Blog

A journey in ProcessOut's flight journal

The payment performance challenges

by Guillaume Merindol on

Today, I’d like to give you insight on some of our nomenclature and even on metrics we use internally to know how good our algorithms are doing. This will be rather short but will give you an insight on what you can work on, and what we are working on.

In payments, the merchant has three axes to work on: authorization rates, fees, and chargebacks.

The problem with the Authorization Rate

The authorization rate is intrinsically the most important stat: it’s essentially measuring the income you should be getting but you’re not. If you could go to 100% authorization rate, you wouldn’t. 100% would mean letting fraudulent payments go through and thus increasing your chargebacks, decreasing the trust banks/networks/acquirers have in you and costing you $10+ per chargeback.

Different merchants should have different standards for a “good” authorization rate. A gaming website’s good standard might be 85%, where as a watch-selling ecommerce might be 95%. That’s the problem. We want a stat that has more meaning, that makes sense across any merchant. A stat that you can aim for 100% without problems.

Easy, instead of dividing by the total number of transactions, you divide by the total number of transactions that weren’t flagged as fraudulent (i.e. some failed transactions). We call this internally the True Authorization Rate (TAR)*. We use this notion when we’re training our algorithms and when we’re measuring merchant performances.

However, we always keep both the AR and the TAR in mind, because counting only non fraudulent-transactions is not exact. Not every flagged transaction would end up as a chargeback. Right now, to count these transactions that weren’t flagged as fraudulent we trust our partners. We realize that there are plenty of fraud-scoring tools right now that are viable, and we estimate this is not (yet ;) how we can unlock you money.

We want to bring all our merchants to 100% TAR. We’re working with some of the biggest ecommerce players in Europe to improve their TAR. Our strength is using their different data to improve each one’s TAR.

*TAR: we’re open to better names

Effective outcome vs deserved outcome

Every point off your TAR is essentially income that should be yours but isn’t because of weird rules or technical integrations in the online payments world. 100% TAR is essentially your deserved outcome, whereas your current TAR is your effective outcome.

You’re making 90 a month, but you’re at 90% TAR. Your deserved outcome is 100 a month. This is why we talk about “unlocking” your money. We’re here to bump that up to a 100. The bigger you get, the more important your TAR gets. A single percentage point on your TAR means millions. This is one of the reasons we generate so much interest amongst big players.

The everything-is-linked problem

Merchants that want to have the best possible payments have to simultaneously juggle with the trifecta: fees, authorization rate, and chargebacks. Just taking care of one of these is hard enough, but having the 3 work in tandem in order to boost your effective income is even harder.

Maintaining the balance is delicate. Here are a few examples of problems:

  • If you increase your AR you might increase your chargebacks, if you increase your chargebacks you can technically increase your fees. (Talking about TAR instead here solves this)
  • If you increase your AR through using multiple acquirers you might increase your fees because your old acquirer is receiving less transactions.
  • If you decrease your chargeback you might actually block too many transactions and lose AR.
  • You can increase your AR by using specialized acquirers/players that are more expensive but let risky transactions go through. (So you’re increasing your fees…)

Mission

Our mission right now (2018) is as follows: Unlock the deserved income (through TAR) you’re not getting whilst bettering your fees, and not touching your fraudulent transactions.