The deals that close share a small set of facts you can observe on the calls: the person who controls the budget has joined a meeting, the buyer has described how they'll get a contract signed, there is a plan with dates both sides agreed to, and more than one person on their side is talking to you. Check those four things against what was actually said, not the stage field, and your forecast gets more honest.
Why the stage field lies
Stages describe what the seller has done. Demo delivered, proposal sent, verbal received. None of those are things the buyer did. A deal can sit in Proposal for four months with a champion who loves the product and no path to a signature, and the stage will say Proposal the whole time. Rep confidence has the same flaw: it tracks how the last call felt, and the last call is usually with the friendliest person at the account.
The fix is to inspect buyer behavior instead of seller activity, and calls are one of the richest and least filtered records of buyer behavior. Notes are filtered through the rep's optimism. Transcripts and scorecards record what was actually said.
Four buyer-side signals worth inspecting
Every qualification framework has more boxes than this, and the extra boxes matter for coaching. For forecasting, four signals do most of the work.
- The economic buyer has been in a meeting. Not "the champion says the VP is supportive." The person who can approve the spend has joined a call, heard the business case, and said something about it. If you can't quote them, you don't have this signal.
- The paper process has been described by the buyer. Ask "walk me through what happens between you deciding yes and a signed contract" and listen for specifics: security review, legal redlines, a procurement portal, who signs. A buyer who can describe it has done this before and intends to do it again. A buyer who shrugs has never bought anything this size.
- There is a mutual plan with dates. Both sides have agreed to a sequence of steps ending in a signature, each with an owner and a date, and the buyer has completed at least one of their steps on time. Plans the seller wrote alone don't count.
- You are multithreaded. At least three people on the buying side have been on calls, including someone outside the champion's team. Single-threaded deals die when the champion changes jobs, gets busy, or loses an internal argument you never heard about.
A deal with all four has materially stronger buyer evidence than its stage alone suggests. A deal with fewer than two signals is still a lead, no matter how many demos you've run.
Step 1: score every open deal against the four, from the calls
Do this once as a cleanup and then weekly as a routine. For each deal above a size threshold you set, answer the four questions with a quote or a "no". The quote comes from a call transcript, a scorecard, or an email the buyer wrote. It does not come from the rep's memory.
The first pass is usually uncomfortable. Expect a good share of the deals in your commit category to have one signal or none. That isn't a rep problem yet. It is the first honest look at the pipeline you've had.
Step 2: run the weekly inspection on the deals, not the reps
Pipeline reviews turn into interrogations when the manager asks "are you confident" and the rep defends the deal. Change the question. Go signal by signal: "Has the economic buyer been on a call? Show me." "What did she say about the paper process?" The rep either has the evidence or goes to get it, and the meeting becomes a planning session about which meeting to book next.
Keep it to thirty minutes for a team of eight. Deals with four signals get a sentence. Deals with two or three get the discussion. Deals with fewer than two get moved, which is the next step.
Step 3: decide what to do with deals that fail the test
A deal without the signals isn't dead, but it isn't forecastable either, and pretending otherwise is how quarters get missed. There are three options, and the rep should pick one on the spot.
- Go get the missing signal. Book the meeting with the economic buyer. Ask the paper process question on the next call. Send a draft mutual plan and see whether the buyer edits it. Give it two weeks.
- Move it out of the forecast. Keep working it, but stop counting it. This is the most common right answer and the one reps resist most.
- Close it lost and write down why. If the champion can't get you a meeting with anyone else after three asks, you've learned something about the account.
Whichever you choose, the deal should look different in two weeks. If it doesn't, the choice was wrong.
Step 4: make the signals visible without a spreadsheet
This routine dies if it lives in a spreadsheet someone updates by hand. The four signals need to be fields on the opportunity, filled from the calls as they happen, with the supporting quote attached. Then the weekly review is a filtered pipeline view sorted by signal count, and the "show me" question answers itself.
If your call notes are already being turned into scorecards, the signals are a subset of what a MEDDPICC or SPICED rubric captures. Economic buyer, decision process, and champion strength map directly. Multithreading is a count of distinct buyer-side attendees across calls. The mutual plan is the one signal that usually needs a manual field, because it lives in a shared document rather than in a conversation.
Where this goes wrong
- Accepting hearsay as a signal. "My champion says the CFO is on board" is not the CFO on a call.
- Inspecting activity instead of buyer behavior. Meetings held and emails sent tell you how hard the rep is working, not whether the buyer is buying.
- Letting the rep grade their own deal. The evidence has to come from the record of the call, or the exercise becomes the old confidence question with more steps.
- Running the review monthly. Deals move in days. A monthly inspection catches problems after the quarter is already decided.
- Keeping the cleaned-out deals in the forecast "just in case". It hides the real number from the people planning around it.
Where Aircover fits
Aircover scores every call against MEDDPICC, SPICED, or your own rubric using live transcription, so economic buyer engagement, decision process, and champion strength show up as fields with the buyer's words attached. That gives you most of the signal count per opportunity without anyone maintaining a spreadsheet. The MEDDPICC page shows how the scoring works during and after the call.