It Looks Done. It Isn't: The Hidden "Workslop" Tax Draining Your Revenue Team

You have seen it, even if you did not have a word for it.

A deal summary that reads smoothly for three paragraphs and then says nothing you could act on. A quarterly business review deck that is beautifully formatted and structurally hollow. A prospecting email so generically competent it could have gone to anyone, because it essentially did. A competitive battlecard that confidently lists eight features and quietly invented two of them. Each one looks finished. Each one, the moment you try to actually use it, hands the real work back to you.

There is now a name for this artifact, and a fast-growing pile of research on what it costs. Researchers at BetterUp Labs and Stanford's Social Media Lab call it workslop: AI-generated content that masquerades as good work but lacks the substance to move the task forward. Coined in a September 2025 Harvard Business Review article, the term describes something every knowledge worker recognizes on sight. Workslop is not the honest bad draft written by a colleague who is in over their head. It is polished output that looks like progress and behaves like a bill you have to pay later.

For RevOps, sales, marketing, and customer success leaders whose teams now run on AI-drafted work, workslop is the productivity tax you are already paying and have not yet measured. It does not appear on a dashboard. It shows up as slower cycles, quiet rework, and a slow erosion of trust between the exact people your revenue engine depends on to hand work to one another cleanly.

What workslop actually is

The defining feature of workslop is not that it is wrong. It is that it looks right. A struggling employee who turns in a thin, obviously incomplete draft creates a problem you can see and manage. Workslop creates a problem you cannot see until you are already several minutes into trusting it. It has the surface texture of competence: correct formatting, confident tone, plausible structure. Underneath, the actual thinking has not been done. It has simply been moved to whoever opens the file next.

That transfer is the whole story. The person generating the artifact experiences AI as a time-saver, because producing something plausible now takes ninety seconds. The person receiving it inherits the cognitive work that was skipped: verifying the claims, reconstructing the logic, figuring out what the document was supposed to decide. Workslop does not eliminate effort. It relocates it downstream, from someone who was measured on output to someone who was not. The tool made producing volume nearly free. It did not make producing substance free, and workslop is what fills the gap.

This matters enormously in a revenue organization, because revenue work is a relay race. An account executive hands discovery notes to a solutions engineer. The SE hands a scoped proposal to deal desk. Deal desk hands a closed account to customer success. Marketing hands leads and messaging to sales. Every one of those handoffs runs on a quiet assumption: that the artifact you just received is real, and you can build on it without re-doing it. Workslop breaks that assumption one document at a time.

The number nobody put in the AI business case

When your company approved its AI spend, someone built a model. It almost certainly counted the hours saved on the producing side: faster emails, faster decks, faster first drafts. It almost certainly did not count the hours lost on the receiving side. That omission is where the workslop tax hides.

The BetterUp and Stanford researchers surveyed 1,150 U.S. full-time employees. Forty percent said they had received workslop in the previous month. Each incident took an average of one hour and 56 minutes to sort out: to decode what the sender actually meant, redo the missing analysis, or bounce it back and wait for a second attempt. Translate self-reported salaries and time into dollars and the researchers landed on an invisible cost of roughly $186 per employee per month. Scale that to a 10,000-person organization and it becomes more than $9 million a year in lost productivity, a figure that does not even attempt to price the downstream damage to morale and trust.

Look at where the slop travels and the organizational risk sharpens. About 40% of workslop moves peer-to-peer, between colleagues at the same level. But 18% is sent up the chain to managers, and 16% flows down from managers and executives to their teams. That last number is the dangerous one. When leaders forward AI-generated work that looks done and isn't, they are not just wasting an hour of a direct report's time. They are broadcasting the standard: this is what "good enough" looks like here. Slop begets slop.

The reason this tax stays invisible is structural. The saved time is legible and lands on the producer's ledger, where the AI business case is scored. The lost time is diffuse and lands on the receiver's ledger, on a different team, in a different budget line, weeks later. The sender saves twenty minutes. The receiver loses two hours. The net is negative, and no single dashboard is positioned to see it.

The real damage is to trust, not the clock

If workslop only cost time, it would be a nuisance. What makes it corrosive is what it does to how people see each other. The same research found that receiving workslop provoked real feelings: 53% of recipients reported being annoyed, 38% confused, and 22% offended. Those reactions do not evaporate when the document is fixed. They attach to the sender.

The trust numbers are the ones every revenue leader should sit with. Roughly half of employees who received workslop said they now viewed the sender as less creative, less capable, and less reliable than they had before. Forty-two percent saw them as less trustworthy. A follow-up analysis found that 74% said the experience lowered their trust in the quality of that person's work, and about a third said they were simply less likely to want to work with that person again. One hollow AI-generated artifact can quietly reprice a colleague's professional reputation.

In a revenue org, that repricing has teeth. When a customer success manager stops trusting an account executive's handoff notes, they do not flag it in a meeting. They quietly re-run discovery the customer already sat through, which is exactly the redundant, low-value experience that erodes accounts. When a sales manager stops trusting a rep's AI-drafted forecast commentary, they re-interview the whole pipeline by hand, which is the coaching time that was supposed to be freed up. When marketing sends sales a "campaign brief" that is really a prompt output, sales stops reading the briefs. Trust is the operating system of a revenue organization, and workslop introduces silent corruption into it. The cost is not the two hours on any single document. It is the tax that gets levied on every future handoff, because now everything from that person gets double-checked.

Why your revenue team is a workslop factory

None of this happens because your people are lazy. It happens because the conditions are engineered to produce it. Adoption is already near-total: roughly 75% of global knowledge workers now use generative AI, and about 78% of them bring their own tools to work, often unsanctioned, according to workplace research from Microsoft's Work Trend Index. The tools are everywhere, and using them is the default.

The support is not. One recent workforce survey found 41% of workers say their employer has done nothing to prepare them to use AI, with only about one in five reporting they had received comprehensive training. So you have a workforce that has been handed, or has quietly adopted, a machine that produces infinite plausible output, and told to be faster with it, and given no standard for what good actually looks like. The incentive math is brutal and rational: if the mandate is "use AI to move quicker" and there is no defined quality bar, the rational play under deadline pressure is to ship the plausible draft and hope nobody looks closely.

Two more findings complete the picture. Even among people using AI, 52% are reluctant to admit they used it for their most important tasks, and 53% worry that leaning on it makes them look replaceable. So the very people most anxious about their value are quietly using AI on their highest-stakes work and hiding it, which means the output gets less scrutiny, not more. In its January 2026 follow-up, HBR named the culprit plainly: the workslop epidemic is largely a management failure, the predictable result of unclear AI mandates and overwhelmed teams pushed to produce more without guardrails, training, or accountability. Blaming the individual pressing "generate" misses where the pressure came from.

This is a mechanism behind the missing AI ROI

Everyone in B2B has by now seen the uncomfortable headline. MIT's NANDA initiative, in its 2025 report The GenAI Divide, found that 95% of enterprise generative-AI pilots delivered no measurable return, despite $30 to $40 billion in enterprise spending. The usual explanations are about integration, data quality, and change management, and those are real. Workslop is the one that hides in plain sight.

Here is the connection. The productivity a tool creates on the producing side gets clawed back on the receiving side, invisibly, and the two never meet on the same spreadsheet. Your AI dashboard shows adoption climbing and drafts flying out the door, and reads that as value. The value is partly an illusion, because a share of that output is workslop that generates two hours of downstream rework for every twenty minutes it saved. The pilot looks busy and lands nowhere. The gap between "our teams use AI constantly" and "we cannot find the ROI" is, in part, the workslop tax, and it will not close until output quality becomes something you actively govern. Notably, MIT also found that buying AI capability from specialized vendors succeeded around 67% of the time while internal builds succeeded roughly a third as often, a reminder that the discipline around how AI gets deployed matters more than the raw capability.

To be precise about the lane: this is not the abstract "is AI worth it" debate, and it is not a security-governance problem about which tools employees are allowed to touch. It is narrower and more actionable. It is a quality-control failure in the work product itself, and quality control is something revenue operations already knows how to run.

How to stop paying the workslop tax

The fix is not less AI. The organizations getting real returns use it heavily. The fix is holding AI output to the same standard you would hold a human's, and building that standard into the workflow instead of hoping it emerges on its own.

Make the author own the output, never the tool. The single most important cultural rule is that "the AI wrote it" is not an explanation and not an excuse. Whoever's name is on the document owns every claim in it, exactly as if they had typed each word themselves. The moment a team internalizes that they are accountable for what they forward, the incentive to ship unreviewed slop collapses, because their reputation, as the research makes vivid, is what is on the line.

Define "done" before you deploy AI, not after. Most workslop exists because nobody specified what a finished version of the artifact actually requires. Write it down. A deal summary is done when it states the next step, the blocker, and the economic buyer, not when it is 300 fluent words. A battlecard is done when every competitive claim has a source. A definition of done turns "looks plausible" back into "is actually useful," and gives reviewers something concrete to reject against.

Use AI to draft, and keep humans on the decisions. The reliable division of labor is to let AI accelerate the low-judgment parts, first drafts, formatting, summarizing raw notes, and reserve human judgment for anything that commits the company: what to recommend, what to promise a customer, what to forecast. Workslop is what you get when AI is allowed to make the call and a human just forwards it. Reverse that, and the tool becomes leverage instead of liability.

Build verification into the workflow, not after it. Asking people to "double-check AI output" as a general virtue does not work, because the whole point of workslop is that it is fluent enough to survive a glance. Put verification where the work moves: a source-check step before a battlecard ships, a required "so what" line on every AI-assisted analysis, a peer review on customer-facing copy before it leaves the building. Friction at the handoff is far cheaper than rework after it.

Measure rework, not just output volume. If your only AI metrics count drafts produced and time saved, you are measuring exactly half of the equation, the flattering half. Start asking the other question in QBRs and pipeline reviews: how often is work getting sent back, redone, or quietly re-created downstream? Rework is the signal that output volume is hiding. You cannot manage a tax you refuse to measure.

Model the standard from the top. Recall that 16% of workslop flows downhill from managers and executives. Leaders set the quality bar with what they personally forward. An executive who sends the team an obviously AI-generated strategy memo they did not read has licensed everyone below them to do the same. The fastest way to raise an organization's standard is for its most senior people to visibly refuse to pass slop along.

The honest counterpoint

It would be easy to read all this as an argument against AI at work, and that reading is wrong. The productivity gains are real, and the companies in MIT's successful 5% are not the ones using AI cautiously. They are using it constantly, inside disciplined workflows. Workslop is not evidence that the tools do not work. It is evidence that a powerful tool without a standard defaults to producing plausible noise.

There is also a real risk of overcorrecting into something worse. Turning every AI-assisted document into a suspect, launching slop witch-hunts, or quietly banning the tools drives usage further underground, which is precisely how you get more hidden, unreviewed output, not less. The goal is not surveillance or shame. It is a shared, stated bar for quality, applied evenly to work whether a human or a model produced the first draft. The variable that separates leverage from liability was never the model. It is the standard the human holds it to, and that standard is entirely within your control.

What to do this quarter

You do not need a task force to start. You need one honest question, asked in your next pipeline review or team meeting: how much of the work moving between us right now looks finished but forces the person downstream to redo it? The answers will be uncomfortable and specific, and they will point straight at your worst workslop leaks.

From there, pick your single highest-volume AI-assisted artifact, the deal summary, the QBR deck, the outbound sequence, the renewal brief, whichever one your teams generate most, and write one page defining what "done" means for it. Put a verification step in front of it. Tell your team, in plain language, that they own what they send regardless of what drafted it. Then watch the rework rate over the next 60 days.

The workslop tax is already being deducted from your revenue team's productivity and, more quietly, from the trust that lets them hand work to one another at all. The AI is not going to develop that standard on its own. That part was always going to be human, and it is the part that turns a tool everyone is using into a return you can actually find.