Shadow accounting has long been a familiar problem in sales organizations. When sales reps don’t have clear, timely visibility into their commissions and progress, they fill the gap themselves, traditionally using spreadsheets. Those homebrewed spreadsheets are often inaccurate and inefficient, but they reflect a real need. Reps have to understand where they stand and what they will earn.
Shadow accounting has always been frustrating, but it’s something compensation admins have known how to deal with. When disputes arose, they generally had a playbook for how to respond.
But in recent years, generalized AI tools like ChatGPT, Gemini, and Claude have become widely available, offering quick, confident answers to almost any prompt. If your “shadow accountant” reps aren’t already using AI, they will be soon, and that shift completely changes the dynamics.
Sales reps can now offload tracking, calculations, and dispute arguments to AI, receiving plausible-sounding answers in seconds, often without realizing the risks they’re introducing. For admins and compensation teams, this means a surge of harder-to-resolve issues, with little precedent or established guidance for how to manage them effectively.
Once shadow accounting becomes AI shadow accounting, many organizations may find themselves missing the old spreadsheets.
In this article, we’ll look at:
- The cause and costs of shadow accounting before AI
- Why AI shadow accounting raises the stakes
- How organizations can address both with the right tools
Shadow accounting before AI
Shadow accounting is the practice of sales reps keeping their own unofficial records to track commissions, progress toward goals, and expected earnings alongside the company’s official systems.
Shadow accounting is a well-documented issue, and it has been for years. Most sales leaders and compensation teams have encountered it, and many have taken steps to reduce its impact. But it’s worth briefly recapping why shadow accounting happens and what it traditionally costs, as that baseline makes clear just how much AI is changing things.
Why sales reps engage in shadow accounting
Sales reps engage in shadow accounting for one simple reason: they need clarity. To plan their work and pace their efforts, reps must understand how daily activity translates into quotas, bonuses, and take-home pay.
When official tools or reports don’t provide timely, detailed, or trusted information, reps are left guessing. Few things are more demoralizing than believing you’re on track, only to discover at the end of a period that the numbers didn’t line up.
Shadow accounting gives reps a sense of control. They create their own records, usually in spreadsheets, to track deals, estimate commissions, and double-check official figures. These records are usually less accurate than the systems they’re meant to supplement, but from the rep’s perspective, imperfect information feels safer than no information at all.
That’s why shadow accounting should be understood as a symptom of a broken system, rather than willful misconduct. It emerges when transparency and trust break down.
The traditional costs of shadow accounting
Shadow accounting has always carried real costs.
Maintaining personal tracking systems wastes time, pulling reps away from selling. Homegrown spreadsheets are built on partial information and informal interpretations of complex plans, making errors and mismatches inevitable. And those mismatches lead to disputes requiring investigation, explanation, and resolution.
This creates friction across the organization. Reps begin to question whether the numbers can be trusted. Admins and finance teams are pulled into recurring reconciliation work. Morale and confidence in the compensation process suffer.
We’ve covered the costs of shadow accounting in depth elsewhere, and they remain a serious concern. But traditional shadow accounting operated within known limits. It required effort. It exposed its assumptions. And when things went wrong, there was usually a visible trail to follow.
Those limits largely disappear once shadow accounting becomes AI driven.
AI shadow accounting introduces new risks
AI shadow accounting is the practice of sales reps using generalized AI tools to calculate commissions, interpret compensation plans, or generate dispute arguments outside the organization’s official compensation systems.
Generalized AI changes shadow accounting from a visible, effort-heavy workaround into something fast, opaque, and much harder to control. When reps turn to AI chatbots, the same gaps that once drove spreadsheet-based shadow accounting now produce larger risks that are less visible and more complex to resolve.
Hallucinations replace visible calculation errors
Accuracy has always been a challenge with shadow accounting, and that remains true in the age of AI. But when sales reps relied on spreadsheets to track commissions, they at least left a paper trail. Errors could be traced back to a misplaced figure, a faulty assumption, or a broken formula. It would take time and effort, but the logic was there to examine. But AI doesn't show its work.
When a sales rep asks an AI chatbot to calculate commissions, the model isn’t applying real compensation logic. It’s inferring that logic from whatever context it was given. If information is missing, contradictory, or ambiguous, the AI fills in the gaps on its own and delivers a response that sounds confident and authoritative, even when it’s wrong.
That gap-filling behavior is fundamental to how large language models work. When an LLM lacks sufficient context, it often generates a plausible answer rather than signaling that it doesn’t know. The model is designed to produce something that sounds reasonable, even when the underlying information is incomplete or wrong.
And the frequency of that behavior can be significant. In benchmark testing across specialized tasks, some LLMs have been shown to produce fabricated or incorrect outputs more than 30% of the time. Those outputs rarely explain their assumptions or intermediate steps. And even when you explicitly ask for them, there’s no way to verify that the steps weren’t generated after the fact to justify a conclusion.
This shift from visible, inspectable errors to invisible hallucinations makes disagreements harder to diagnose, explain, and resolve.
Hidden AI usage corrodes trust
There are two ways shadow accounting can take shape with AI. Sales reps can either be open about their use of AI, or they can use those tools without disclosing it.
Neither approach is ideal. But when reps are transparent about using AI, managers and admins at least have context. They can push back on incorrect assumptions, clarify what AI should and shouldn’t be used for, and begin setting clear policies around acceptable use.
The greater risk is when AI use goes undisclosed—and that appears to be the more common scenario. In one recent survey, 59% of employees admitted to using unapproved AI tools at work without their employers’ knowledge. Hidden use of AI is a reality most organizations are already dealing with.
That lack of visibility quickly undermines trust. Managers don’t know what they’re evaluating. Reps feel frustrated when conclusions they believe are well founded are challenged. And conversations become defensive instead of collaborative.
Compensation depends on shared logic and shared facts. When AI enters the process unofficially, that shared foundation starts to erode. Without transparency about how conclusions are formed, trust breaks down, and every dispute becomes more difficult than it needs to be.
Unauthorized AI use risks security and privacy violations
While traditional shadow accounting already carried security risks (particularly if sales reps stored sensitive information in personal spreadsheets), those risks were relatively contained. Files lived on individual machines or were shared internally, with limited exposure beyond the organization. AI shadow accounting opens up a whole new avenue of exposure.
When reps use generalized AI tools to check commissions or prepare disputes, they’re uploading potentially sensitive information—like deal values, customer names, contract terms, payout structures, or internal performance metrics—into external systems with limited privacy guarantees. Once that data leaves the system of record, it’s no longer governed by the organization’s access controls, audit trails, or security policies.
Most sales reps aren’t trying to bypass safeguards or expose confidential data. They just want to understand their pay. But generalized LLMs aren’t designed to handle compensation data safely, and they make no distinction between low-risk questions and highly sensitive financial details. A simple prompt to double-check a commission can introduce significant compliance, contractual, and regulatory exposure. And leadership may not even be aware that it’s happening.
Dispute volume increases
Shadow accounting used to take a good bit of effort—which meant it had built-in deterrence. Reps had to create and maintain spreadsheets, interpret complex compensation plans, track deals over time, and apply their own formulas, which meant that only the more motivated issues tended to turn into formal disputes.
That’s no longer the case with AI.
Today, a rep can paste a few details into a chatbot, ask a question, and within seconds receive a polished response that appears to explain what happened and why a payout might be wrong. The answer may be much less accurate than their spreadsheet would have been, but it's presented with confidence and seems actionable.
As a result, the barrier to raising a dispute drops dramatically. When generating an explanation becomes easy, more explanations will be generated. Even routine curiosity about progress or expected earnings could turn into a perceived issue once an LLM produces a convincing explanation, increasing dispute volume and putting additional strain on compensation teams.
Disputes become harder to resolve
With traditional shadow accounting, disputes were usually grounded in visible logic that reps had to think through themselves. A spreadsheet might misapply a rate, omit a deal, or misunderstand a rule, but the rep could generally explain how they arrived at their conclusion. Admins could walk through the assumptions, identify where things broke down, and resolve the issue based on shared facts.
With generalized AI, that shared footing disappears. AI-generated conclusions lack a clear chain of reasoning, so the rep submitting the dispute may understand their argument as little as the chatbot does. When asked to explain their logic, they may just repeat what the AI said, without having reasoned through the compensation plan or the underlying data.
That puts admins in a trickier position. Instead of reconciling two interpretations of the same logic, they have to reconstruct the entire scenario from scratch just to establish a baseline. The assumptions and gaps in the AI’s response aren’t visible, and there’s no reliable way to interrogate how the conclusion was reached. Of course, admins won’t always recognize AI-generated disputes—and the rep may not ever mention that their reasoning is simply, “Claude told me so.”
When admins finally craft a thoughtful response, nothing stops reps from going back to the chatbot for a counterargument to that response. Or the next response. Or the next.
This means longer resolution times and more back-and-forth. Even when the official payout is correct, disproving an AI-generated claim can feel adversarial and unsatisfying. Disputes stop being collaborative efforts to understand the numbers and become exercises in dismantling conclusions that were never grounded in the compensation system.
Admins are tempted to use LLMs in response
Generalized AI tools are appealing because they promise quick answers with minimal effort, and that appeal isn’t limited to sales reps. Compensation admins are no more immune to the temptation than anyone else, especially when the work in front of them keeps getting harder.
As AI shadow accounting becomes more common, admins feel the effects directly. Disputes arrive more frequently, and each one takes longer to resolve—all while deadlines and day-to-day responsibilities remain the same. In such an environment, AI can feel like a natural solution. If more disputes are coming in, then using an LLM to draft a response can seem like the obvious way to keep up.
You can easily end up with AIs endlessly arguing with each other through humans too busy to keep them in check. Disputes drift further away from the compensation system. Instead of being grounded in real plans, rules, and calculations, they become an exchange of competing machine-generated explanations. In the long run, this adds even more work, as admins are forced to untangle conclusions that are increasingly disconnected from reality.
AI itself isn’t the enemy here, and the impulse to reach for AI is understandable. The real issue is context.
To produce accurate, trustworthy results, AI needs access to the full compensation plan: the rules, effective dates, eligibility logic, adjustments, and how all of those pieces relate to one another. When that context is missing, the model fills in the gaps on its own. That’s where hallucinations come from in generalized AI chatbots: they’re being asked to reason about complex incentive compensation plans without the context they need and without being designed for that level of specificity.
At Performio, we’re integrating purpose-built AI inside the compensation platform, where that context already exists. Because it’s designed for this environment, our AI understands your actual plans, data, and configuration, so it can explain payouts, support dispute resolution, and guide responses based on real logic rather than inference.
Performio’s purpose-built AI gives admins a way to manage growing complexity without creating yet another shadow process to clean up later.
Legitimate disputes get drowned out
As frustrating as disputes can be, some are far more valid than others. Sales reps who identify real errors that actually affect compensation are doing the company a service by bringing them to light. These are the disputes compensation teams want to see and resolve, because they improve accuracy and trust across the organization.
As AI shadow accounting lowers the barrier to generating disputes, legitimate issues get buried in the noise. Increased dispute volume makes it harder to quickly separate real problems from weak or unfounded claims. Admins end up spending more time just establishing a baseline understanding, leaving less time and attention for the disputes that really matter.
A steady stream of low-quality claims can make admins more skeptical of disputes as a category. Reps with legitimate concerns feel like they’re being brushed aside. And trust erodes on both sides, even if no one is acting in bad faith.
AI shadow accounting prevents the right disputes from receiving the attention they deserve.
Morale and retention suffer
The cumulative effects of AI shadow accounting take a toll on morale. Sales reps feel confused about their earnings, frustrated by disputes that go nowhere, and unsure which explanations they can trust. Those who raise legitimate concerns feel ignored or lumped in with weaker claims.
Compensation teams feel overwhelmed by rising dispute volume, harder investigations, and conversations that seem increasingly adversarial. Admins feel like they’re constantly on the defensive, forced to justify correct outcomes instead of focusing on improving the system. Everyone ends up doing more work with less clarity and fewer satisfying outcomes. And trust erodes as the process becomes increasingly unreliable.
When compensation is unpredictable, motivation suffers. Reps lose confidence in the connection between performance and pay. Managers struggle to coach effectively when they can’t clearly explain outcomes. And that uncertainty undermines engagement and productivity, especially among high performers who expect transparency and fairness.
The long-term risk is retention. Talented reps don’t stick around in environments where compensation feels confusing or contentious. Experienced admins burn out under the weight of constant disputes and reactive work. What starts as an attempt to save time creates a culture of frustration that’s difficult to reverse.
How Performio eliminates the need for AI shadow accounting
Shadow accounting—whether with spreadsheets or AI—is a response to a broken system. Sales reps aren’t being intentionally reckless; they’re simply filling in the gaps when they haven’t been given the clarity they need to feel confident in their commissions.
So the solution isn’t to impose a blanket ban on AI, but to provide a better alternative. When you remove the conditions that make shadow accounting necessary, and ensure AI is used safely and responsibly, the temptation to rely on risky external tools fades. AI belongs inside the compensation system, where it can operate with the right context and controls.
Performio’s incentive compensation management software gives sales reps the transparency they need to understand their earnings without resorting to parallel tracking. It equips admins with tools to ensure accurate payouts and streamlined dispute resolution. And it integrates purpose-built AI that was specifically designed for incentive compensation.
Performio gives sales reps full, trustworthy transparency
Performio gives sales reps real-time visibility into their performance and compensation. Reps can see what they’ve earned, how close they are to their goals, and how individual deals contribute to their payouts. That clarity removes the uncertainty that drives shadow accounting in the first place.
When reps don’t have to guess, they don’t need workarounds. There’s no incentive to maintain spreadsheets, upload data into external AI tools, or rely on unofficial interpretations. Instead, reps can go to their dashboards to see the same data and logic the organization uses to calculate their pay.
This shared visibility builds confidence and removes the need for shadow accounting. Reps spend less time second-guessing outcomes and more time focused on selling.
Performio equips admins to efficiently respond to disputes
Performio gives admins a clear, authoritative view into how compensation outcomes are produced. Instead of digging through spreadsheets, configurations, or disconnected systems, admins can trace payouts directly to the plan components and transactions that shaped them.
When disputes arise, admins don’t have to debate interpretations or dismantle opaque conclusions. They can walk through exactly what happened and why. Legitimate issues can be identified and corrected quickly, while faulty claims can be addressed directly, without escalation or defensiveness.
When they’re working in Performio, admins spend less time in reactive triage mode, freeing them up to focus on the strategic and operational priorities that actually move the business forward. By giving admins clarity and control, Performio ensures compensation remains fair, consistent, and trustworthy, even as plans grow more complex.
Performio’s purpose-built AI provides a safe, accurate, and reliable alternative to generalized AI tools
AI can play an important role in incentive compensation, but only when it’s applied in the right places and with the right constraints. The problem with AI shadow accounting isn’t the technology itself. The problem is that generalized AI tools are being asked to interpret compensation calculations they don’t understand.
Incentive compensation management software should include AI, but it has to be implemented the right way. As Performio CEO Grayson Morris explained in his recent open letter:
“The divide won't be between those who have AI and those who don't. It will be between software applications that have the architecture to support AI, and those that are simply layering a chatbot over a mess of custom scripts or formulas.”
Long before the current wave of AI hype, Performio invested in a standardized, extensible ICM core with a component-based architecture that models incentive compensation in a way both humans and AI can understand.
Instead of relying on custom scripts or formulas, compensation plans in Performio are assembled from well-defined components with clear intent. Plans, rules, transactions, and adjustments all live in a consistent data model that captures not just the math, but the business logic behind it. That structure gives AI the context it needs to reason accurately about how compensation actually works.
As a result, Performio’s AI doesn’t have to infer meaning or fill in gaps. It explains outcomes based on real configurations and data. When a payout looks unexpected, the system can trace it back to specific plan rules, eligibility logic, or adjustments and explain the result. Every answer is grounded, traceable, and verifiable.
By keeping AI inside a governed, purpose-built platform, Performio avoids the risks introduced by AI shadow accounting. Sensitive data stays protected. Outputs can be trusted and explained. And AI reduces noise and workload instead of amplifying confusion.
Used this way, Performio’s AI strengthens the compensation process rather than undermining it.
Don’t let AI create a bigger shadow
You don’t have to avoid AI entirely, but you do need to be sure it’s being used in the right place, with the right context, and with the right guardrails. When reps have transparency, when admins have confidence in the numbers, and when AI is implemented correctly, AI shadow accounting (along with regular shadow accounting) loses its reason to exist.
Performio was built to provide that clarity and control. By combining real-time visibility, rigorous compensation logic, and purpose-built AI designed for incentive compensation, Performio helps organizations eliminate both traditional shadow accounting and its AI-driven successor.
To see what Performio can do for your organization, request a demo today!
Patrick McCarville is a Solutions Engineer at Performio, where he helps enterprise organizations modernize their incentive compensation processes with a focus on scalability, accuracy, and cross-functional alignment. With years of experience in the sales performance management space, Patrick brings a practitioner’s perspective shaped by his prior role as a Senior Sales Compensation Consultant and Account Manager at OpenSymmetry. There, he led complex implementations, guided global teams through compensation system transformations, and partnered with HR, Finance, and Sales leaders to align comp strategy with business goals.