Revenue Operations and the Ghost State

The Second Number #

A salesperson once added a field to Salesforce called "Real Close Date."

The system already had a Close Date field. It was required on every opportunity and appeared in the forecast, the pipeline report, and most conversations about the quarter. Salespeople updated it during weekly reviews, especially when their managers were watching. Real Close Date lived a few columns over, and that was where the salesperson entered the date they actually believed.

I found the field while trying to understand why opportunities kept moving backward in time. At first it looked like a data-governance problem. Two fields describing the same event, one of them unofficial, created by someone without permission and used inconsistently across the team. The ordinary response would have been to remove it. Then I compared the two and found the unofficial field was more accurate.

I have since stopped being surprised by it. I have seen versions of that second field across companies and years, reinvented under different names by people who never met each other, which is the part worth paying attention to. This essay is written from that accumulated view rather than from any one company's records. The scenes are composites, the pattern recurring, the particulars no one's in particular.

The field had no executive audience, no connection to compensation, and no role in the story being told about the quarter. Nobody had asked it to stay optimistic through Friday's pipeline call. Close Date was the answer the salesperson gave the organization. Real Close Date was the answer they gave themselves, and the second one was what they planned their week around.

So they were maintaining two versions of every deal, one for the forecast and one they could actually use, at roughly twice the effort of keeping a single set. The private field sat three columns from the field it was hedging, built in the company's own software, available to anyone who scrolled right.

Nobody scrolled right. The more useful assessment was sitting in the system the whole time, unauthorized and unrecognized, which meant the company had no way to act on it and the salesperson had no way to raise it without admitting what the official field was for. Revenue Operations tends to appear inside that distance.

Three Pipeline Numbers #

I once watched a company prepare for a board meeting where three teams brought three different pipeline numbers, each produced in good faith.

Sales had built its number from individual seller forecasts and manager judgment. Marketing had counted pipeline associated with campaigns, using an attribution model the team had spent months refining. Finance had applied historical conversion rates and several adjustments that made sense once the analyst walked through them slowly.

The disagreement began as clarification. Someone asked whether renewals were included. Someone else questioned the attribution window. A spreadsheet opened in the corner of a laptop, followed by a quiet recalculation. For a few minutes, everyone behaved as though the numbers would converge once the definitions were cleaned up, and then they didn't.

The company had developed parallel descriptions of the same quarter. Each one answered a slightly different question, served a different audience, and carried the incentives of the team that produced it. None was obviously false, and taken together they were unusable.

The meeting could have continued. Meetings usually do. A number would have made it onto the slide, partly because a board deck requires one and partly because choosing is easier than explaining why no single answer exists.

This is where RevOps often gets described as the owner of data quality. The description is technically correct and still misses most of the work. The harder task is deciding which differences matter, which can coexist, and which prevent the company from acting coherently.

A pipeline number used to plan sales capacity may need to differ from one used to evaluate campaign performance. Problems begin when those numbers share a name, travel without their assumptions, and arrive in the same room pretending to describe the same thing.

What the Spreadsheet Knows #

The easiest places to see these fractures are usually outside the official system.

A Customer Success manager maintains a renewal tracker because the dashboard runs two weeks behind. A regional sales leader keeps a spreadsheet containing the opportunities they expect to slip, then updates Salesforce only after the forecast call. Finance builds a separate bookings model because contract structure can't be represented cleanly in the CRM.

From a distance, these systems look redundant. They create duplicate work, inconsistent definitions, and dependencies on files maintained by one person. They also tend to appear at exactly the point where the official system stops helping someone do their job.

Rather than treating every shadow system as rebellion, I've learned to read it as evidence of where the formal process has lost contact with the work. Sometimes the cause is technical, and the required field can't express the distinction an operator needs. Sometimes the official system updates too slowly. Often the problem is social: a person will write an honest assessment in a private file that they would never enter into a field reviewed by their manager, the CFO, and the compensation team. The system asks for a forecast while quietly punishing uncertainty, so people create somewhere else to think.

That distance contains useful information, and it also creates risk. A private tracker can preserve nuance that a standardized field would flatten. It can also become the only place where anyone understands the business, and the more accurate it becomes, the more dangerous its isolation becomes. RevOps usually enters after both facts are already true.

Cleaning the Signal #

The obvious solution is to bring the unofficial information into the official system. Add the field, define the process, assign an owner, build the dashboard. Sometimes that works. Other times the act of formalizing the signal changes it.

The honest renewal-risk tracker becomes less honest once it feeds an executive report. The second close date loses its value once managers begin asking why it differs from the first. A qualitative note becomes a dropdown, then a metric, then a target. People learn how the new field will be interpreted and adjust what they enter.

The organization gains consistency while losing some of the candor that made the information useful. This is one of the less comfortable parts of systems work, since better structure doesn't automatically produce a better description of reality. Structure changes behavior, especially when the information being collected has consequences, and a clean field can end up less truthful than a messy spreadsheet.

Leaving everything informal creates a different failure. Knowledge remains trapped with individuals. Definitions drift. The latest version becomes difficult to locate. When the person maintaining the tracker leaves, the company discovers that an important operating process was stored in someone's habits.

RevOps lives between these outcomes. The role has to make information durable without draining it of meaning, standardizing enough for the company to act while preserving enough local context for the signal to remain worth acting on. There is no permanent calibration.^[1]

The Official Version #

Where RevOps reports influences how that calibration gets made.

Under a CRO, the function stays close to the frontline and may absorb the optimism of the commercial organization. Under a CFO, it may privilege auditability and historical performance while discounting information that arrives through judgment. Under a CEO, it can become a clearinghouse for every disagreement the rest of the organization has failed to resolve. None of these arrangements produces neutrality. The official number is usually the number produced by the system with enough authority to end the conversation.

RevOps participates in that authority. It chooses definitions, establishes cutoffs, decides which exceptions deserve accommodation, and determines which local practices should be removed. Those choices may be reasonable, although they still shape the reality the company later experiences as objective. I've seen definitions become more rigid because the quarter needed a stable forecast, and I've seen them soften because the rigid version stopped resembling how deals actually moved. Both decisions could be defended.

That is what makes the role political, even when the work appears technical. A stage definition changes who gets credit. A territory rule determines whose judgment prevails. A churn policy can alter the apparent health of the business without changing a single customer outcome. The math may be correct while the boundary around the math is chosen.

RevOps can't stand outside that process and guard an untouched version of the truth. The function is part of the machinery that produces the company's shared description of itself, and the responsibility comes from making those choices visible enough to be questioned.

Memory at the Edges #

Organizations preserve outcomes more reliably than reasons. The territory model remains while the argument that shaped it disappears. Approval thresholds survive long after the pricing environment that justified them. A field stays required because removing it feels riskier than asking why it exists.

Shadow systems often carry some of this missing context. An extra column records the exception the official model can't explain. A note beside a formula preserves the reason for an adjustment. A private tracker contains categories inherited from a problem that everyone else has forgotten. Separated from their history, these artifacts can look irrational, and some of them are, while others are scars from a collision nobody wrote down. The challenge is learning which kind you've found.

RevOps becomes a form of institutional memory here, though memory alone isn't enough. Old decisions don't deserve protection merely because they're old. The work is to recover enough of the original conditions to understand whether a rule still protects the company or has become another source of drift, which requires curiosity before cleanup.

The Room Changes #

Most of the time, these questions remain hidden beneath ordinary work. Someone requests a dashboard. A leader asks for a new stage. Finance wants the forecast reconciled before the board meeting. A salesperson complains about an approval step. The request arrives in the language of process, and then the conversation starts.

The dashboard request turns out to be a disagreement about ownership. The new stage is an attempt to give uncertainty a place to sit where someone else can see it. The forecast problem reflects two teams using the word "pipeline" to mean different things. The approval complaint reveals that people have been routing around a pricing policy for six months.

The nature of the job changes once the underlying disagreement becomes visible, which can make RevOps look strangely inconsistent from the outside. On one day, the work is technical. On another, it resembles mediation, organizational archaeology, or diplomacy conducted through field definitions. The activities differ because the fractures differ.^[2]

Sometimes that means creating one definition. Sometimes it means allowing two definitions to remain, then naming the difference clearly enough that nobody mistakes one for the other. Shared reality doesn't require everyone to see the business from the same position. It requires enough translation between positions that the organization can still decide what to do.

The Cost of Being Right #

The person closest to the data often sees a distortion before anyone wants it named. A deal sits in commit past any defensible reading of its history, held there by someone who has been right about exactly this before. The conversion rate everyone quotes comes from the best quarter, which was also the quarter the model was built in. Somebody redrew a territory to keep one relationship intact and the quota math never caught up. There's a renewal risk the account team hasn't documented, because documenting it would put it in front of the customer's procurement lead. Discount thresholds stay where they are because lowering them would mean auditing every exception granted since the last pricing change. None of these were stupid when they happened. Correcting the number may be simple, while naming what produced it is harder.

There is a third container, and it's a person. The field held what the forecast couldn't. The spreadsheet held what the dashboard couldn't. What neither can hold is an assessment that has no evidence behind it yet, and those go to whoever has demonstrated they won't immediately do something with them.

Some version of this job is being the psychiatrist for the go-to-market organization. The comparison is narrower than it sounds. What matters about that room is that saying the true thing in it carries no immediate consequence outside it, and the absence of consequence is what makes the next session possible. The same arrangement operates here. The disclosures arrive before the evidence would let anyone act on them. A champion has left and the deal is being carried by someone who does not return calls, weeks before that shows up anywhere in the pipeline. A renewal is already gone in the judgment of the person closest to it, while every official signal still reads green. A regional leader has stopped believing the number and wants to know, without putting it in writing, what you intend to do about that. An attribution model survives because it once won an argument nobody remembers having. None of this arrives through process. It arrives because people have learned that telling you something does not automatically make it happen to them.

That access is part of the system, not a courtesy extended around it, and it exists on a condition. What builds it is the accumulated evidence that disclosure doesn't automatically become exposure, and that evidence can only be assembled out of instances where you had something and didn't use it. Each quarter you sit with what you know and don't convert it into consequence, the disclosures get earlier and less guarded, which is the same as saying the function gets more accurate. Spending it works differently. Escalating well can leave you with more standing than you started with, but you find out at the moment you commit, and the version of the access that consisted of never having escalated is gone either way.

The closest thing I've found to describing it is the flagger at a road crew, holding a paddle with SLOW on one face and STOP on the other. There is no authority in the stick. It is a person standing in front of moving traffic, and it works only because drivers accept the signal as binding. Turn the sign for something that didn't warrant it and they start rolling through.^[3]

Which sets a trap for anyone patient enough to fall into it. There is always a case that this quarter isn't the one. The distortion is small, the relationship is worth more than the correction, the number will probably self-correct by March. That reasoning is available every quarter and can't be tested in the present, which means the person exercising careful judgment and the person who quietly stopped being willing to escalate years ago look alike from outside and, more uncomfortably, from inside. Wait long enough and the access has become something else, which is being pleasant to talk to.

The cost isn't incidental to the role. A version of the function that paid nothing would have nothing to spend, and the restraint that makes an escalation land is the same restraint that looks like complicity while it's happening. I've held things back for a quarter and been right. I've held things back for a quarter and watched the correction cost more than it would have in October. Neither felt different at the time.

One Usable Description #

The company with three pipeline numbers did eventually choose one for the board deck. It wasn't the discovery of a hidden, objectively correct answer. We agreed on a definition, documented the exclusions, and made the adjustments visible. The other views remained because they served purposes the board number couldn't.

Marketing still needed to understand campaign influence. Sales still needed manager judgment. Finance still needed a model that connected pipeline to the rest of the operating plan. The improvement came from knowing which number was answering which question.

For a while, the descriptions met. That is about as stable as shared reality becomes inside a company. Incentives continue pulling. People leave. Markets change. A definition that once clarified the business begins to obscure it. Someone builds a spreadsheet to recover what the official system no longer sees.

Then the second number appears again. Sometimes it appears as a field. Sometimes it's still only something a person will say to one other person and hasn't yet found a place to sit. Either way, its existence doesn't automatically mean the system has failed. It means someone has encountered a gap between the description they were given and the decision they have to make.

The unofficial field may be wrong. It may be self-serving, outdated, or understood by only one person. It may also contain the first honest description of something the company hasn't learned how to say in public.