The complexity is usually compliance (eg KYB, sanctions checks, AML) and while payment processors have done a lot to reduce the time and complexity it can still be a long and painful process.
In my experience (I’m CTO at a payment network) technical integration is <10% the time for a merchant doing this kind of volume moving processors.
Finally, what you described has happened. A lot of payment processors have been heavily influenced by Stripe’s DX and align relatively closely, or at least closely enough to make transition easier. The challenge is that few even medium sized merchants just use the core ‘move money’ APIs that are ‘easily’ replaceable, they use things like reporting/reconciliation, anti-fraud and a host of other products that make up an ‘ecosystem’ of products that in combination is very hard to replace quickly.
They've even implemented the load balancing idea: https://www.spreedly.com/smart-routing-for-payments
Pricing is in the hundreds-per-month plus cents-per-transaction, so they're not a fit for all businesses. But if you collect high-value transactions and keep cards on file (for subscriptions, SaaS, etc.), I highly recommend building on top of them rather than on top of an individual processor, as well as maintaining multiple merchant accounts with different providers. Business continuity is the name of the game!
It can be implemented, it has been implemented. Let's say that you'd like to move from one PSP to a new one, but you also have old PSP manage storing credit card information for you. All you have is a weird token, which represents your costumers credit card, but locked to you store. When you switch, you send the list of tokens to the PSP, which will then exchange information with the new PSP. Lastly the new PSP will send you a set of replacement tokens.
Granted it's not super automated, and not something PSPs will do, unless you're moving between them, but they can do it.
Most people don't realize is that paying with credit cards mean dealing with as many as four companies. The store, obviously, then the payment service provider, an acquirer and then the actual credit card company (and then maybe your bank).
The acquirer is normally pretty invisible. Some companies acts as both PSP and acquirer, but some PSP also use the same acquirer, or support multiple. In the later case, you can actually switch PSP, but keep the acquirer. In these cases I would guess that you could avoid doing having to transfer raw credit card data between PSPs.
That JSON-dump includes the card number, expiry, and billing address.
The very sad thing is that bitcoin was designed as a proposal to circumvent this specific problem. Be your own bank and stuff like that.
Then bitcoin and other cryptocurrencies became what they became... Basically speculative bubbles.
It's true that modern cryptocurrency is largely a pyramid scheme, but it could have been so much more.
This is why we (note the royal we) can't have nice things. It is also why becoming a financial services provider is a late-stage growth milestone of a business. Once you get past compliance, monitoring, and risk management boss, offering a line of credit or other finance products is an exercise in bookkeeping, and scales relatively trivially given sane accounting practices and managing your liabilities/issuing demographic.
If as a business, you achieve that and are still even remotely true to your original roots, congratulations, you're The Man(TM) now.
Doing your own PCI compliant credit card storage system instead of relying on your payment processor's tokenization system is not too difficult, but is a bit of a pain in the ass.
The biggest problem I see to switching between payment processors is the Visa and MC stored credential requirements. A few years ago they changed to a system where when you store a credit card you have to set a flag on the first transaction that tells Visa or MC that you are storing the card.
When you do subsequent transaction using the stored card, you have to tell them you are using a stored card and send them a reference to that original flagged transaction.
Most payment processors I've used give you a transaction ID that is something they generate, not the transaction ID that the Visa or MC network generates. If you are doing your own card storage that can lock you into using that same processor for future on-file and recurring charges on that card, because only that processor knows how to map the transaction ID you know (the one that processor generated) for the original transaction to the Visa or MC transaction ID.
Note that even if your current processor provides a way to get the underlying Visa or MC transaction ID there may be no way to use that with another processor. The other processor's APIs that take a transaction ID are going to expect one of their transaction IDs.
I'm not sure how widespread this is. Visa and MC announced their "Stored Credential Framework" several years ago. As the deadline approached for when it would become required, too many major payment processors were not ready and so it was delayed a year. By the end of that year there were still major ones not ready (Authorize.Net for example still hadn't even said what their plans were) and so it was delayed even more.
The processor we use was ready by the end of the first delay and we got our system converted to work with that, and I stopped following what was going on with the others so have no idea if it now actually is required everywhere or if there are major processors that still have an extension.
ProviderS. Plural.
Yes it will cost more to integrate with a different API, but not an enormous amount.
Given Stripe is pretty popular, it feels like for many it is becoming the "S3" of payments in terms of an API. Would it be viable for others to implement the same API on the basis of compatibility, portability and interoperability?
In the past, I recall dealing with systems where you'd have multiple backend processors for major external functions, and you'd load balance which provider was used. Some might have more favourable pricing, and be a preferred partner, while others might be potential future preferred partners being tried out (see how many failures you get from them, compared with the current preferred partner).
It seems that, if this was possible to implement (card tokenization obviously being an issue here), you could have multiple independent backend providers implementing the Stripe API as a provider, and you load balance across them.
Just like how you could use another S3 API provider for storage if one is giving you hassle, or not offering satisfactory performance or support. Sure, you have migration work to do, but having the same API and underlying properties should speed things up and let you get new transactions onto the new storage ASAP.
ACME is like this for SSL - there's now multiple CAs implementing the one API. Could we get there for payments too?