Payment Orchestration
Payment Orchestration vs. a Single PSP: When the Switch Pays for Itself
Picture a finance lead at a subscription business who notices that card approvals dip about a point and a half every time volume spikes. Her processor’s answer is that the decline rate is within normal range. It probably is. It is also costing her more in lost renewals than the processor’s entire fee, and she has no way to route around it because every transaction goes to one place.
That is the single-PSP problem in one paragraph. One provider means one set of approval behavior, one outage profile, one fee schedule, and one negotiating position. For a lot of businesses that trade is fine. For others it is the most expensive line on the P&L that nobody reviews.
Payment orchestration is the alternative: a layer that sits above your processors and decides, per transaction, where it goes. This piece compares the two honestly, including the cases where orchestration is the wrong answer, and gives you a way to run the numbers on your own volume.
What Each One Gives You
| Single PSP | Orchestration layer over two or more PSPs | |
|---|---|---|
| Integration | One API, one contract, one dashboard | One API to the layer; the layer holds the processor connections |
| Approval rates | Whatever that processor achieves on your mix | Route by card type, region, amount, or history to the processor that approves best; retry declines on a second processor |
| Outages | You are down when they are down | Automatic failover to a second processor |
| Fees | One schedule; leverage limited to volume | Least-cost routing where processors differ; real leverage at renewal |
| Payment methods and regions | Limited to what the PSP supports | Add a local method or regional acquirer without re-platforming |
| Card data | Tokens belong to the PSP; leaving means re-tokenizing or losing stored credentials | Vault-independent tokens; switch processors without losing cards on file |
| Reporting | Clean, single source | Consolidated across processors, if the layer does it well |
| Operational load | Low | Higher: rules to maintain, a second contract, reconciliation across providers |
The market has mostly voted. S&P Global Market Intelligence’s 451 Research 2025 merchant study, as reported by Bluefin, found 90% of merchants now work with two or more payment providers, and 62% prefer a multi-provider setup, up from 50% in 2023. Having several processors is normal. The open question for most businesses is whether to manage them with a proper orchestration layer or keep stitching them together by hand.
Where Orchestration Pays
The benefit is concentrated in five situations. If none of them describe you, skip to the next section.
1. Your Approval Rate Gap Is Measurable
If you have ever tested a second processor and seen a different approval rate on the same traffic, you already have the number that matters. A half-point difference on a high-volume business is real money, and routing captures it per transaction rather than per contract.
2. An Outage During Peak Would Be Expensive
Every processor has had one. The question is what an hour of downtime costs you on your busiest day. For a retailer in late November or a subscription business on billing day, failover and cascading retries alone can justify the layer.
3. You Sell in More Than One Region or Currency
Cross-border processing fees run far higher than domestic, and local acquiring usually approves better. An orchestration layer lets you add a regional processor for one market without touching the checkout.
4. You Are Negotiating a Renewal
A processor that knows it is your only option prices accordingly. A credible second connection changes the conversation, and the layer makes it credible.
5. You Want to Own Your Payment Data
In the same 451 Research study, 71% of merchants above $100 million in revenue strongly agreed that owning and controlling payment data independently from their processors is a priority. Vault-independent tokenization is the mechanism. Without it, every stored card is a reason you cannot leave.
Where It Does Not
Three situations where the honest answer is to stay on one PSP, at least for now.
One market, modest volume, stable approvals. If you sell domestically, your approval rate is steady, and a day of downtime would hurt but not wound, the layer adds a vendor and a set of rules without a return to match.
No one to own it. Routing rules, reconciliation across providers, and processor relationships need an owner. If payments is a part-time job for someone in finance, orchestration becomes a second part-time job nobody asked for.
The real problem is somewhere else. Declines caused by a bad checkout form, missing 3-D Secure configuration, or stale card data on subscriptions will not be fixed by routing. Fix those first; they are cheaper and the improvement is larger.
Three Ways to Get There
If orchestration is the right call, there are three ways to build it, and they are not interchangeable.
| Path | What it is | Fits when | Watch for |
|---|---|---|---|
| PSP-native multi-processor features | Your main provider offers routing to additional processors inside its own platform | You want failover and some routing with the least change | You are still inside one vendor’s walls; tokens and data stay theirs |
| Orchestration vendor | A third-party platform with prebuilt processor connectors, routing rules, and vault | You need many connectors fast and do not want to maintain them | Another contract and fee; routing logic limited to what the platform exposes; their roadmap is not yours |
| Custom orchestration layer | A layer you own, connected to the processors you choose, with rules written for your business | Payments is strategic, your routing logic is specific, or you need control over data and tokens | Needs engineering to build and run; you own PCI scope decisions; plan it like a product, not a project |
For most mid-market businesses, the first path is the right first step, and it can be taken this quarter. The second and third become worth it when routing logic gets specific or when data ownership starts to matter at renewal. Choosing between those two is its own decision, and we cover it in build vs. buy for payment orchestration.
How to Run the Numbers
You do not need a vendor’s ROI calculator. Four lines on a spreadsheet will tell you whether to keep going.
Approval uplift. (Expected approval rate gain, in points) × (annual card volume) × (gross margin on recovered sales). Use a conservative gain; a test with a second processor gives you a real one.
Outage exposure. (Revenue per hour on your peak day) × (hours of downtime you have actually experienced in the last two years, or a defensible estimate).
Fee delta. For cross-border or specific card types, the per-transaction cost difference between your current processor and the alternative, × volume in that segment.
Cost of the layer. Platform fees or build cost, plus the time of whoever owns it, plus the second processor’s minimums.
If the first three lines together are not comfortably larger than the fourth, stay where you are and revisit in a year. If they are, the next question is which of the three paths, and that depends mostly on how specific your routing needs are and whether anyone inside the business can own it.
The cheapest experiment. Before any platform decision, run 10% of traffic through a second processor for a month using whatever your current PSP allows. The approval rate difference you measure is the only number in this article that is yours.
Where to Start
If you want the four lines filled in with your own numbers, that is what a payments assessment does: we look at your approval rates, fee schedule, outage history, and data position, and tell you which path pays and which does not. Request a free payment strategy assessment, or see how we build custom orchestration layers.
Newsletter
Subscribe to newsletter
Subscribe to receive the latest blog posts to your inbox every week.