How do I build a real bottom-up forecast in a 50-rep org?

Don't ask AEs to estimate pipeline. Ask them to list every deal (deal name, company, amount, close date, next step, owner, confidence). Then YOU bucket the deals by stage and probability yourself.

Reps are terrible forecasters: Gartner's 2024 sales forecasting study found that AE-submitted forecasts miss actuals by 18% on average, and only 45% of forecasted deals close in the forecasted quarter (https://www.gartner.com/en/sales/research). Your bottom-up forecast lives in your CRM, not in a manager's gut.

How to build it (weekly cadence, not monthly—see /knowledge/q12 on cadence design):

Step 1: Data pull. Export your CRM: deal name, contact, amount, stage, last activity date, next step, AE-submitted confidence. One row per deal. If your CRM can't export this in 5 minutes, fix that first (see /knowledge/q33 on CRM hygiene minimums).

On Salesforce, this is a single Opportunities report filtered to IsClosed=FALSE with the columns above; on HubSpot, a Deals export with the same fields. If you can't get to a one-click view, the rest of this process is theater.

Step 2: Clean & validate. Check 4 things:

• Amount is set. No $0 deals. Clari's 2024 State of Revenue benchmark shows 23% of pipeline in median orgs has missing or stale amounts (https://www.clari.com/resources/state-of-revenue/).
• Stage is filled. No "other" or blank stages.
• Next step exists. If a deal has no next step, it's dead—move to closed-lost.
• Last activity is <21 days ago. If older and you're forecasting on it, that's lying. Bridge Group's 2024 SaaS Sales Report shows deals with >21 days of inactivity close at 11% the rate of active deals (https://www.bridgegroupinc.com/saas-report). See /knowledge/q18 on activity-based pipeline scoring.

Step 3: Assign confidence by stage, not by rep opinion. Don't trust AE confidence scores—pull historical close rates instead.

• Stage 1 (Discovery): 5–10% confidence, average cycle 30 days
• Stage 2 (Proposal): 20–25% confidence, average cycle 21 days
• Stage 3 (Negotiation): 45–50% confidence, average cycle 14 days
• Stage 4 (Commit/Legal): 70–75% confidence, average cycle 7 days
• Stage 5 (MSA Signed): 95% confidence, average cycle 3 days (Clari benchmark for commit-stage close: 95% within 30 days, https://www.clari.com/resources/state-of-revenue/)

These percentages must come from YOUR historical data. Pull 12 months of deals and calculate: "Of deals that entered Stage 2 in the past year, what % actually closed?" Use that real number, not industry benchmarks (see /knowledge/q41 on stage-conversion math). Important: calculate on a *dollar-weighted* basis, not a deal-count basis—a 30% deal-count win rate in stage 3 can hide a 15% dollar-weighted rate if your big deals slip more than your small ones.

Step 4: Segment by rep, deal age, and size. Create a pivot showing:

• How many deals each AE has at each stage
• How many of those are >21 days old (at risk)
• Total revenue by stage
• Deals >$50K segmented separately (those need 1:1 review, not aggregate math)

For a 50-rep org, expect ~600–900 open opportunities (Bridge Group 2024 reports the median enterprise AE carries 12–18 active opps; https://www.bridgegroupinc.com/saas-report). Anything materially outside that range means hygiene is broken.

Step 5: Calculate three forecasts.

Present all three to leadership. Never claim "we're making $X" when you mean "if everything closes." Say "conservative is $2.4M, realistic is $3.1M, upside is $4.2M." Board framing: "Commit is conservative; my realistic is the number we hit if Q-2 cohort behaves like Q-1; upside is if our two stage-4 outliers both land." That sentence buys you credibility no spreadsheet can.

Step 6: Reconcile bottom-up vs top-down. Your CFO has a top-down number from the board (ARR target / 4). If your realistic bottom-up is >15% below top-down, that's the gap you have to close with new pipeline this quarter. If it's >15% above, your stage probabilities are too generous—recalibrate.

Pavilion's 2024 Compensation Report shows the median CRO misses quota by 9% precisely because bottom-up and top-down are never reconciled (https://www.joinpavilion.com/compensation-report). See /knowledge/q26 on quota-coverage math.

Step 7: Flag red deals. Any deal >$50K that's over age-target for its stage. Pull the AE in and ask: "This deal is 40 days old in stage 2. What's changed since day 20? Is it still real?" Update it or close it (see /knowledge/q07 on deal-review questioning).

Do this every Friday. Takes 2 hours for a 50-rep org (or 30 minutes if you automate it with Salesforce Reports/Power BI). Once AEs realize you're checking the data independently every week, they stop sandbagging on Fridays and actually work harder.

Bear Case (why this can still fail): This whole approach assumes your stage definitions are stable and your historical close rates are still predictive. If you (a) just changed your sales process in the last 6 months, (b) entered a new segment, (c) changed pricing >10%, or (d) had >20% AE turnover, your historical stage probabilities are garbage and a bottom-up forecast built on them will be confidently wrong.

In those cases, run a top-down forecast from booked ARR + new logo run-rate for one quarter while you rebuild the historical baseline. Also: if your sales cycles are <14 days (PLG/SMB), stage-weighted forecasting doesn't work—use a cohort-based run-rate model instead. If leadership uses your conservative number as the new target, reps will learn to hide deals from you—commit number must stay separate from stretch.

And there is a deeper Goodhart's-Law failure: once reps know which fields you grade (next step filled, activity <21 days), they will mechanically tick those boxes without doing the underlying work. Counter this by spot-checking 5 deals/week against actual call recordings or email threads—if the 'next step' on file doesn't match the last real customer interaction, that's data fraud, not a forecasting error.

TAGS: forecasting,sales-ops,bottom-up,50-rep-org,revenue-ops