How to Analyze Deals and Forecast with AI
Most sales forecasts miss because they rely on how reps feel about deals rather than what the data shows. This playbook teaches you how to use AI to score deal health, spot stalling opportunities before they slip, and build pipeline reviews grounded in evidence instead of optimism. You will learn to compare AI projections against your own judgment, investigate the gaps, and present recommendations your manager can act on immediately.
This playbook covers the how. For the why and what, see the
skill definition
.
Developing Start here. Build the foundation.
- Run a weekly pipeline health check every Monday morning. Export your open opportunities from your CRM into a spreadsheet and run this prompt: 'For each deal below, flag any that show warning signs: no activity in the last 14 days, no next step scheduled, single-threaded (only one contact engaged), missing key qualification data (budget, timeline, decision process, or competition). List each flagged deal with the specific warning sign.' Review the output, verify it against your CRM, and address the top 3 at-risk deals before your weekly pipeline review.
- Build a deal scoring checklist with 8-10 criteria that predict whether a deal will close. Include: confirmed budget, identified decision-maker, established timeline, defined decision process, multi-threaded engagement (3+ contacts), recent activity (within 14 days), clear next step scheduled, and competitive position understood. Score each open opportunity 0 or 1 on each criterion and calculate a percentage. Any deal below 50% needs an action plan before you include it in your forecast. Update scores weekly and track whether they correlate with actual outcomes over a quarter.
- Before every pipeline review with your manager, prepare a one-page summary for each deal using this prompt: 'For this opportunity [paste CRM data and recent meeting notes], provide: (1) current deal status in one sentence, (2) the single biggest risk to closing, (3) what has changed since last review, and (4) the specific next action I plan to take with a deadline.' Print or share this summary before the meeting so review time is spent on coaching and strategy instead of status updates.
Proficient Build consistency and rhythm.
- Compare your forecast against AI-generated projections every two weeks. Run your pipeline through a prompt: 'Based on these deal attributes [paste deal data including stage, age, activity recency, number of contacts, and deal size], estimate the probability of each deal closing this quarter. Explain your reasoning for any deal you rate below 50%.' Compare the AI estimates with your own commit/upside/pipeline designations. For every deal where you disagree with the AI by more than 20 percentage points, write a one-sentence explanation of why you believe differently. Bring these disagreements to your manager. They are the most productive coaching conversations you can have.
- Create a deal velocity tracker to catch stalling patterns early. In a spreadsheet, track each deal's stage entry date and calculate days-in-stage for every opportunity. Run a monthly prompt: 'Here are my open deals with their current stage and days-in-stage. Our average sales cycle is [X] days. Flag any deal that has spent more than 150% of the average time in its current stage. For each flagged deal, suggest three possible reasons it may be stuck and one action to unstick it.' Review the output, pick the most likely diagnosis for each deal, and execute the suggested action within the week.
- Build a win/loss pattern analysis at the end of each quarter. Export your closed-won and closed-lost deals with full attribute data and run this prompt: 'Compare these won and lost deals across these dimensions: deal size, sales cycle length, number of stakeholders engaged, number of meetings held, competitive situation, and industry. Identify the 3 strongest predictors of winning and the 3 strongest predictors of losing.' Present the findings to your team with specific examples from real deals. Use the patterns to update your deal scoring criteria for the next quarter.
Mastered Operate at the highest level.
- Build a forecast accuracy tracking system. Each week, record your committed forecast number and the AI-generated projection side by side. At the end of the month, compare both against actual bookings. Create a simple chart showing your forecast accuracy and the AI's forecast accuracy over 6 months. Identify where each source tends to be wrong. If you consistently overestimate certain deal types or stages, build that correction factor into your future forecasts. Share this analysis with your manager to demonstrate data-driven forecast discipline.
- Create strategic pipeline recommendations for your team's quarterly business review. Run your full team pipeline through a prompt: 'Given this pipeline [paste aggregated data], recommend (1) which deals to accelerate with executive sponsorship or additional resources, (2) which deals to de-risk by widening stakeholder engagement or addressing specific objections, and (3) which deals to deprioritize because the probability-weighted return does not justify continued investment. Support each recommendation with specific data from the deal attributes.' Present these recommendations with the supporting evidence in a structured slide or document that your leadership team can act on in the review.
- Develop a rolling 90-day pipeline health dashboard that your team reviews monthly. Track five metrics: pipeline coverage ratio (pipeline value divided by quota), average deal health score, average days-in-stage by stage, forecast accuracy trend, and new pipeline generated versus pipeline closed. Run a monthly prompt: 'Based on these pipeline health trends over the last 3 months, identify the 2 most concerning trends and recommend specific actions to address them.' Use this dashboard to shift pipeline reviews from deal-by-deal status updates to strategic discussions about where to invest selling effort.
Unlock Skill Progression
Coaching Personalized to your current level
Progress Tracking Across every skill area
Mastery Validation Evidence-based, not guesswork