The forecast call, rebuilt around three questions

You have sat in the forecast call where the ninth rep in a row calls a deal “still tracking” and the VP nods along, because 30 deals remain and 40 minutes are left on the clock. Nobody asks when the buyer last replied. By Friday the VP has sent finance a number that no one on the call would bet their own money on.

Leaders in Gong’s 2022 survey of 928 sales and revenue professionals spent an average of 4.9 hours a week on forecasting, and only 22% called their process efficient. Fifty-two percent said their organization regularly misses its forecast by more than 10%. Xactly’s 2024 survey of 405 North American sales and finance professionals found four in five leaders had missed a forecast in the past year.

A status recital rewards the rep who tells the best story. It gives the manager no structured way to separate a deal that moved from a deal that was described well, so the manager ends up grading confidence. Ask three questions, in the same order for each deal, and you put evidence where the story used to be.

Question one: what changed since last week?

Only deals with a change get airtime: a new forecast category, amount, close date, stakeholder or risk. The rep names the evidence behind the change, such as a buyer email, a calendar invite, a signed order form or a redline from legal. A deal with no change gets one line in the notes, and the manager moves to the next one.

Close-date pushes deserve the hardest look, because a pushed date is the cheapest early warning you will get. Ask who moved the date and what they said. If the rep moved it without a buyer conversation, log the new date as a guess until the buyer confirms it in writing.

Question two: what has to be true for it to close?

The rep lists the conditions and, for each one, the person on the buyer side who confirmed it: budget approved, signer named, security review scheduled, procurement engaged. A condition that only the rep believes goes in the notes as a hope.

Give procurement its own line, with a contact name and the date of the last conversation. Forrester’s State of Business Buying 2026, built on a survey of nearly 18,000 business buyers, found procurement professionals are decision-makers in 53% of buying cycles. A commit deal with zero procurement contact is a best-case deal with good posture.

Question three: what are we doing about it before the next call?

The manager closes each deal discussion with one action, assigned to an owner with a due date and typed into the next-step field while the call is live. Next week, the manager opens the call by checking those actions before anyone presents a number. Requests for help belong here too: an executive sponsor call, a pricing exception, a legal escalation or a reference customer.

Cadence and agenda

Run it weekly and keep it about the current quarter. Pipeline creation and next-quarter coverage get their own monthly review, so they stop eating the forecast hour.

• Monday by noon: reps update amount, close date, category and next step, and answer the three questions on each commit and best-case deal.
• Tuesday: managers run 20-minute inspections with each rep on the largest deals and anything that changed category.
• Wednesday: a 45-minute team call. Five minutes checking last week’s actions, five on the number, 25 on deals that changed, five on red flags and five on requests for help.
• Thursday: managers submit commit and best case with a one-line reason per change, and the CRO rolls up for finance on Friday.

Inputs and manager prep

The manager preps from a snapshot comparison of this week’s pipeline against last week’s, filtered to deals whose category, stage, amount or close date moved. That report is the agenda. Before Wednesday, the manager flags any rep answer that lacks a named buyer. AI deal signals such as risk scores and slip predictions go in as prompts for question one, and the rep still owes the evidence.

The CRM fields that matter carry history: close date with a push count, forecast category with change dates, amount history, next step with a due date and the date of the last buyer activity. Pull buyer emails and meetings into the CRM automatically, so the last-activity date records what the buyer did. Ebsta’s 2025 benchmarks found 44% of the contacts sellers interact with are missing from their CRM, and a quarter of those missing contacts are senior decision-makers.

Make closed-lost reasons mandatory while you are in the fields. Clari’s March 2025 study of 10 million opportunity records from more than 100 enterprises found 98% of companies failed to track closed-lost reasons consistently, which leaves you forecasting this quarter without knowing why last quarter’s deals died.

Commit means four things

Publish the definition before the quarter starts. A deal sits in commit only when the buyer has confirmed all four.

1. The close date, in writing, from someone who can sign or knows who will.
2. The signer, by name, who has attended at least one meeting.
3. The paper process, with legal review, security review, procurement and the purchase order mapped to finish before the close date.
4. The amount, matching a proposal or order form the buyer has seen.

Best case holds deals with a credible path and at least one missing condition, named. Track each rep’s commit accuracy by quarter and share it with them. A rep whose commit lands close to the actual has earned lighter inspection.

Red flags

• The close date has moved twice, or has moved into the last two weeks of the quarter.
• One buyer contact has carried the deal for the past 30 days.
• The close date is less than 30 days out and the rep has not met the signer or procurement.
• The category moved up while the stage and next step stayed put.

On one red flag, the manager pulls the deal out of commit until the rep brings new evidence. On two, the deal goes back to pipeline.

Where AI forecasting helps

Put AI tools on capture and pattern work. They log the emails and meetings reps skip. They also flag deals that went quiet and project the quarter from historical conversion by stage. In Gartner research published in September 2024, Steve Rietberg and Tyler Huguley argue that AI can reduce seller burden while increasing forecast accuracy.

Gong’s 2022 survey found organizations using insights from customer interactions were twice as likely to report accuracy within 5% of their monthly forecast as those relying on seller-entered data (22% vs. 11%). Gong sells that capability, so weigh the finding accordingly.

Where it misleads

Models learn from history, so a new product, a price change, a new segment or a new sales motion breaks the pattern they were trained on. Activity volume can pass for momentum when one champion is the only person replying. Sentiment models can score politeness as buying intent. A projection carried to the second decimal looks more certain than a manager’s round number, even when it rests on dirtier inputs.

Bad data makes each of those failures worse. In a Gartner survey released in February 2020, only 47% of respondents believed their organization had high-quality data, and 13% called it poor. Reps also learn what moves a score, and a team that books meetings to lift a deal score has trained the model to flatter it.

Run the AI projection as a third forecaster. Log the rep roll-up, the manager call and the AI number each Thursday, then score all three against the actual when the quarter closes. In Gong’s survey, only 24% of leaders trusted their reps’ forecast commitments and 58% trusted their own. After two quarters you will know which of the three numbers has earned your trust, and you can ask whoever owns the other two to explain the gap.

The first few calls will run short and feel awkward, because a rep with nothing new has nothing to say. Decide before Wednesday whether you will fill that silence with another story or move to the next deal. Your reps will know which one you chose by the third deal.