What Most RevOps Efforts Get Wrong

One client had more than 113,000 account records in Salesforce. We tried to connect those records to identifiable companies and couldn't find a match for 70% of them. That was the data behind every report the company produced.

The surveys suggest this is a familiar problem. Openprise and RevOps Co-op's 2025 State of RevOps survey found that 99% of practitioners struggle with technical data issues. Just 11% described their data as excellent. In Validity's 2025 survey of 602 CRM users, 76% said less than half of their CRM data was accurate and complete.

The central mistake in most RevOps efforts is building reports before verifying the records behind them. Teams put money into dashboards and forecasts, and now into AI agents that use the same unproven data. Precise-looking reports give those numbers a credibility the underlying records haven't earned.

Mistake 1: Treating reporting as the job

Dashboards, forecast rollups and board decks are the outputs most RevOps teams get measured on. The work follows those measures. When a new CRO wants more visibility, the response is a better dashboard. The data feeding it rarely gets the same scrutiny.

RevOps should reach reporting at the end of the work. Think of a dashboard as a window into a room: making the window larger won't tidy what's inside. It will give you a clearer, more colorful view of the same mess.

Matching and validating data seldom produces something a leader wants to look at, so the budget goes elsewhere. You don't hear "we matched 40,000 accounts to real companies" in a board presentation. Quarter by quarter, the reporting improves while the records beneath it deteriorate.

Mistake 2: Measuring the CRM instead of the market

A dashboard rate has a numerator and a denominator. In RevOps, that denominator usually comes from the CRM: accounts worked, leads created or opportunities opened. The market itself never enters the calculation.

We worked with an IT-services firm whose offer fit 18,775 companies. Salesforce contained about 24% of that market. At the same time, 41% of its Salesforce account records represented companies outside the offer's fit. Reports used a CRM that left out most of the market while including plenty that didn't belong in it.

Coverage has to be measured before RevOps can report it. With only a quarter of the market in the CRM, conversion rates, win rates and pipeline ratios describe that quarter alone. The remaining three-quarters are absent from the reports, and that absence goes unnoticed.

Mistake 3: Treating what reps typed as what buyers did

CRM entries usually come from people who have a reason to make them. A booked demo, a new opportunity, an advanced stage or a checked "pain confirmed" box tells you what a seller did or believed. It doesn't establish a buyer's decision.

Leadership at a B2B fintech called its 67% demo-to-opportunity conversion "very strong." Yet 80% of those opportunities stalled in the first stage, which was where SDRs earned their pay, and 5% closed. The dashboard numbers were accurate. What they captured had been shaped by the compensation plan.

Incentives aren't the only issue. Validity found that 37% of respondents said employees at their companies fabricate data "to tell the story they think decision-makers want to hear." That story then gets rolled up by RevOps, turned into a forecast and presented to the board.

When a RevOps team uses seller-entered records as evidence of buyer progress, its reporting describes the sales team's paperwork rather than the market.

Mistake 4: Nobody owns the seams

A typical revenue stack combines the CRM with engagement, conversation-intelligence, data, forecasting, routing and billing tools. Large B2B enterprises in the 2026 LXA and LeanData survey used an average of 37 tools. Even so, 74% favored best-of-breed tools over a single platform.

Each system handles the data within its own boundaries. Engagement software relies on its activity log; forecasting software relies on opportunity fields; data providers rely on their contact records. Vendors respond to data problems by asking customers to bring more of the stack onto their platform. Gartner now groups many of these vendors in the Revenue Action Orchestration category.

The gaps occur where those systems meet. An activity never reaches the CRM. A meeting has no associated opportunity. A stage moves forward without a buyer event, or the CRM deal value disagrees with billing. Because no vendor owns both ends, none takes responsibility for the break. RevOps is left reconciling spreadsheets ahead of forecast calls, often with the reconciliation logic held in one person's head.

Salesforce's State of Sales, Seventh Edition found that 51% of sales leaders using AI said technology silos delay or constrain their initiatives. Time spent reconciling is only part of the cost: these gaps also obstruct the work built on that data.

Mistake 5: Letting AI write to the record without a log

People and administrator-defined rules used to be the only sources of CRM writes. Several vendors' AI agents can now change the same records at once. They update fields, log activity, send outreach and move stages.

LeanData's 2026 survey of its customers shows why that matters. Among 157 go-to-market practitioners, 93% had deployed at least one AI agent. Nearly one in three didn't know how many agents were acting on their records. Thirty percent had discovered record actions with no audit trail, and 27% had seen multiple tools or agents reach out to the same prospect.

Dan Gottlieb at Gartner captured the risk: "If those systems are fragmented, the agents will scale the fragmentation." Adding agents to an unverified record makes it harder to trust that record and defend the numbers drawn from it.

What to do instead: verify before you report

RevOps still needs to report. First, though, it needs to understand the data. Teams commonly begin by defining meetings, opportunities and stages. Those definitions come too early: you need to know what you have before deciding what to count. I'd establish the market, assess the records and then work through the following sequence.

1. Build a clean TAM

Begin with the market. Specify the conditions a company must meet to buy your offer, then use a company database to build the list. Don't derive it from the CRM. Include each company once at the parent level and assign a standard company identifier to every record. Leave uncertain fits on the list, labeled unknown.

What good looks like: a maintained, clean list covering every company that fits your offer, ready to serve as the reference for the rest of the work.

2. Assess your data against that TAM

Assess where the data is reliable and where it falls short. Use identifiers to match CRM accounts to the TAM; names aren't enough. Connect each account you can validate to an actual company, count duplicate records and identify those with no match.

A clean TAM makes that assessment useful immediately. A record linked to a company in the TAM can be validated; an unlinked record is an exception to investigate. Without editing a field, you can already distinguish what belongs in the dashboards from what should be excluded.

What good looks like: you can demonstrate the match rate, duplicate count and portion of the CRM that falls outside your market.

3. Let what you found drive the definitions

With the assessment in hand, write definitions that cover the records you actually have. Decide which accounts belong in reporting, which records to exclude and the reasons for exclusion. Define a meeting, an opportunity and a stage. Sales, marketing and finance should contribute; one person should document the decisions and maintain them.

What good looks like: each number on the executive dashboard has a documented, versioned definition grounded in records you can validate.

4. Verify the numbers you report

Present reported and verified values for every number sent to the board. The reported value comes from the CRM. Its verified counterpart is the portion you can substantiate: linked to a TAM company and an actual buyer action within a specified window, with evidence someone outside the team can check. Begin with open pipeline and determine how much of its value meets that test. That proportion is the number's trust score, the most honest measure in the building.

What good looks like: board-level figures display the reported amount, the verified amount and the limits of what each figure can establish.

5. Reconcile the revenue chain across tools

Trace each deal from touch to meeting, opportunity, stage, booking and invoice. At each transition, compare the systems on either side. Identify missing CRM activity, meetings without opportunities and stage advances without buyer events. Include repeatedly shifting close dates and values that disagree with billing. Resolve the source of a break so fixing the record isn't the end of the work.

What good looks like: you find breaks within a day and prevent the same ones from recurring.

6. Log every write, whoever makes it

Maintain a single log for changes to revenue records. Identify the writer of each change, whether a person, an automated rule or an agent from any vendor. Raise an alert if multiple agents act on the same prospect or an agent replaces a rep's entry.

What good looks like: no writes go untracked, and you can answer "who changed this?" for every consequential field.

7. Then build the dashboards

At this point, build the reports with verified and reported values shown together. Expect the dashboards to look less impressive initially: some of their previous figures weren't true. A smaller figure that holds up to scrutiny is more useful than a larger one you can't substantiate.

What good looks like: forecast calls begin with verified pipeline and focus on buyers' actions.

It's never going to be perfect

There is no final cleanup after which the data stays clean. Companies merge and people move jobs. Reps keep entering records, while each added tool or agent contributes its own. Be skeptical of anyone selling a project with a promise of perfect data at the end.

The achievable goal is much better data whose quality you can measure. Matching against a clean TAM validates a large share of the records straight away. The others become identifiable exceptions rather than disappearing into averages. Keep matching incoming records, reviewing exceptions and updating the TAM as the market changes. Over successive quarters, more of the figures you report can be verified.

Knowing which numbers you can defend, and increasing that proportion each quarter, matters more than reaching perfect data.

The question for your next forecast call

When your team presents pipeline this week, what proportion can it support with evidence of a buyer action in the last 30 days?

Without an answer, the forecast review is a review of the team's story.

Here's what to take away:

• Most RevOps work puts reports on top of unverified records. Improving those dashboards makes the underlying problem appear more credible.
• The five mistakes are making reporting the job, using the CRM as a substitute for the market, equating rep entries with buyer behavior, assigning nobody to the gaps between systems and allowing unlogged AI writes.
• Work in sequence: establish a clean TAM, assess the CRM against it, write definitions based on that assessment, verify reported figures, reconcile the revenue chain and record every write. Build dashboards after that work. Data will never be perfect; what should improve each quarter is the share of your numbers you can defend.

Verify before you report.

Three pilot Revenue Truth Audits

Growth Orbit is running three pilot audits this quarter for companies with a CRM and at least two other revenue tools. Over two weeks of read-only work, we compare reported and verified pipeline, identify the top 20 breaks in the revenue chain and show how much of the market is visible in the CRM.

Pilot engagements are priced from $7,500 to $12,500, depending on scope and data complexity. Request a Revenue Truth Audit.