Payments as a Growth System

Payments as a Growth System

I’ve digested quite a few posts and articles about the Sainsbury’s/Argos deal since my last post and the more opinions I read on it the more it reinforced that the topics I was circling in that article weren’t specific to Argos or Sainsbury’s at all — they’re industry-wide.

That’s led me to a simpler question which I’m sure many people in Payments have asked themselves before: if payments is the bit that gets you paid, why is it so often the bit that comes after everything else when enterprise merchants are investing in their business systems?

The bottom of the funnel problem

Payments gets talked about like it’s the last step of the journey — the bit after the real work is done, handled by whichever gateway is cheapest or whichever provider procurement signed off fastest. It’s filed next to compliance and IT security: necessary, slightly boring, best not to poke it too much because PCI DSS is scary and nobody wants to own an audit finding.

I understand where that instinct comes from. But it’s backwards. Payments isn’t the bottom of the funnel. It’s the connective tissue between every promise a business makes to a customer and every promise it makes to its bank, its regulator and its shareholders. Get the architecture right and it becomes a genuine growth lever — faster checkout, higher authorisation rates, fewer abandoned baskets, cleaner data flowing back into everything else you’re trying to build. Get it wrong, or simply leave it alone because it’s frightening, and it becomes the ceiling on how far any transformation programme can actually go.

I’ve watched this play out from both sides now — inside a major retailer doing exactly this kind of separation and rebuild, and inside a global processor watching dozens of other businesses try to localise, replatform or modernise their acceptance stack. The pattern repeats. The businesses that treat payments as core architecture, sat close to the commercial strategy, are the ones whose transformation programmes actually land on time and on budget. The ones that keep it at arm’s length, bolted on and reviewed once a year for compliance, are the ones still fighting the same integration fires eighteen months later.

The reflective balance

None of this is a call to fix everything at once. Budgets are finite, and no CFO is signing a blank cheque for payments architecture just because I think it deserves more attention. The real challenge isn’t “invest more in payments” as a rule — it’s judgement. Weighing the case for foundational investment against every other pressing priority — the new storefront, the loyalty relaunch, the AI pilot the board has already announced — and being honest about which trade-offs are reversible and which aren’t.

Underinvest in the plumbing and you cap what everything else can achieve. Overinvest in it relative to where the business actually is and you’ve built a beautiful platform nobody asked for, at the expense of something that would have moved the needle sooner. The skill is reading that balance correctly for your own business, not applying a rule of thumb borrowed from the last transformation programme someone happened to work on.

Where this collides with AI

Which brings me to the thing I keep seeing right now, and the reason this matters more in 2026 than it did five years ago.

Agentic AI is the frontier everyone’s scrambling towards. Boards want an AI strategy. Vendors want to sell you an agent. And there’s real capability there — I’ve spent enough time building with these tools to know that. But an awful lot of the companies chasing it haven’t finished the basics. They’re running fragmented architecture where the payments platform can’t talk cleanly to the order management system, which can’t talk cleanly to fraud, which can’t talk cleanly to finance. No agent, however capable, fixes that. Agentic AI orchestrates decisions across systems — and if those systems don’t hold consistent, trustworthy data about a transaction in the first place, the agent just makes bad decisions faster and with more confidence than a human would have.

Put rubbish in, and you get rubbish out. That’s not a new idea, but it gets forgotten the moment “AI” enters the conversation, as if the label alone solves the underlying plumbing problem.

This is the bit I want to dig into properly: the risk of a business skipping pre-season and expecting to win the league — jumping straight to AI before it’s stabilised the fundamentals underneath it. I’m pulling that thinking together into a proper paper over the next few weeks, using exactly this kind of transformation and carve-out scenario as the working example. If that’s a conversation you’re already having on your board or in your transformation programme, I’d like to hear how you’re framing it.

More to come soon.