Funnel Engineering
Your funnel is not your website. It is the system that converts qualified interest into committed revenue.
A sales funnel is a measurement instrument. Each stage answers a question. If the answer is wrong, the deal doesn't advance. If the answer is never asked, the deal stalls and nobody knows why.
Overview
| Attribute | Value |
|---|---|
| Purpose | Design, measure, and optimize the stages deals move through from qualification to close |
| Trigger | Quarterly funnel review, or when conversion rate drops below target at any stage |
| Frequency | Continuous tracking, quarterly architecture review |
| Duration | Ongoing — 30 min/day pipeline review, 2-4 hours quarterly redesign |
| Owner | Sales Leadership (human strategy) + RevOps (AI analytics) |
| Output | Stage-by-stage conversion data, bottleneck diagnosis, optimization actions |
Human Role: Stage design, conversion strategy, deal progression decisions AI Role: Pipeline analytics, forecasting, velocity tracking, bottleneck detection Spectrum: AI-Assisted
Prerequisites
Tools Required
| Tool | Purpose | Access |
|---|---|---|
| CRM pipeline | Visual deal tracking by stage | Sales CRM |
| Sales forecasting | Predict deal close probability | Sales Forecasting algo |
| Analytics dashboard | Conversion metrics by stage | CRM reporting |
Knowledge Requirements
- Understanding of buyer's decision process (not your selling process)
- Historical deal data (close rates, cycle times, deal sizes)
- Product/market fit assumptions validated through lead qualification
Inputs
| Input | Source | Required? |
|---|---|---|
| Qualified leads | Lead Qualification | Yes |
| Historical deal data | CRM | If available |
| Buyer journey understanding | Customer interviews, ICP research | Yes |
| Revenue targets | Business plan | Yes |
Process
Phase 1: Design the Funnel
Duration: 2-4 hours (annual, refresh quarterly) Responsibility: Human-led
The funnel architecture should mirror the buyer's decision process, not your internal org chart.
Step 1.1: Define Stages
Each stage represents a BUYER decision, not a seller activity.
| Stage | Buyer Question | Evidence Required | Exit Criteria |
|---|---|---|---|
| Lead | "Should I learn more?" | Responded to outreach | Accepted a conversation |
| Discovery | "Do they understand my problem?" | Attended discovery call | Problem confirmed, mutual interest stated |
| Qualified | "Is this worth my team's time?" | Shared internal requirements | Decision maker engaged, budget discussed |
| Proposal | "Does the solution match my needs?" | Reviewed proposal or demo | Specific feedback given |
| Negotiation | "Can we make the terms work?" | Active discussion on price/terms | Terms agreed in principle |
| Won | "Let's do this" | Signed agreement or PO | Payment or commitment received |
| Lost | "Not now / not this" | Explicit decline or 30-day silence | Reason captured, learning documented |
Step 1.2: Set Stage Metrics
| Stage | Target Conversion | Target Duration | Deals Expected |
|---|---|---|---|
| Lead → Discovery | 40-60% | 1-2 weeks | ? |
| Discovery → Qualified | 50-70% | 1-2 weeks | ? |
| Qualified → Proposal | 60-80% | 1-3 weeks | ? |
| Proposal → Negotiation | 40-60% | 1-2 weeks | ? |
| Negotiation → Won | 50-70% | 1-2 weeks | ? |
| Overall: Lead → Won | 5-15% | 4-12 weeks | ? |
Step 1.3: Work Backward from Revenue
Revenue target: $X/quarter
Average deal size: $Y
Close rate: Z%
Deals needed: X / Y = N deals
Pipeline needed: N / Z = P pipeline deals
Leads needed: P / stage_conversion_rates = L leads
This math tells you exactly how many leads to generate. If the numbers don't work, the problem is either deal size, close rate, or volume — and you know which to fix.
Phase 1 Output: Funnel architecture with stages, metrics, targets
Phase 2: Track Pipeline Health
Duration: 30 minutes daily, 1 hour weekly Responsibility: AI-led tracking, human interpretation
Step 2.1: Daily Pipeline Review
| Check | What to Look For | Action |
|---|---|---|
| Stalled deals | Any deal in same stage >2x target duration | Ask: what's blocking? Who needs to act? |
| Overdue follow-ups | Tasks past due date | Complete today or reschedule with reason |
| New entries | Leads that entered pipeline today | Verify qualification, assign next action |
| Stage changes | Deals that advanced or regressed | Celebrate advances, diagnose regressions |
Step 2.2: Weekly Pipeline Metrics
| Metric | Formula | What It Tells You |
|---|---|---|
| Pipeline value | Sum of all deal values × probability | Expected revenue |
| Pipeline coverage | Pipeline value / quota | Need 3-4x coverage minimum |
| Velocity | (# deals × avg value × win rate) / avg cycle | Revenue per time period |
| Stage distribution | Count of deals per stage | Healthy = pyramid shape |
| Average deal age | Days since deal created | Aging = stalling |
Step 2.3: Funnel Shape Diagnosis
| Shape | What It Looks Like | Diagnosis |
|---|---|---|
| Pyramid (healthy) | Many at top, few at bottom | Normal — leads filter through stages |
| Cylinder | Equal at every stage | Not qualifying enough — too many deals advancing |
| Inverted pyramid | Few at top, many at bottom | Pipeline will dry up — lead gen needed |
| Hourglass | Many top, few middle, many bottom | Bottleneck in middle stages — qualification or proposal problem |
| Flat | Very few at all stages | Pipeline emergency — stop everything else, generate leads |
Phase 2 Output: Pipeline health report with diagnosis
Phase 3: Optimize Conversions
Duration: 2-4 hours quarterly Responsibility: Human analysis, AI data
Every stage has a conversion rate. Improving any one rate compounds across the whole funnel.
Step 3.1: Identify the Bottleneck
The bottleneck is the stage with the lowest conversion rate relative to target.
| Stage Transition | Target | Actual | Gap | Priority |
|---|---|---|---|---|
| Lead → Discovery | 50% | ? | ? | ? |
| Discovery → Qualified | 60% | ? | ? | ? |
| Qualified → Proposal | 70% | ? | ? | ? |
| Proposal → Negotiation | 50% | ? | ? | ? |
| Negotiation → Won | 60% | ? | ? | ? |
Fix the bottleneck first. A 10% improvement at the bottleneck has more impact than 10% everywhere.
Step 3.2: Diagnose Root Cause
| Bottleneck Stage | Common Causes | Diagnostic Questions |
|---|---|---|
| Lead → Discovery | Poor outreach, wrong ICP | Are replies relevant? Do they ask questions? |
| Discovery → Qualified | Weak discovery, no pain confirmed | Did they name the pain unprompted? |
| Qualified → Proposal | Value not clear, stakeholders not aligned | Did the decision maker attend? |
| Proposal → Negotiation | Price shock, feature gap, wrong frame | What specific objection killed it? |
| Negotiation → Won | Competing priorities, internal politics | What changed between "interested" and "silent"? |
Step 3.3: Run Improvement Experiment
Pick ONE bottleneck. Design ONE change. Measure for ONE cycle.
HYPOTHESIS: If we [change], then [stage] conversion will improve by [%]
EXPERIMENT: [Specific change to process, messaging, or qualification]
MEASURE: [Stage conversion rate, n=20+ deals]
DURATION: [One sales cycle, typically 4-8 weeks]
DECISION: If improved → standardize. If not → diagnose further.
Phase 3 Output: Identified bottleneck, root cause, improvement experiment
Phase 4: Forecasting
Duration: 30 minutes weekly Responsibility: AI-led prediction, human adjustment
Step 4.1: Stage-Weighted Forecast
| Stage | Deal Count | Avg Value | Stage Probability | Weighted Value |
|---|---|---|---|---|
| Discovery | ? | ? | 10% | ? |
| Qualified | ? | ? | 25% | ? |
| Proposal | ? | ? | 50% | ? |
| Negotiation | ? | ? | 75% | ? |
| Total pipeline | $? |
The Sales Forecasting algorithm provides more sophisticated predictions using deal velocity, engagement signals, and historical patterns.
Step 4.2: Commit vs Upside
| Category | Criteria | Expected Close |
|---|---|---|
| Commit | Verbal agreement, terms discussed, timeline confirmed | This month |
| Best case | Proposal reviewed, positive feedback, decision pending | This quarter |
| Pipeline | Qualified but not yet in active evaluation | Next quarter |
| Upside | Early stage, could accelerate | Future |
Phase 4 Output: Revenue forecast with confidence levels
Outputs
| Output | Format | Destination |
|---|---|---|
| Pipeline health report | Metrics dashboard | Weekly sales review |
| Stage conversion data | CRM analytics | Quarterly optimization |
| Revenue forecast | Weighted pipeline | Leadership reporting |
| Bottleneck diagnosis | Analysis + experiment plan | Process improvement |
Downstream Consumers
| Downstream Workflow | What It Needs | Link |
|---|---|---|
| Customer Lifecycle | Won deals entering onboarding | Lifecycle |
| Revenue forecasting | Weighted pipeline data | Business planning |
| Lead Generation | Pipeline gap → volume needed | Lead Gen |
Success Criteria
Quality Metrics
| Metric | Target | Measurement |
|---|---|---|
| Forecast accuracy | Within 15% of actual revenue | Monthly comparison |
| Stage probability calibration | Stage % matches actual conversion | Quarterly calibration |
| Deal aging alerts | 100% of stalled deals flagged | CRM automation |
| Win/loss capture rate | 100% of closed deals have reason | CRM data quality |
Performance Metrics
| Metric | Target | Timeframe |
|---|---|---|
| Pipeline coverage | 3-4x quota | Monthly |
| Sales cycle length | <12 weeks (avg) | Monthly |
| Win rate | >30% overall | Quarterly |
| Deal velocity | >$X revenue per week | Monthly |
Failure Modes
| Failure | Symptom | Diagnosis | Solution |
|---|---|---|---|
| Happy ears | Every deal is "looking good" | No objective stage criteria | Define evidence required per stage |
| Sandbagging | Pipeline always low, deals appear at close | Reps hiding deals until certain | Create safe-to-forecast culture |
| Pipeline rot | Large pipeline, low close rate | Stale deals inflating numbers | Monthly pipeline hygiene — kill or advance |
| Stage skipping | Deals jump from Lead to Proposal | Skipping qualification | Enforce gate criteria before advancement |
| Single-thread | All deals through one contact | One person leaves, deal dies | Map buying committee, multi-thread |
Context
- Lead Qualification — What feeds the funnel
- Customer Lifecycle — What happens after the close
- Sales CRM & RFP — The pipeline tool
- Sales Work Chart — Human/AI split for pipeline management
- Process Optimisation — The improvement methodology
- Landing Pages — Where inbound leads enter the funnel