Why does fixing the order-to-cash process basics matters more than implementing AI?

The AI era is forcing leaders to rethink what “transformation” really means. Yet in many successful businesses, order-to-cash, a process that turns customer demand into revenue and revenue into cash, is still held together by manual re-keying, fragmented documentation and informal workarounds.

The result is subtle operational drag: disputes that linger, invoices that slip, forecasting that can’t be trusted, and cash that arrives late or is applied by hand. Fixing this isn’t about silver bullets. It’s about seeing the process end-to-end, restoring ownership across teams, and rebuilding the foundations, where automation supports the business rather than disrupting it, and AI becomes a multiplier on top of a process that works.

Walk into almost any large organisation, and you will find two stories being told at once.

The first is the story we like to tell ourselves: transformation is accelerating, data is the new currency, and the next decisive advantage will come from intelligence; faster decisions, smarter automation, predictive insight.

The second story is the one playing out, quietly, in the operational backrooms of the same business: orders arrive in formats that don’t quite fit; the “single source of truth” turns out to be three sources and a spreadsheet; exceptions are handled by the same handful of people because they know how the long, manual and often error prone process works.

But which is true? It is possible for both stories to be true. In fact, it is increasingly common. That is what makes the current AI moment so interesting. Boston Consulting Group’s latest AI Radar report shows 65% of CEOs rank accelerating AI among their top three priorities (bcg.com, 2026).

I understand the pull. I share the belief that AI will change how work gets done, and at speed. But I also spend time with organisations where, in one of the most business-critical flows in the supply chain, the fundamentals are still operating on goodwill and manual effort. The basic mechanics of turning demand into revenue, and revenue into cash, can be far less modern than the rhetoric suggests.

In very normal, very successful businesses, this is still happening every day. Orders come in by email, PDF or portal. Someone re-keys them. Shipping notes don’t quite match the order. Invoices go out late or with errors. Disputes bounce between teams. Cash is applied manually. And a lot depends on a few people who “know how things work”.

We are talking about optimisation and intelligence while still struggling with a basic operational question: How do we deliver the right things, invoice the right things, and get paid, without manually managing every step along the way?

Part of the problem is ownership, or rather, the lack of it. Orders, invoices and cash do not live in one team.

• Sales cares about getting the order in and pricing.
• Operations cares about shipping the product and efficient resource planning.
• Finance cares about invoicing and getting paid.
• Customer service deals with disputes.
• IT owns the systems underneath it all.
• Each team is doing its job.

The issue is that the end-to-end flow often belongs to no one. When there is no single view of the full chain, the process ends up held together by informal glue: email threads, spreadsheets, manual troubleshooting, manual exception management, and people filling in gaps that shouldn’t exist. APQC benchmarking data cites organisations with manual processes spend 3-times more per invoice than automated organisations (APQC, 2023)

This can appear workable for a long time. But it is fragile. It works until the business starts selling more products, into more countries, through more channels. Then the improvisation that was once manageable becomes a constraint.

The aspirational business growth is actually exposing where the seams are starting to split.

At this point, automation and increasingly AI often get positioned as the answer. Yet these issues are not hard to fix because the technology doesn’t exist. They are hard because fixing them means changing how teams work together, understanding processes that are often hidden or wrongly assumed, and surfacing workarounds people rely on. In other words, order-to-cash problems are rarely “software problems” first.

They are;

• design problems.
• governance problems.
• accountability problems.
• data problems.
• And they sit in the gaps between functions.

This is where AI narratives can become unhelpful. If orders are inconsistent, invoices don’t line up, and data is late or incomplete, AI does not help. It just gives you faster answers to the wrong questions. It accelerates the visible layer without repairing the foundations beneath.

There is a phrase that I find myself returning to: you can’t optimise what you can’t trust. If the organisation cannot trust the order data, cannot trust that fulfilment events are captured consistently, cannot trust that references align across documents, then any “intelligence” applied downstream becomes guesswork. It may look sophisticated, but it will not be reliable. In financial operations, unreliability is not a minor inconvenience. It becomes a risk.

The unglamorous truth is that the work that changes everything is rarely exciting. It is the work of making inputs clean and consistent, reducing manual exceptions, reducing manual touchpoints, and creating visibility across the whole flow. It is the work of building a transaction process that can scale.

That process starts earlier than most people assume. It starts at order capture, where the organisation decides whether it will accept ambiguity as a normal cost of doing business or whether it will insist on clarity upfront. Clean, ERP-ready orders are not a nice-to-have. They are the first control point in the chain. If the order is inconsistent, every downstream step becomes more expensive: fulfilment becomes harder to reconcile, invoicing becomes harder to validate, disputes become harder to resolve, and cash becomes harder to apply.

The process continues through fulfilment, where the business needs a consistent thread of evidence that what was ordered is what was delivered, and then onto invoicing, which should be the controlled output of a reliable flow rather than a manually created task, built from partial information. And it continues through dispute handling and cash application, where the organisation either has structured workflows and shared visibility, or it relies on inboxes and memory.

This work is not glamorous precisely because it is foundational. Yet it is also the work that unlocks everything else. When friction reduces, people feel it quickly. When the basics become stable, automation becomes support rather than disruption. And when the basics become stable, then AI begins to make sense. Not as a rescue plan, but as a multiplier.

The hardest part of order-to-cash is not usually the endpoints. Leaders can describe how an order should be taken and how revenue should be recognised. They can describe invoicing, payment terms and collections. The challenge comes in the messy middle: everything between customer intent and cash in the bank, where the lived process diverges from the diagram.

This messy middle is where complexity accumulates.

• It is where small inconsistencies compound. It is where a missing reference becomes a delayed invoice, which becomes a dispute, which becomes withheld payment.
• It is where a unit-of-measure mismatch becomes a pricing discrepancy, which becomes an operational correction, which becomes a customer complaint.
• It is where a late shipment triggers a partial invoice, which triggers a short pay, which triggers manual cash allocation, which triggers incorrect ageing reports, which triggers poor forecasting.

If the messy middle is invisible, improvement becomes guesswork. Organisations tend to focus on what they can see: a new billing tool, a new analytics layer, a new automation pilot. But the real impact comes from making the flow visible end-to-end and then identifying where friction causes the most damage.

That is why I believe the right starting point is not “fix everything at once”. It is to make informed decisions about what to fix first, based on reality. Fixing these processes means making the real process visible, identifying where friction causes the most damage, deciding deliberately where to fix the problem (whether in the ERP, in EDI, in middleware, or in the way teams collaborate) and then sequencing improvements so each change reduces the burden on the next.

There is another reason the basics matter more now than they did even a few years ago: compliance expectations are moving towards greater structure and auditability.

Across Europe, the shift towards digital reporting and e-invoicing is accelerating. The European Commission announced the adoption of the VAT in the Digital Age (ViDA), (Eurpean Comission, 2025) package on 11 March 2025, with progressive rollout until January 2035. Whatever your view on policy, the operational implication is clear: invoice data and transaction reporting are becoming more standardised, more controlled, and more scrutinised over time.

In the UK, the government’s consultation response on promoting electronic invoicing states that, as announced at Budget 2025, the UK will introduce mandatory e-invoicing for all VAT invoices from 2029. Professional bodies have reinforced that direction and outlined how the requirement is expected to apply across VAT transactions in scope.

This matters because it reframes the “basics” as something more than operational housekeeping. Clean, consistent, traceable transactions are becoming a compliance requirement. Audit trails are no longer a by-product; they are a design goal. In that environment, informal processes that rely on emails and spreadsheets are not just inefficient. They increasingly look like a risk.

It is worth pausing on the psychology here. AI is exciting. It signals progress. It promises speed. It makes leaders feel they are moving with the market. The basics of order-to-cash are the opposite:

• They are painstaking.
• They require cross-functional alignment.
• They expose uncomfortable truths about how work actually happens.
• They surface workarounds that people rely on.
• They often demand decisions about standards, ownership, and governance that have been avoided for years.

That is exactly why they are so powerful.

This isn’t just about buying tools. It is about deciding that friction isn’t “just how things are” and that people’s time and the business’s cash are too valuable to waste. It is about accepting that the most effective transformation programmes feel “boring” in the best way: fewer exceptions, fewer disputes, fewer frantic chases for information, and a more reliable view of performance.

And it is about sequence. When organisations try to optimise before they stabilise, they create frustration. When they stabilise first, optimisation becomes natural. This means putting the basics in place so automation doesn’t feel threatening; it is a reliever of friction, frustration and failures. When the basics are in place, AI becomes not a headline, but a practical capability applied safely to clear use cases.

Fixing order-to-cash isn’t about small tweaks or silver bullets. It’s about understanding the real process end-to-end and making informed decisions about what to fix first within the process as a whole. The reason I insist on that framing is that it returns transformation to its proper place: not as a technology race, but as a discipline of operational design.

Everyone already knows the process isn’t great. The difference is choosing to face it end-to-end, and to improve it deliberately with one informed decision at a time. That choice is what separates organisations that build resilience and cash performance from those that simply add more layers of tooling on top of friction.

AI will undoubtedly change the way we work. But if we want it to matter in order-to-cash, we need to be honest about the foundation it sits on. Fix the basics, and AI becomes a multiplier. Ignore the basics, and AI becomes a faster route to confusion.

The most advanced organisations are not the ones chasing the newest idea first. They are the ones who have the discipline to make the fundamentals coherent so that when they do invest in intelligence, it has something reliable to work with.

While the pressure to “accelerate AI” is continually rising, that may be the most quietly radical leadership decision of all.

Ready to improve your order-to-cash workflow? Click the button below and schedule a 20-minute discovery call.