Buy first, sign up later
Rebuilding checkout so a first-time buyer can pay before they finish creating an account.
Most people buying a ticket on TicketMax have never used TicketMax before. In August 2026, 77% of buyers were new accounts. That makes sign-up the single biggest thing standing between someone wanting a ticket and having one.
The problem
The old checkout wasn't broken. The order summary and the account step lived on the same screen, with log-in and sign-up as tabs so nobody got sent away mid-purchase.
But a new buyer completed three sequential steps before paying: name, email, phone and date of birth; then an SMS verification code; then a full address. Only then could they choose Pix or card and pay.
Every one of those fields was legitimate. Together, they asked someone to open an account with a company they had never used, in order to buy one ticket.
The flow also asked a question buyers often can't answer: do you already have an account? People forget. Guessing wrong drops them into a password reset in the middle of a purchase.
What changed
One field replaced the log-in-or-sign-up decision
Checkout now opens with a single CPF field. If the system recognises it, the buyer goes straight to payment. If it doesn't, they're a new customer — and the system knows that without asking them to know it.
Registration split into what the purchase needs and what the account needs
To pay: full name, email, phone, and the billing address the payment itself requires. Phone verification, password, date of birth and the account's own address moved to after the purchase.
Ticket access became the reason to finish
The purchase completes, but tickets only appear in the account once registration is done, with reminder emails until it is. Nobody is blocked from buying, and nobody has to be chased before the money is in.
Buyers without a CPF have a way through
An explicit option switches to passport, so foreign buyers aren't stopped by a Brazilian tax ID.
Results
The express checkout shipped on 4 July 2026.
Of the accounts created in a month, the share that went on to buy a ticket. June, before the change: 75.2%. August: 84.4%.
That comparison is conservative. June's cohort had three months to convert; August's had one.
That's the cost of the trade, and the number I keep watching.
What's honest about the 9-point jump. Part of it is mechanical: an express account only comes into existence because someone was already paying, so it enters the data with a purchase attached. Separating the two routes in August, accounts created at checkout converted at about 87% and accounts created the old way at about 80% — both above June's 75%. The gain isn't only a change in mix, but some of it is, and the number shouldn't be sold for more than it's worth.
The change didn't cost payment reliability. July, its first full month, had the quarter's highest card approval rate.
What the data showed next
The more useful part of reading three months rather than one was separating what the redesign caused from what it didn't.
Card declines run near 40% in all three months, including the two before the change. That is chronic and structural, not something checkout layout created. Pix abandonment did spike in August — but 266 of the 350 abandoned codes came from a single event, a low-ticket impulse purchase that came to dominate the month catalogue. A composition effect, not a product regression.
One finding changed how I'd approach the next version: the first durable record of purchase intent is the transaction itself, which is created too late to measure the top of the funnel. Someone who hit the old sign-up wall and left never generated a row anywhere. Persisting a checkout-started event is the instrumentation I'd want in place before changing this flow again.
What's next for this flow
Data entry and payment selection still happen on a separate screen from ticket selection. The next version brings both into the sidebar, under the order summary, so buying never changes screens. It isn't built yet — engineering priorities are elsewhere right now.
