A forecast can be mathematically tidy and still rest on weak opportunity records. The problem appears when a close date has rolled forward for three months, an amount comes from an early estimate, or a stage reflects seller confidence without a recorded buyer action. The forecast calculation cannot repair those inputs.
Salesforce reports that only 35 percent of sales professionals completely trust the accuracy of their organization's data. Clari's 2024 survey of 420 B2B sales and revenue leaders found that 71 percent reported incorrect or hidden forecast and pipeline details. These are survey findings from software vendors, not universal error rates. They still describe a practical control problem: managers need a repeatable way to test the records beneath a forecast.
The phrase sales forecasting data quality describes the condition of the opportunity data used to estimate future revenue. A useful weekly control checks whether each material deal has a current stage, amount, close date, buyer contact, recent activity, and next step supported by evidence. It sends only exceptions to a manager, so forecast judgment stays with the person accountable for the number.
Contents
- What the research says about forecast inputs
- Separate missing, stale, and unsupported data
- Define evidence for each opportunity field
- Run the weekly exception review
- Keep judgment with the forecast owner
- Measure whether the control works
- Use the findings without overstating them
- Sources
What the research says about forecast inputs
Salesforce's published statistics draw on its State of Sales research and report the 35 percent complete trust figure. Trust is a perception measure. It does not establish how many CRM fields are wrong, but it does show that most respondents stopped short of complete confidence in the data available to them.
The Clari 2024 Revenue Leak Report gives a more specific view of pipeline problems. In its survey, 65 percent of senior revenue operations leaders said hidden or incorrect forecast and pipeline details were a main reason their organization experienced what Clari calls revenue leak. Across respondents, 49 percent reported an inability to diagnose deal progression, another 49 percent were unsure where pipeline dropoff happened, and 60 percent reported slipped deals.
Clari defines revenue leak broadly, and it sells revenue software. Its loss estimates should not be treated as an audited financial benchmark for an individual company. The operational survey responses are more useful here than the headline economic estimate because they point to observable conditions in a CRM: details are hidden, progression is hard to diagnose, and dates slip.
Forrester's November 2024 pipeline progression article describes the data needed for sales and marketing coordination. Sellers must maintain stages, buying group members, deal values, and expected close dates before teams can share and act on opportunity insight. The article is not a forecast accuracy experiment. It supports a narrower conclusion: cross-team pipeline work depends on current opportunity fields.
Together, the sources justify a record control, not a promise that one checklist will produce a particular forecast accuracy percentage.
Separate missing, stale, and unsupported data
Forecast reviews often label every problem as bad data. That makes correction harder because three different defects require different action.
Missing data means a required field is blank. A deal has no next step, no expected close date, or no identified contact. The remedy is to find the source evidence or return the record to its owner. Filling the field with a guess only hides the defect.
Stale data was once plausible but no longer describes the opportunity. The expected close date passed, the last activity is older than the team's threshold, or the named contact changed roles. A stale record needs confirmation, a revised value, or removal from the active forecast.
Unsupported data contains a value without enough evidence. A seller may move a deal to proposal because a call felt positive, while the CRM contains no confirmed requirements, buying participants, or scheduled buyer action. The field is populated, but a reviewer cannot reproduce the decision.
This classification gives a sales pipeline virtual assistant a precise preparation role. The support person can locate blanks, calculate age, compare dates, and attach source activity. The opportunity owner then supplies context or changes the record.
Define evidence for each opportunity field
A team should write the acceptance rule before auditing records. Otherwise reviewers make a new decision for every deal and the control becomes another subjective meeting.
| Forecast input | Acceptable evidence | Exception condition |
|---|---|---|
| Stage | Recorded buyer action that matches the stage definition | Stage changed without the required buyer action |
| Amount | Current quote, product scope, or documented estimate basis | Amount has no source or conflicts with the latest scope |
| Close date | Buyer timeline, procurement event, or agreed decision date | Date passed or moved without a note |
| Buying group | Named contacts and their known roles | Material role is unknown or no active contact is recorded |
| Next step | Specific action, owner, and due date | Vague note, missing owner, or overdue action |
| Recent activity | Logged call, email, meeting, or buyer event | No qualifying activity inside the team's age threshold |
The stage rule deserves special attention. Stage names such as discovery or negotiation sound clear but often hide different practices across sellers. Define the buyer action needed to enter a stage and the condition that moves a deal out. A sent proposal is evidence that a document left the company. It is not evidence that the buying group reviewed it.
The amount rule should preserve uncertainty. If scope is not settled, record the approved estimating basis and distinguish it from a signed or quoted amount. A precise number without a source can create more confidence than the deal supports.
Run the weekly exception review
Run the control on the same day each week, before the management forecast call. Start with opportunities expected to close in the current period, then include large deals from the next period and any record whose amount exceeds the team's materiality threshold.
The reviewer exports or filters the required fields and applies the written rules. Each exception gets a defect type, the field involved, the evidence found, the record owner, and a correction deadline. Keep the source link or activity identifier in the CRM when the system supports it. A private audit spreadsheet that cannot be traced back to the opportunity creates another data copy to maintain.
Use three queues:
| Queue | Meaning | Action |
|---|---|---|
| Ready | Required inputs pass the evidence rules | Include for manager review |
| Correctable | Evidence exists, but the record is blank or stale | Update before the forecast cutoff |
| Judgment required | Evidence is conflicting or the deal status is uncertain | Send to the opportunity owner and manager |
Do not let the support reviewer decide that a deal will close. The reviewer decides whether the record meets an acceptance rule. This boundary makes routine checking delegable without transferring commercial judgment.
A weekly cadence also avoids an end of month cleanup rush. Teams looking for practical ownership patterns can use the sales pipeline support warning signs to identify recurring maintenance that is competing with seller time.
Keep judgment with the forecast owner
Forecast ownership belongs to the manager or revenue leader accountable for the submitted number. Record maintenance can sit elsewhere. Confusing those roles either traps managers in routine cleanup or gives a support operator authority the role should not have.
The reviewer can verify that the close date has a source, identify an overdue next step, find a missing contact, and show that the last activity exceeds the agreed threshold. The opportunity owner explains buyer context. The manager decides whether to include the deal, adjust the forecast category, or request more evidence.
Escalation rules should name the decision owner. For example, a correctable field returns to the seller by a set cutoff. A conflicting amount goes to the sales manager. A duplicate account goes to CRM operations. A permissions problem goes to the system administrator. Naming the owner prevents exceptions from becoming a shared list that nobody clears.
Changes also need an audit trail. Record the previous value, new value, source, editor, and time for material forecast fields when the CRM does not already retain field history. The log helps a manager distinguish a genuine buyer timeline change from repeated date rolling.
Measure whether the control works
Forecast accuracy is an outcome measure, but it is affected by market changes and seller judgment as well as record quality. Pair it with process measures that the weekly control can influence directly.
| Measure | Calculation | What it tests |
|---|---|---|
| Required field completion | Records with all required inputs divided by records reviewed | Missing data |
| Stale close date rate | Reviewed records with passed or unsupported dates divided by records reviewed | Date maintenance |
| Evidence pass rate | Material fields meeting written evidence rules divided by fields checked | Support for entered values |
| Overdue next step rate | Active deals with an overdue next action divided by active deals reviewed | Follow through |
| Correction turnaround | Median time from exception to accepted correction | Ownership and response |
| Date movement frequency | Count of close date changes per opportunity | Repeated slippage |
Track the measures by week and owner, but do not use a small sample to rank sellers. The first goal is to find workflow defects. If many records lack buying group roles, the stage criteria or CRM layout may be at fault. If corrections arrive after the forecast cutoff, the review schedule or ownership rule needs attention.
Sample the accepted records as well as the exceptions. A monthly check of records marked ready can catch a reviewer who treats populated fields as sufficient evidence. That check protects the control from becoming a box filling exercise.
Use the findings without overstating them
The three sources use different methods and answer different questions. Salesforce reports sales professional perceptions across its research sample. Clari reports responses from 420 B2B sales and revenue leaders and frames them through its revenue leak model. Forrester offers operational guidance on pipeline progression. None of them proves that a particular weekly audit will increase revenue or achieve a fixed forecast accuracy rate.
Apply the evidence to the problem it can support. The surveys establish that trust, visibility, progression, and slippage are common reported concerns. The Forrester guidance identifies fields that cross-team pipeline work needs. The control in this article is an operational inference built from those findings.
Test it against the team's own records. Set the field rules, run the review for four forecast cycles, and compare exception rates, correction time, date movement, and forecast outcomes with the starting baseline. Keep the control if records become easier to verify and managers spend less time reconstructing deal history. Change it if the checklist produces work without revealing material exceptions.
If your forecast meeting begins with missing fields and stale dates, book a free consultation. SalesSupportStaff can help define the preparation queue, evidence rules, and escalation path while your managers retain forecast authority.
Sources
- Salesforce, Sales Statistics. Source for the reported share of sales professionals who completely trust their organization's data.
- Clari, 2024 Revenue Leak Report. Survey of 420 B2B sales and revenue leaders used for reported pipeline visibility, progression, and slippage findings.
- Forrester, Three Critical Steps Frontline Marketers Must Take to Support Pipeline Progression. Source for the operational fields needed to coordinate pipeline progression.