Payment orchestration, open payments, and whether the rails are ready for agents: Peter Dougherty of Spreedly

Payment orchestration began as a narrow idea: you can use more than one payment processor at the same time. Peter Dougherty's argument is that the concept has outgrown its name, and that the more urgent question is whether the world's payment infrastructure can survive what comes next.

Dougherty is president at Spreedly, which he describes as a universal adapter for payments: one interface connecting to over 140 gateways, with more than $50 billion in transactions running through the platform annually. He joined in 2024 after more than a decade scaling Lightspeed Commerce.

In his first twelve months, Spreedly quadrupled year-over-year growth in annual recurring revenue while holding operating expenses flat.

The right people at the right problems

Asked how, Dougherty's answer is about assignment rather than tactics.

He is a believer in having the right people pointed at the right challenges. When he encounters a business problem or opportunity, the first thing his mind reaches for is who is going to be working on this, and whether the skillset and personality fit.

His summary: everybody is great at something, nobody is great at everything.

And the first step in placing someone is understanding what motivates them. When you ask people to do hard things, money is part of the equation and not the whole one. What matters is when they feel like they are winning, and what makes them feel that way.

Then match that to the shape of the problem. Is this operational, customer experience, engineering, sales? His observation about the payoff: once people are excited and energized, magic happens.

One-way doors and room to learn

Michael's follow-up is the practical constraint. People are sometimes excited by things they do not yet know how to do, and sometimes you cannot afford someone learning on the job.

Dougherty's framework is one-way doors and two-way doors, which he credits to a well-known founder.

The question is which decisions are existential to the business. That is where there is no room to learn on the job. Everything else may be important without being existential, and the useful follow-up is how much runway you have before it must be solved, and what happens if it goes wrong.

Then the more textured version, which is about the shape of a person rather than a binary. Everyone has a broad set of skills at different levels of proficiency. His example: someone excited by marketing who is strong on performance marketing and softer on brand, or the reverse. Which of those matters more to your company right now?

So you hire for the thing that is critical today, accept that they will learn the other, and check that the thing they need to learn is something they want to learn.

His phrase for why this is hard and interesting: people are wonderful tapestries of skills and experiences.

How to decide what is existential

Michael asks how a leader avoids treating every problem as the sky falling.

Dougherty's aside is worth having on record: every executive you have ever met is also privately anxious about whether their bets will work, regardless of how calm they appear.

His method starts at the top. What is the overarching objective of the business? Is this a growth story or a profitability story? Are you taking market share or optimizing cost?

Then, unavoidably, the money. Work with the CFO, look at the income statement, and find where the opportunities actually are. New customer acquisition, upsell, retention, or a failure to launch during onboarding.

The example he gives of the trap is a good one. In high-growth organizations you often hear that revenue growth needs fixing and therefore the sales team needs revamping. And it might not be the problem at all. It might be something entirely different that shows up as a sales problem.

Only then do you decide what skills and people you need, and, critically, how those skills interact with each other.

Building a team you cannot hire all at once

The interaction question is where Dougherty's answer gets genuinely useful.

You are not hiring seven leaders in the same week. It happens organically over time. You hire two, you already have someone in place, you are looking for more.

Which means you are solving for the strengths and weaknesses of the people already on the team. His framing: you build the team as a whole rather than hiring individuals.

What Spreedly did and did not need

His worked example is the go-to-market build.

Spreedly has been in market for fifteen years and effectively invented the category, which means substantial brand power already.

So when they expanded the go-to-market engine, brand building was not the requirement. What they needed was velocity and operations: moving from a team built for $20 million in revenue to one built for $100 million.

Which changed what skills to hire for. In a different context they would have needed brand builders. Here he walked into a market-leading brand and needed people focused on operations and scaling.

Why orchestration became open payments

Dougherty's definition of payment orchestration is the simple one: the concept that you can use more than one payments processor at the same time. Spreedly was central to creating it over the past decade or so.

What changed is scope. As the concept matured through the mid-market and enterprise, it expanded well beyond credit card processing to include transaction fraud monitoring, open banking, and know-your-customer and know-your-business checks.

Which is why the category name shifted. Payment orchestration is credit-card-centric. Open payments describes the whole payment stack being multi-provider.

Michael's reaction is the honest marketing verdict: say orchestration and eyes glaze over. Open payments is better.

The market question before the operating question

Asked how he evaluates a company at an inflection point, from $20 million toward $100 million, Dougherty puts a question ahead of anything operational.

Is the market large enough to sustain a billion dollar company?

In Spreedly's case his answer was yes, based on a macro trend in enterprise payments: companies moving away from a single provider, wanting the benefits of best-of-breed while retaining the simplicity of a single integration.

Only once that is validated does the operating work begin.

Can the rails handle agents

The most forward-looking part of the conversation, and the reason this episode will age well.

Dougherty's question is whether the existing payments infrastructure of the world is ready for the volume of transactions that is coming.

His reasoning is simple and hard to argue with. If agents can move, and humans can already buy quickly, and a great deal of commerce happens online, then agents will do it a hundred times faster.

So the question he keeps returning to is not whether agentic commerce happens, but whether the infrastructure underneath it can absorb it. He describes it as a live conversation the company is having, and one they are enthusiastic about.

Two steps forward, one back

Dougherty's account of his own path is unpretentious. He did not end up in the president's seat on purpose. It was a journey, with the privilege and frankly the luck of getting to make a lot of mistakes and learn from them.

His framing of a career: two steps forward, one step back, learning at every step, and having had the privilege of taking that step back many times as well as taking two or three forward.

And his position on people who claim otherwise: when someone says they do not make mistakes or do not know the answer to something, they are lying. What matters is how you take the challenge, repackage it, pivot and learn.

The 5 things I took away from this conversation

1. Ask who before you ask what. Peter's instinct on any business problem is to start with who will be pointed at it. That is the reverse of how most planning works, where the plan is written and the staffing is a consequence, and it is probably why the growth and the flat cost line happened together.

2. Motivation first, then match the problem shape. Understanding when someone feels like they are winning is the input that determines whether they should be on an operational problem or a customer problem. Skills tell you what someone can do. Motivation tells you what they will do well.

3. Existential decides who gets to learn on the job. The one-way and two-way door test, applied to staffing rather than to decisions. Non-existential problems with runway are where people grow. Existential ones are not, and knowing which is which is the leader's job.

4. The sales problem is often not a sales problem. Peter's warning about high-growth companies deciding revenue is soft and therefore the sales team needs rebuilding is one I have watched play out. Find it in the income statement before you find it in the org chart.

5. Agents will transact faster than the rails were built for. This is the question I had not been asking. Everyone is discussing whether AI agents will buy things. Peter's question is whether the payment infrastructure can carry the volume when they do.

FAQ

What is payment orchestration? The ability to use more than one payment processor simultaneously, routing transactions across providers rather than depending on a single integration. Companies adopt it for redundancy, better margins, and expansion into markets a single provider cannot serve well.

What are open payments? The broader successor to payment orchestration. Where orchestration was largely about credit card processing, open payments covers the whole stack being multi-provider, including fraud monitoring, open banking, and identity and business verification.

Why would a company use more than one payment provider? Redundancy, so a single provider outage does not stop revenue; better economics, since routing can be optimized; and geographic coverage, since no single provider is strongest everywhere. Dougherty's framing of the demand is wanting best-of-breed with the simplicity of a single integration.

How do you decide who to assign to a problem? Dougherty starts from what motivates the person and when they feel they are winning, then matches that to the nature of the problem. He uses the one-way and two-way door distinction to decide whether a given problem has room for someone to learn on the job.

Is payment infrastructure ready for AI agents? That is Dougherty's open question. His reasoning is that agents will transact far faster than humans, at volumes the existing infrastructure was not designed for, which makes capacity rather than capability the more urgent issue.

Also mentioned

• Spreedly, its 140-plus gateway connections and $50 billion in annual transaction volume
• Lightspeed Commerce, where Dougherty spent more than a decade before joining
• One-way doors and two-way doors, applied here to staffing rather than strategy
• Open banking and identity verification, the capabilities expanding orchestration into open payments

Listen to the full episode

Peter Dougherty on Between Two COO's

Between Two COO's is hosted by Michael Koenig. Subscribe on Apple Podcasts, Spotify, or wherever you listen.