Payments Analyst

Why work at Higgsfield AI?
Higgsfield AI is the fastest-scaling generative AI company in history, hitting $1B in annual revenue run rate, 30M+ users worldwide, 6M+ generations per day, and powering 390 of Fortune 500 brands.
We're building at the absolute frontier of AI-powered video creation and next-generation creative tools. Joining Higgsfield means becoming part of a high-impact team shaping the future of AI-native experiences, at a company that isn't just moving fast, but rewriting what fast looks like.

What This Role Means at Higgsfield
We sell subscriptions and credit packs to consumers in over a hundred countries, and today almost all of it moves as cross-border card payments in USD . That has three consequences, and this role owns all three.
We are declined more than we should be. In some of our largest growth markets roughly one attempt in three fails, and a meaningful share of those are not "no money" — they are cross-border blocks, card limits, stale credentials on renewal, and issuers who do not recognise us.
We pay more than we should. Our blended cost of acceptance is the single largest line below compute, and the majority of the network component sits on transactions our processor classes as cross-border.
We are not on the rails our customers use. In two of our fastest-growing markets, the dominant local payment method is one we do not offer at all.
You own the measurement and the argument behind fixing all of it.
You make sure:
We know our true authorization rate — at the unit the business actually cares about, not the one that is easy to compute

Every basis point of payment cost has a name, a geography and an owner

A decline is a diagnosis, not a statistic

When we propose a new rail, a new entity or a new processor arrangement, the number underneath it holds up to Stripe's own analysis and to our board's

What You Will Do
Authorization & decline analysis
Own authorization rate end to end: by country, issuer, BIN, card brand and funding type, plan price point, first payment vs renewal, and payment method.

Build and maintain the decline taxonomy — what each decline code actually means, which are recoverable, which are terminal, and which are us rather than them.

Measure at the right unit . An invoice that succeeds on the third retry is a success, not two failures and a success, and reporting it the other way has made teams optimise the wrong thing.

Separate soft declines we can recover from hard declines we cannot, and put a number on the recoverable pool.

Cost of acceptance
Own the blended take rate and its decomposition: interchange, scheme fees, processor markup, cross-border and currency-conversion components, disputes, and the fixed per-transaction pieces.

Find where the cost actually sits. Averages hide it — the cost is concentrated by corridor, by card type and by whether the transaction is domestic.

Build the evidence for commercial conversations with our processor, and keep it current enough to reuse.

Local payment methods, entities and routing
Size the case for new rails market by market: what share of local commerce runs on them, what our current card performance costs us, and what we would realistically recover.

Model merchant-of-record and local-acquiring arrangements against the alternative of our own entities, honestly — including the fee we would pay for the privilege.

Design and read the experiments that prove it, with the product analytics team. A new payment method is an A/B test, not a launch.

Renewals, dunning and involuntary churn
Own the retry and dunning strategy as a measurable system: when to retry, how often, through which rail, and when to stop.

Quantify involuntary churn separately from voluntary. They look identical in a cancellation count and they need completely different fixes.

Track the health of stored credentials — network tokens, card-account-updater coverage, expiring cards — because in a subscription business that is silent re