The deal that was "on track" for six weeks
Consider an illustrative opportunity sitting in the forecast at the same stage for six straight weeks. Every week the seller reports the same update: demo went well, proposal is out, waiting to hear back. The stage in the CRM says "proposal sent," which sounds like forward motion. Nothing about that stage tells you whether the customer's security review has a date, whether the budget owner has actually agreed to move forward, or whether the deal is quietly stalled behind a dependency nobody on your side controls.
"Proposal sent" describes something your team did. It says nothing about what the customer has agreed to do next. For a VP of revenue or RevOps leader reviewing a forecast, that gap between activity and commitment is where a lot of end of quarter surprises come from.
Why this happens
A pipeline can be configured around the seller's actions: discovery done, demo done, proposal sent, contract sent. That structure exists because it is easy to track and easy for a rep to update. It is not built around the customer's actual decision process, which typically involves steps the seller does not fully control, a security review, a budget approval, a legal sign-off, or a competing internal priority.
Because the stage reflects seller activity, a rep can honestly move a deal forward in the CRM without the deal actually being any closer to closing. This is not usually dishonest reporting. It is a mismatch between what the system tracks and what actually determines whether the customer buys.
Mapping the real buying process, not just the stage
Before trusting a forecast, separate three things for each opportunity expected to close in the period you are reviewing:
- What our team has done, demo delivered, proposal sent, follow up call completed
- What the customer has explicitly committed to do next, with a date if one exists
- What is still unresolved and blocking the customer's decision
This mapping needs to happen at the account level, not just at the stage level. Two deals can share the same CRM stage and be in completely different positions relative to an actual close.
A worked example
Picture an illustrative opportunity: a mid-market software deal expected to close this month. The demo happened three weeks ago. The proposal went out two weeks ago. The rep has followed up twice. On paper, this looks like solid, consistent seller activity, and the stage has not moved backward.
Underneath that activity, the customer's security review has not been scheduled, because the customer's security team has a backlog and has not confirmed a date. Separately, the budget owner attended the demo but has not yet said the number is approved internally. Neither of those facts shows up in "proposal sent." Both of them determine whether this deal closes this month or slips to next quarter.
If the account owner has to state the answer explicitly, either they know the security review date or they do not, either the budget owner has confirmed or has not, the gap becomes visible immediately. Repeating the seller's activity instead of answering these questions lets the deal sit at "on track" indefinitely while the actual blocker goes unaddressed.
The fix: build the customer commitment field into your opportunity record
Add a field to the opportunity record, separate from the stage, that requires the account owner to record the latest customer confirmed next step and its date. Not a hoped for date. A date the customer themselves has agreed to. If there is no such date, the field should say so plainly rather than being left blank or copied from last week.
Pair this with a second field for the specific unresolved dependency, what has to happen before the customer can say yes, and who on the customer's side owns resolving it. This does not need new software. A required field in your existing CRM or a simple weekly note attached to each opportunity is enough to start surfacing the gap.
Review this at your next forecast call. For each opportunity expected to close in the period, ask for the customer confirmed next step and the date attached to it. If the answer is a description of what your team did instead, that is the signal the deal needs more scrutiny before it stays in the forecast at its current probability.
What AI can prepare, and what the account owner still confirms
AI can be genuinely useful here as a first pass, pulling together the permitted records for an opportunity, activity history, meeting notes, email threads, and assembling them into a summary that highlights whether a customer commitment or a customer dependency has actually been mentioned recently. It can flag commitments that have passed their agreed date or opportunities without recent customer evidence, using a review interval suited to your sales cycle.
What it should not do is decide the deal is on track or off track. That judgment belongs to the account owner, who has to confirm the actual status from the customer relationship, not accept a confident sounding summary as a substitute for that confirmation. The role of automation here is to make the missing evidence obvious faster, not to replace the person who has to go get the answer.
A forecast entry that makes the next decision possible
For the illustrative software opportunity, a useful record would say:
- Seller activity: demo complete; proposal sent. Links to the dated records.
- Customer commitment: no confirmed security-review date. Budget approval unconfirmed. Do not substitute a seller's requested date.
- Dependency: security review and budget approval must finish before the customer can proceed.
- Owners: account owner seeks confirmation; customer contacts own their internal decisions. Mark an unknown customer owner as unknown.
- Next review: a date for the account owner to return with evidence or explicitly report that it is still missing.
- Forecast decision: the forecast owner reviews the expected close date and category against these facts. The AI summary does not change them automatically.
The record can say "unknown." Requiring a filled field without permitting an honest unknown encourages a guess. If AI proposes a commitment, make its source and date available beside it, and check that the customer actually made the commitment rather than merely receiving the request.
How to test your forecast for this gap
Pick a defined set, such as every opportunity currently forecast to close in the current month, and for each one ask for the latest customer confirmed next step and the unresolved dependency. Count how many opportunities have a real answer to both questions versus how many only have a description of seller activity.
Do this for a full forecast cycle, not a single spot check, since the same gap tends to reappear the next month if nothing about the process changes. Freeze the forecast set at the start of the review, including each expected close date and whether customer evidence existed at that time. At period end, count the opportunities that closed as expected, slipped, were lost or remained unresolved. Do not remove slipped deals from the starting set or rewrite the old close dates.
For each evidence group, divide opportunities closed within the original period by all opportunities in that group at the snapshot. Show the counts as well as the rate. Keep segments and sales-cycle lengths comparable. Small groups and other differences can explain a gap, so treat the comparison as a signal to investigate, not proof that adding a field caused an improvement.
The common failure to watch for
The most common mistake is adding the new fields to the CRM and then not enforcing them at the forecast review. If a rep can still say "it's on track" without being asked for the customer's confirmed next step, the old habit wins, and the new field becomes one more piece of unused data. The fields only work if the forecast conversation is structured around them every time.
A useful forecast explains what still has to happen and who on the customer's side has to make it happen. It should not let consistent seller activity stand in for a customer's actual commitment to buy. If your revenue team is seeing deals stall at the same stage without a clear reason, that is worth mapping before changing your CRM or adding another sales tool. See how we approach this kind of buyer side visibility problem in AI sales systems, and how we scope any engagement in our method.
To review your own forecast process with FlowChainLabs, Book your call.