Feed the Constraint or Feed the Top: Forecasting a Funnel Fix
Objective: Given a 4-stage funnel from a mattress DTC brand, forecast the paid-customer outcome of two competing budget proposals: more top-of-funnel spend vs. fixing the identified constraint stage.
You're presenting to the CMO. Sales wants to double paid traffic. The ops lead wants budget to fix checkout financing options instead, believing that's the real constraint.
Model both proposals against the actual funnel numbers and forecast which produces more paid customers.
Before you start
What you'll need
Free path (everything below is enough to finish)
Free, sufficient for a 4-stage funnel model with two scenario columns
Free, the source of the actual conversion data the forecast is built on
The process
2 steps
Step 01 of 02
The lesson's TOC pass computes conversion rate at every stage; the stage with the biggest gap versus benchmark is the constraint, and adding more top-of-funnel spend just makes that bottleneck worse.
Given the funnel below and DTC mattress benchmarks (visit-to-cart 4%, cart-to-checkout-start 55%, checkout-start-to-purchase 68%), which stage has the largest benchmark gap?
Procedure
- Import funnel-data.csv with visits, cart adds, checkout starts, purchases
- Compute this brand's conversion rate at each stage
- Subtract the benchmark rate from the actual rate at each stage to find the largest negative gap
Stage Actual Benchmark Gap Visit to Cart 4.2% 4.0% +0.2pt Cart to Checkout start 54% 55% -1pt Checkout start to Purchase 41% 68% -27pt
Healthy
A funnel where every stage sits within a few points of benchmark, no single stage is a standout constraint.
Unhealthy
A funnel with one stage sitting 27 points below benchmark while every other stage is roughly in line, that stage is the constraint governing total output.
What this means
Checkout-start-to-purchase, 27 points under benchmark, is the constraint. Visit-to-cart is actually fine, doubling traffic would only pile more people into the same broken checkout step.
So what do I do about it?
| Symptom | Action | Effort |
|---|---|---|
| Checkout-start-to-purchase is 27 points below benchmark while top-of-funnel is on target | Redirect the proposed traffic budget into a checkout audit (financing options, shipping cost visibility, guest checkout) instead | half day |
Step 02 of 02
The lesson's worked example shows doubling traffic doubles signups but the paid-customer gain is smaller than fixing the actual conversion constraint at current traffic.
At 100,000 monthly visits, doubling traffic to 200,000 at current stage rates versus lifting checkout-start-to-purchase from 41% to 60% (still below the 68% benchmark) at current traffic, which produces more purchases, and by how much?
Procedure
- Build scenario A: 200,000 visits at current stage rates through to purchase
- Build scenario B: 100,000 visits with checkout-start-to-purchase raised to 60%
- Compare total purchases for each scenario
Scenario A (double traffic): 200,000 visits, 4,536 purchases Scenario B (fix checkout, same traffic): 100,000 visits, 4,987 purchases Scenario B produces about 451 more purchases with zero added media spend
Healthy
A forecast where the constraint-fix scenario matches or beats the traffic-doubling scenario, confirming the fix is the higher-leverage spend.
Unhealthy
Presenting only the traffic-doubling scenario's raw signup increase without modeling it through to purchases at the still-broken checkout stage, which overstates the actual gain.
What this means
Fixing the constraint beats doubling the media budget here, and it costs nothing in incremental ad spend, only the checkout-audit hours. The CMO's decision has a number attached now, not two competing opinions.
So what do I do about it?
| Symptom | Action | Effort |
|---|---|---|
| Two teams are pitching opposite budget asks with no shared forecast | Require both proposals to run through the same stage-by-stage model before the budget meeting | half day |
Final deliverable
A two-scenario forecast model comparing traffic-doubling vs. constraint-fix, with a one-paragraph recommendation for the CMO.
See a reference example
Allbirds checkout forecast (excerpt) Scenario A, double paid spend: +3,100 forecasted purchases Scenario B, fix cart abandonment email flow: +3,800 forecasted purchases at current spend Recommendation: fund Scenario B first; re-evaluate traffic spend next quarter once the constraint moves.
Success criteria
You're done when you can:
- Correctly identifies the constraint stage using the benchmark gap, not just the lowest raw rate
- Forecast models both proposals through to purchases, not just to the proposal's own stage
- Recommendation is quantified, not just directional