I have sat in enough forecast calls to know the ritual.
A rep says the deal looks good. A manager says they feel okay about it. Someone asks if procurement is a risk. Everyone nods in a thoughtful, slightly haunted way. Then the deal slips.
That is not forecasting. That is group meditation.
The teams that get sharper over the next few years will not be the ones with prettier forecast dashboards. They will be the ones running better deal inspection.
The market keeps pointing in that direction. Outreach’s 2025 sales report says opportunities closed within 50 days have a 47% win rate, versus 20% or lower after that threshold, and it also shows a 10-point win-rate lift on larger deals supported by AI-assisted execution. Meanwhile, Salesforce’s 2026 State of Sales announcement says 87% of sales organizations already use AI somewhere in the process and 54% of sellers have used agents.
To me, that means the real question is no longer “Should we use AI in sales?”
It is “Where does AI make the manager smarter?”
My answer: deal inspection.
What deal inspection actually is
Not pipeline hygiene. Not rep surveillance. Not a second CRM.
Deal inspection is a structured way to ask:
is this deal real?
what is missing?
what is being assumed?
what evidence supports next-stage confidence?
That is much healthier than “How do you feel about it?”
The seven-question rubric I’d use
Every meaningful opportunity gets reviewed against the same rubric:
Question | Green | Yellow | Red |
|---|---|---|---|
Is there a real business pain? | quantified and urgent | acknowledged but vague | mostly implied |
Is there a compelling event? | time-bound reason to act | soft timing | no real deadline |
Do we know the buying process? | clear stages and owners | partial visibility | mostly guesswork |
Are we multithreaded? | multiple stakeholders engaged | one thread with some exposure | one-threaded |
Has value been quantified? | ROI / risk case exists | rough estimate only | no business case |
Is there a mutual next step? | scheduled and specific | verbal only | none |
Has risk been surfaced? | objections and blockers named | partial risk view | optimism is doing all the work |
If you only did this once a week, your forecast quality would improve fast.
How AI fits in without becoming obnoxious
I do not want AI deciding whether a deal is real.
I do want AI to prepare the inspection.
Feed it the call notes, emails, meeting summaries, and CRM fields. Then ask it to produce:
a first-pass score on the seven questions
missing evidence
contradictions between notes and stage
suggested next questions for the rep
top slip risks
Now the manager is not spending half the meeting gathering context.
They are spending it on judgment.
That is the right split.
A practical weekly cadence
Monday
AI summarizes every stage-2+ opportunity using the same rubric.
Tuesday
Managers review only the yellow and red deals.
Wednesday
Reps leave each inspection with one concrete action:
book the evaluator
quantify the business case
confirm procurement step
secure the next meeting
surface the blocker
Friday
RevOps reviews patterns:
which stages leak most?
which missing fields correlate with slips?
which objections appear late too often?
Now deal review becomes a learning system.
A concrete example
Let’s say a rep says:
“The buyer loved the demo, and they said this is a priority.”
Cool.
Inspection should immediately ask:
who said it was a priority?
what happens if they do nothing?
who else is involved?
what step is already on the calendar?
what would finance say if we pulled them in today?
If those answers are fuzzy, the stage is probably too advanced.
That is not pessimism. That is accuracy.
My founder take
Forecast theater usually rewards confidence over evidence.
That is dangerous in slower markets.
I would much rather run a team where reps know:
optimism is welcome
but evidence gets the final vote
AI can help here because it is patient. It will review every note. It will notice missing pieces. It will flag contradictions. It will not get socially tired.
Managers still make the call.
They just make it from a better starting point.
What I’d do next week
I would take the top 20 open deals and run one brutal inspection pass.
Not to punish reps.
To answer:
which deals are actually strong?
which deals only sound strong?
what single next action would improve win probability?
That is the beginning of a serious forecast culture.
Not more pipeline adjectives.
Better deal evidence.
