Bottoms-Up Forecast

A bottoms-up forecast builds the revenue prediction from individual deals. Instead of "we'll grow 20% this quarter" (top-down guess), you start with "here are 47 deals, here's each one's probability, here's the math." The forecast is only as accurate as your deal-level assessments.

The principle: every forecast is a probability-weighted sum. The quality depends on two things: accurate deal amounts and honest probability assessments. Most teams get the amounts right and the probabilities wrong. Fix the probabilities and the forecast fixes itself.

The Forecast Formula

Basic weighted forecast

Forecast = Sum of (Deal Amount × Close Probability)

Example:
  Deal A: $50K × 80% = $40K
  Deal B: $30K × 50% = $15K
  Deal C: $25K × 30% = $7.5K
  Deal D: $40K × 20% = $8K

  Weighted forecast: $70.5K
  Upside (if all close): $145K
  Commit (80%+ deals only): $40K

Forecast categories

Building the Forecast

Step 1: Qualify every deal

For each deal in pipeline, answer these questions:

Step 2: Assign probabilities

Use stage-based defaults, then adjust per deal.

Step 3: Build the forecast

1. List every deal with close date in the forecast period
2. Assign probability using stage default + deal-specific adjustment
3. Calculate weighted value per deal
4. Sum into forecast categories (commit, best case, upside, pipeline)
5. Compare total to target

Forecast output

Forecast Accuracy

Measuring accuracy

Forecast Accuracy = 1 - |Actual - Forecast| / Target

Example:
  Forecast: $600K
  Actual: $550K
  Target: $600K
  Accuracy: 1 - |$550K - $600K| / $600K = 91.7%

Accuracy benchmarks

Accuracy rules

• Track accuracy over time. A VP who consistently under-forecasts by 15% is more useful than one who's randomly off by 25%. Consistent bias can be corrected. Random error can't
• Separate commit accuracy from total accuracy. If commit always hits but best case always misses, you have a categorization problem, not a forecasting problem
• Measure per-rep accuracy. Rep A forecasts within 10%. Rep B is off by 30%. The team forecast looks OK because they cancel out. Individual accuracy reveals who needs coaching
• Never inflate to hit target. If the honest forecast is $450K against a $600K target, forecast $450K and communicate the gap. Inflating the forecast delays the reaction time leadership needs

Common Forecast Errors

Systematic errors

Deal-level red flags

Forecast Cadence

Weekly forecast rhythm

Cadence rules

• Forecasts update weekly. Not daily (too much churn, too much time). Not monthly (too late to react). Weekly balances accuracy with effort
• Every deal above $20K gets individual review. The manager should know every deal above threshold by name. Below threshold, manage by category
• Commit is a commitment. If a rep puts a deal in commit and it doesn't close, that's a miss. Commit should only contain deals the rep would bet their job on. This discipline makes the commit number reliable

Measurement

Pre-Forecast Checklist

• [ ] Every deal has an updated close date (not a default or stale date)
• [ ] Every deal has a defined next step with a specific date
• [ ] Probabilities reflect deal-specific signals, not just stage defaults
• [ ] Deals with no activity for 30+ days are cleaned or marked at-risk
• [ ] Close dates have been challenged ("what specifically drives this date?")
• [ ] Single-threaded deals are flagged and probability reduced
• [ ] Commit only contains deals with confirmed budget, timeline, and decision-maker
• [ ] Upside deals have a realistic path to close within the period
• [ ] Total weighted forecast compared to target (coverage ratio)
• [ ] Historical accuracy reviewed to calibrate current forecast

Anti-Pattern Check

• Forecasting from the top down and calling it bottoms-up. "We did $500K last quarter, we'll do $550K this quarter." That's a guess, not a forecast. A bottoms-up forecast starts with individual deals and sums them. If you can't name the deals, it's not bottoms-up
• Every deal at 50% probability. This usually means reps don't know the real probability or are defaulting to "maybe." Require evidence for each probability level. What has the buyer said or done that supports this number?
• Commit that misses by 30%. If commit deals regularly don't close, the commit category has lost meaning. Tighten the criteria. Only deals with verbal yes, pricing agreed, and contract expected this period go in commit
• No deal-level review by management. The VP reads the roll-up number without reviewing individual deals. The number is only as good as the underlying deal assessments. Review the top 20 deals individually
• Forecast never changes week to week. Either nothing is happening (unlikely) or the team isn't updating their deals. If the forecast number doesn't change for 3 weeks, the data is stale
• Mixing CRM stage with forecast category. Stage 4 does not automatically mean "best case." Stage is where the deal is in the process. Forecast category is how likely it is to close this period. A Stage 4 deal with a Q4 close date is not a Q2 forecast item
• Pipeline cleanup only happens at quarter end. 40% of pipeline is zombie deals. The forecast is inflated by $800K in deals that will never close. Clean pipeline continuously, not quarterly. Weekly pipeline hygiene prevents forecast distortion