Execution Benchmarking: A Fair Grade at Every Transaction Size

A small transfer and a large one are no longer held to the same standard. What's new in v1.3, and the data underneath it.

The biggest upgrade to Execution Benchmarking since we launched it is live, and it changes the grade on a large share of the payments you send.

Every transaction is now graded against comparable payments on the same corridor and direction, at a similar size, using what those payments actually cost to deliver. The grades as still Competitive, In Line, or Off Market, and your full history has been re-scored on the new methodology. A $40 payment and a $40,000 payment are no longer measured against the same reference.

If you've ever looked at a small transfer flagged Off Market and thought the number couldn't be right, this update fixes this.

What you get

Four upgrades that compound.

• Your small transfers are graded fairly. Transactions are now compared inside size bands, so a small payment is measured against other small payments on the same corridor and direction.

• A flat fee no longer reads as bad execution. Providers price delivery as a flat fee plus a percentage, and the methodology now separates those before grading. You're scored on how well the money landed rather than on how much you sent.

• The reference is what actually happened. A quote is what a provider says a payment will cost while a completed transfer is what it actually cost. The reference is now built from what comparable payments achieved on the network, with quotes used only when there isn't enough completed flow to be honest about. If that is the case, we flag it on the card.

• Each corridor gets its own tolerance, recalculated daily. A deep, stable corridor gets a tight band and a volatile one gets a wider band, based on how each actually behaves. Where data is thin, the band widens and the confidence tier says so.

What you do with it

The most common use is catching a provider before it drifts. A provider that was fine in March gets a little worse each month, and without a benchmark the first you hear of it is a customer complaint. When the grade moves, you can go back to the provider while the problem is still small.

The same reference works in the other direction, when a customer asks you why a payment cost what it cost. You can point to the network reference for that corridor and size on that day instead of a spreadsheet of your own past transfers

Over a longer window, the grades tell you where to add capacity. A corridor that keeps grading Off Market quarter after quarter has made the business case for a second provider before you've spent engineering time on an integration that might not pay off. The corridors that keep grading Competitive are the ones you can leave alone.

A reference no single provider can build

Mid-market rates are free from a dozen sources, but nobody delivers at mid-market. The number that matters is what a payment actually cost to deliver into a bank account in that country on that day, and that number can be difficult to get.

A provider only sees its own book. It knows what it charged you, and nothing about what the transfer next to yours cost through someone else. Your own data has the same limit in a different shape: it's a sample of one participant, so it can't separate a corridor getting worse from a provider getting worse. Rate feeds publish quotes, which are what someone was willing to advertise rather than what settled.

Borderless.xyz sits in a different position. Completed deliveries flow across many corridors and many providers on the same connection, measured the same way at the same time. That comparison set exists in exactly one place, because no single client, provider, or data vendor can see across the others.

The reference is built from what those providers actually delivered, and they're providers we connect to rather than run or price, so you're graded against their results, pooled and measured identically for everyone on the network. You can also import transactions you've run through providers outside the network and we'll grade those on the same basis, so the benchmark covers your whole book.

It gets sharper from here

Every corridor we add, every provider that joins, and every client who imports their history makes the reference denser and the grades more precise. The sooner your own transactions and your imported history are in it, the sooner your grades sharpen too. Where the data is thin today, the confidence tier says so plainly rather than dressing a weak signal as a strong one, because a benchmark that overstates its certainty is worth less than no benchmark at all.

This is the second major version in five months, and it won't be the last. We'll publish what changes and why, the same way we did here.