Revenue platforms currently have a reputation issue. When something goes wrong, it’s typically assumed to be a technical issue: “There was a bug in the workflow, there was an integration that didn’t work, there was a field that’s missing, there was an automation that didn’t fire.” While these things are indeed problems, they are rarely what’s causing the underlying issue. The underlying problem usually lies elsewhere, in what passes for the calm layer where business principles coalesce into assumptions, approvals morph into myths, and systems start enforcing decisions nobody remembers agreeing to. This is the realisation that came not as a result of a production issue, but because of the regular check of approval queues, which started acting weirdly. Nothing was explicitly problematic, no errors, no failed jobs, or notifications or warnings of any kind, but the deals were behaving erratically; some of them were getting closed in a split second. Others were hanging on without rhyme or reason. That’s not a software problem. That’s just a governance problem. When Systems Enforce the Rules Better Than the Organisations Remember Them Revenue systems are rare; they not only transfer data from one place to another but also convey intent. Each approval route, eligibility determination, discount policy, contractual term, and partner qualification is a result of an individual’s decision at some point in time, which is often conditional upon a specific set of circumstances. These choices accumulate over the years. An exception added regionally to fill a quarter. A temporary provision included to prevent an important merger. A manual override introduced since the data was not yet ready. Each individually makes sense in its own right. Together, they constitute something that can only be termed policy debt. These systems, unlike humans, never forget. They continue to enforce rules despite the context having ceased long ago, due to the fact that the rules exist on multiple platforms, contract systems, CRM layers, approval engines, and data pipelines, which means the drift goes unseen until the business feels the friction that it cannot comprehend. In this particular case, the mean time to move a standard agreement through legal approval had crept up by over 30% in twelve months, despite no change in policy. The system was working as intended. The organisation had simply forgotten why it had designed it this way. Why Dashboards Lie About Governance Health Throughput is measured in most revenue dashboards through cycle time, volume, and conversion rates. Of course, those are useful, but they’re all lagging indicators. They tell you that something is slow but never why it is slow. Governance failures are not represented as peaks. Governance failures are represented as variance. Two identical transactions went completely different ways., approvals piled up for no reason that anybody could explain, aanual reviews were introduced, when automation should have been present. Such patterns can be easily overlooked since they have the appearance of edge conditions. Actually, they are indicators of an internally inconsistent rule system. At this scale, inconsistency is costly. In one of the environments I was operating in, the fact that manual legal review was creeping back in on a significant percentage of transactions was costing several million dollars a year in lost savings that the automation was supposed to provide. The system was still working. It simply wasn't working as it was supposed to anymore. Governance as Architecture The error most organisations make is believing that governance is a matter layered on top of systems rather than embedded in them. Procedures are written out in documents, exceptions are addressed at a meeting, and systems are supposed to adapt. That inversion will not scale. The only way to regain predictability is to make governance, just like architecture in software development, have clearly delineated boundaries, ownership, versions, and carefully managed change. Which begins with asking difficult questions. Which are decisions that should be deterministic, and which should remain discretionary? What are the forms of approvals related to risk management, and those linked to lack of trust? Which rules remain relevant and consistent with the actual way revenue is made? When these questions can be answered honestly, the technical tasks become easy. Approval flows shorten, duplicate checks go away. Data models become simple. In one scenario, when the number of conditional approval paths was reduced by approximately 40%, there was no reduction in compliance but a significant reduction in cycle time and forecasting uncertainty. This progress was not caused by any new tools. It came from removing rules that had ceased to make sense. The Hidden Cost of Manual Overrides Manual intervention appears safe, It provides the sense of control that we humans possess. Manual intervention may be considered essential in revenue systems due to the need for flexibility. It’s also the site of governance failure. It creates an implicit rule with each manual override that is not codified anywhere. Over time, the implicit rules far surpass the number of explicit rules. The system falls back upon institutional knowledge, which is the weakest link that an architecture could depend upon. Whenever the override levels had crossed even 15 to 20% of total transactions in the environment, the accuracy levels of forecasts materially diminished. By then, the data used for subsequent models just represented the enforceable policy rather than policy and ad-hoc decisions. This is a concept of entropy masquerading as agility. In some quarters, this compliance philosophy has been characterised as One myth is that more robust governance results in slower revenue. The exact opposite is usually the case. When an environment is more predictable, more people choose to comply because they don't feel the need to find ways to work around what is going on. Standardised clause libraries, hard thresholds for approvals, and clear rules on eligibility logic reduce risk not by adding friction, but by removing ambiguity. Audits become easier. Reviews become faster, and as a result, trust improves. In another modernisation project that I worked on, the result was a substantial reduction in unapproved variations of contracts, once the system made the right process easier than doing it the wrong way. The response time from lawyers also went down, though this happened due to a change, not an increase, in risk tolerance, since the system was finally consistent with actual policies, rather than exceptions that had accreted. The compliance actually improved as the system ceased to surprise people. What Revenue Platforms Actually Need to Optimise for Most GTM technology discussions focus on speed. Faster approvals. Faster contracts. Faster onboarding. Speed matters, but predictability matters more. Revenue leaders are not worried about slow systems. They are worried about systems which behave erratically under Stress. The aim here is not speed. The aim here is predictable speed. This comes from being able to align three different things that are normally handled independently: policy, data, and operations. When these three things get out of alignment, nothing can compensate for the lack of alignment, even automation. The harsh reality is that too many revenue models fail not because they are underscaled, but because they are over-governed in the wrong areas and under-governed in the right areas. Repairing this would demand less ambition and more discipline. Fewer features. Fewer exceptions. Fewer heroic rescue missions. More simplicity. The Lesson That Stuck One of the key takeaways when you think deeply about revenue systems is how they operate like financial ledgers and less like software products. Small errors compound. What was decided in the past is important, and reversibility is expensive. Treat governance as architecture. Design it intentionally. Revisit it regularly. Delete what no longer reflects reality. Do that, and the systems become less brittle. Deals cease to stall out because someone or some system decided they needed to. Forecasts quickly settle into place. Teams are less likely to point at tools as scapegoats for what isn’t a technology issue in the first place. This is not change; It is maintenance when it is done with conviction, and maintenance, in revenue-generating systems, is where the magic lies.

This story contains AI-generated text. The author has used AI either for research, to generate outlines, or write the text itself. 

Dash

The Architectural Mistake That Turns GTM Platforms Into Unreconcilable Ledgers

About Author

Comments

TOPICS

THIS ARTICLE WAS FEATURED IN

Related Stories

Why Partner Data Became My Toughest Engineering Problem

What Are Convolution Neural Networks? [ELI5]

The Noonification: Have U Been Pwned? (1/12/2023)

Goldman Sachs, Data Lineage, and Harry Potter Spells

People are still crazy about Python after twenty-five years

10 Questions to Consider when Setting up a Corporate A.I project

Why Partner Data Became My Toughest Engineering Problem

What Are Convolution Neural Networks? [ELI5]

The Noonification: Have U Been Pwned? (1/12/2023)

Goldman Sachs, Data Lineage, and Harry Potter Spells

People are still crazy about Python after twenty-five years

10 Questions to Consider when Setting up a Corporate A.I project

Light-Mode

Classic

Newspaper

Minty

Dark-Mode

Neon Noir

Minty

HN StartUps