil, through the scale of Pix, its instant payment system, and its collection and return mechanisms, offers a useful precedent: it is worth looking at what happened. Brazil illustrates operational risks; it does not by itself anticipate how SPEI, CoDi or DiMo will behave in Mexico, so it is worth validating each scenario against your own flows and contracts.

What is being discussed in Mexico, and what is not yet law

According to Forbes México, the federal government has a goal: that, in the first year of application of a new Ley de Economía Digital para Pagos Digitales y Electrónicos (Digital Economy Law for Digital and Electronic Payments), about half of transactions be digital; the article attributes it to the head of the digital transformation agency [1]. The official press release from the Presidency on the initiative does not publish that figure [1]. As a starting data point, the 2024 National Financial Inclusion Survey (ENIF) reports that 85% of people use cash most often for purchases of up to 500 pesos [9]; it is a measure of people, not of transactions.

Three things are worth separating. First, progress: the Chamber of Deputies approved the initiative on October 6, 2026 and sent it to the Senate; its committees issued a favorable report on October 14 and the full Senate approved it on October 20, 2026, so it now goes to the Executive; publication in the Diario Oficial (Mexico's official gazette) is still pending [2]. Second, status: as of October 20, 2026, it is not yet a law in force, and it should not be treated as an obligation until it is published in the Diario Oficial; the legislative status should be confirmed at the time this article is published. Third, the goal: it is a calculation by the Executive that Forbes attributes to an official, not a legal target or an obligation in force, and we have not seen a methodology behind it. It is a reasonable signal of where the environment is heading, not a sanctioned calendar or an obligation for your company.

Nor is it necessary to wait for the law to be published in order to act. Mexico already has instant payment means (SPEI, CoDi and DiMo: Banxico's, Mexico's central bank, real-time transfer, QR-collection and mobile-payment schemes, which are not equivalent to one another: the actual experience depends on the means and on the participant), and both the article cited and the official press release speak of making use of the existing capabilities of SPEI and CoDi for instant transfers [1]. Here we will not use adoption figures for any of the three: we have not been able to contrast them with Banxico's official source, and we prefer to describe the effect in qualitative terms rather than repeat a number without backing. The useful question for your committee is not how much the volume is going to grow, but what part of your operation depends on the volume staying where it is.

What Brazil has already lived through

Pix launched on November 16, 2020. Figures from the Banco Central do Brasil (BCB) allow us to size an expansion that took place in under six years.

Scale. In November 2020 there were 41.2 million users with at least one key registered in the Pix directory; as of September 30, 2026 there were 186,974,734 (almost 187 million) [3]. It is a measure of registrations, not of active users or of share of the population, and we say so with that care. In volume, the BCB's EPAE series (which excludes cancellations, returns, withdrawals and transfers between accounts at the same institution) recorded 15.9 million transactions in November 2020 and 5.4 billion in a single month, September 2025, with a monthly value of nearly 1.7 trillion reais [4]. In the Central Bank of Brazil's open data, which follow a different methodology, September 2026 recorded about 7.17 billion transactions [11]. The contrast shows that an instant payment system can reach billions of transactions a month; it does not, by itself, show that every company has moved from the daily batch to continuous processing, and that is validated against its own flows.

Recurring collections. In June 2025 Pix Automático was launched, which allows recurring collection from individuals on behalf of a company; it came into operation on June 16, 2025, and its availability depends on each participant and on whether it acts as payer or as receiver [5]. The BCB already publishes statistics with a category of its own for these collections, including those from company to company and from person to company [8]. In September 2026 there were about 8.9 million such transactions, just over one in every thousand Pix transactions that month [8][11]. It is still small, but it shows that instant recurring collection has stopped being a promise: it already moves money. For a company with subscriptions, rents, fees or premiums, it is worth testing whether fixed dates (the twice-monthly payday, or quincena, and the due date) concentrate events; the Brazilian figure alone does not quantify that effect.

Collections with data to reconcile. Pix Cobrança is Pix's collection instrument. It is initiated with a QR code —static, dynamic or composite— or with Pix Copia e Cola (the copy-and-paste code), and it serves both immediate payments and payments with a due date; in the latter mode, the charge carries a due date and may include interest, fines and discounts [6]. According to the BCB manual, the dynamic QR code makes it possible to reconcile each payment through the transaction identifier [6]. It is a capability of the scheme; taking advantage of it depends on the company keeping that identifier and matching it against its own records.

Returns. Brazil also built the Mecanismo Especial de Devolução (MED, Special Return Mechanism), a procedure for contesting payments in specific cases of fraud or operational failure; it is not a general channel for commercial disputes or buyer's remorse, and its outcome depends on assessment and on the funds available. Two facts weigh for a CFO. First, since September 1, 2026, whoever received a legitimate payment and suffered an improper return has up to 80 days, no longer 30, to contest it; whoever sent the payment and was a victim of fraud already had 80 days [7]. Second, in 2025 an average of 9.3% of the contested value was recovered (a figure from a BCB technical presentation; a historical reference for 2025, not a current rate) [7]. Read it calmly: it does not mean Pix is unsafe, it means that in the cases of fraud or operational failure that the mechanism covers, a return request can be opened on a payment received, up to months later, and that recovering the money once it has left is hard. Prior prevention and early detection help; later mechanisms can recover funds in full or in part.

Where it breaks: not in the collection

With that context, the three stress tests.

1. Reconciliation stops being a next-morning chore

A process that reconciles in batch against the previous day's bank cutoff works as long as payments arrive at known hours and volumes. When money arrives at any minute, the "unidentified" items accumulate, and the time they spend without an owner grows longer. Three concrete problems appear first:

• Payments with no reference. A customer pays with a wrong reference, or without one. The operation needs a written rule, an owner and an agreed resolution time for every payment with no reference.
• The same notice, twice. A payment notification that arrives repeated, because of a network failure or a retry, cannot credit the revenue twice. It is the same principle we developed in the blog about APIs: that retrying must not duplicate.
• Bursts. It is worth framing as a capacity hypothesis that due dates and payment dates may concentrate events, and testing it against the company's own histories and contracts. A system that receives payment notices must tolerate the peak, not just the average.

2. The accounting close becomes sensitive to the minute

When payments arrive instantly, the period cutoff moves onto uncomfortable ground: the collection at 23:58 on the last day of the month, the one that crosses midnight between two periods, the one received on a weekend. The auditor's question stops being "does the month reconcile?" and becomes "which period does each item belong to, and how do you prove it?"

There is also a tax consequence worth reviewing with your tax area: if your sales are invoiced on credit or in installments, review the match between the payment received and the payment complement (the tax receipt issued when a payment is received on an invoice billed on credit terms); the SAT (Mexico's tax authority) allows issuing it for each payment or grouping the same recipient's payments for the month [10]. Which payment method to record for each means, and how to match the complement with a payment that arrived without a prior invoice, are questions for your accountant and your tax advisor, not statements of ours.

3. "Collected" may stop meaning "final"

This is the test that is least anticipated. In a batch scheme, the money is confirmed days later and the claim usually arrives within known processes. In an instant one, the money is available at once, which helps cash flow, and it can also leave at once. A later adjustment may require an accounting review when there is an applicable case of fraud, return or operational error; it is not an automatic consequence of every instant payment.

What Brazil shows is that, where there is a risk of claims, the company should agree with Accounting, Risk and Compliance which states and evidence it requires to handle identified collections, open investigations and confirmed adjustments; for example, an intermediate state between collected and final. Not every collection is subject to an MED claim. Without that definition, the ERP may recognize revenue that can still be reversed, and one month's close may be corrected in the next. For a bank or a fintech, moreover, the challenge is twofold: registering the claim without breaking the close, and stopping in time the money that is heading to a bridge (transit) account.

What to prepare before volume rises

Before deciding whether you need a larger program, your team can run these six checks without waiting for the law:

1. Measure your starting point. What proportion of your collections arrives today without a reference that identifies them (in the scenario, 3%: 24 items today, 60 when digital collections reach 2,000), and how much of your reconciliation is still manual. Without that number, any automation goal is a wish.
2. Make each collection travel with its identifier. A unique reference per collection, from the moment it is generated, and a written rule for what happens to the payment that arrives without it.
3. Test what happens with the repeated notice. Test that a repeated notice does not generate a second revenue entry or a second accounting posting. It is tested, not assumed.
4. Define the revenue states. Which states it needs (for example, collected, in possible claim and final, according to its accounting policy), with who authorizes the step from one to the other and within what time.
5. Rehearse a hard close. Simulate collections at 23:58 on the last day, a long weekend and a payday peak. See where it breaks before the auditor does.
6. Review your availability. If your collections system, your core banking system or your provider have nighttime maintenance windows, instant payment makes them visible.

What is best avoided is waiting for the volume to arrive in order to discover which of these six was the weak one. In the scenario we opened with, the difference between 24 and 60 manual items is not noticed in Monday's operation; it is noticed when someone asks why the close ran three days late. That merchant needed, from the start, a written rule for the payment that arrives without a reference and a close test before the 60 items arrived on a Saturday night.

Six questions for your next committee

• If half of my collections were digital and instant, how much of my reconciliation would still be manual?
• What percentage of my collections arrives today without a reference?
• Can my system receive the same payment notice twice without duplicating the revenue?
• If a customer disputes a payment that has already been reconciled, who resolves it and within what time?
• Does my monthly close hold up with collections received at 23:59 on the last day?
• Does my core system provider, or my collections provider, operate at any hour, or does it have a nighttime window?

If two or more of these questions have no owner, evidence or verifiable answer, treat it as an operational risk worth assessing before volume grows.

What Hábil does

At Hábil we address, in general terms, the problem of money arriving faster than the company's ability to explain it: we help review traceability across collections, ERP, invoicing and banks: reconciling each collection with its reference, resolving unidentified items with clear rules and reaching a close that can show which period each movement belongs to. It is the line covered by our product Cuadre; its fit and scope depend on each company's flows, constraints and controls. And because reconciliation depends on collection systems, ERP, invoicing and banks understanding each other, it is also the subject of our page Unify your operations and your channels.

If you want to know where your close would break if your digital collections multiply, we offer a first assessment conversation, at no cost, over WhatsApp: reviewing together the map of the integrations around your operation and the points where time, money or audit trail might be getting lost, based on your current flows, contracts and controls. If it makes sense to go deeper, the scope is agreed with you.