Where you are
This is the course's centerpiece session. The metrics course taught you the four gates as numbers; b1-b3 taught you the screens and paths behind them. The simulator below welds it all together: one flow, four gates, four fixes. In b5-b6 you go deep on the two highest-leverage screens (the PDP and checkout), and in b7 you run this machine in reverse - from a broken number back to a broken screen.
The journey is one machine with four dials 10 min live
Mango Lane's month, as a flow: 360,000 sessions, 40% reach a PDP, 25% of those add to cart, 45% of those start checkout, 60% of those pay. Multiply through: 9,720 orders, 2.7% session CR, $660,960 GMV. Each rate is a dial - and because the flow is multiplicative, turning any single dial moves the end of the line by the same proportion.
LiveWhy fixes compound4 min▶
Gate improvements multiply, they do not add. Improve PDP penetration by 7.5% (40 to 43) and add-to-cart by 12% (25 to 28) and orders rise by roughly 1.075 x 1.12 = 20.4% - more than 7.5 + 12 would suggest for big moves. Four modest fixes below compound to +47% GMV. This is why journey work beats one-big-redesign work: several small, owned, testable gate deltas stack.
Self-studyWhere the four fixes come from3 min read▶
The lever sizes in the simulator are illustrative teaching values, but each fix is a real, well-documented pattern:
- Smarter search & nav - killing zero-result searches and improving autocomplete raises the share of sessions that ever reach a product page (b2-b3 showed search-path sessions convert about 2.7x browse-path).
- Richer PDP - better galleries, visible reviews, and size guidance raise add-to-cart (b5 measures engaged-PDP sessions converting 2x plain ones).
- Persistent cart + nudges - carts that survive across sessions and gentle reminders raise checkout starts.
- One-page checkout + wallets - fewer form fields and wallet payments raise completion (b6 walks the checkout micro-funnel; Baymard's research catalogs exactly which frictions kill it).
Read the machine at rest ★ 8 min · everyone reads
Here is the live flow. Before touching anything, read it end to end and match every number to what you already know from b1-b3 and the metrics course - same store, same month, same 9,720 orders.
{
"sessions": 360000, "aov": 68,
"stages": [
{ "label": "Sessions" },
{ "label": "PDP sessions", "rate": 0.40, "rateLabel": "PDP penetration" },
{ "label": "Cart sessions", "rate": 0.25, "rateLabel": "add-to-cart" },
{ "label": "Checkout sessions", "rate": 0.45, "rateLabel": "checkout start" },
{ "label": "Orders", "rate": 0.60, "rateLabel": "completion" }
],
"levers": [
{ "stage": 1, "to": 0.43, "label": "Smarter search & nav", "desc": "PDP penetration 40% to 43% - fix zero-result searches, better autocomplete" },
{ "stage": 2, "to": 0.28, "label": "Richer PDP", "desc": "add-to-cart 25% to 28% - gallery, reviews, size guidance up front" },
{ "stage": 3, "to": 0.50, "label": "Persistent cart + nudges", "desc": "checkout start 45% to 50% - cart survives sessions, gentle reminders" },
{ "stage": 4, "to": 0.66, "label": "One-page checkout + wallets", "desc": "completion 60% to 66% - fewer fields, wallet payments" }
]
}
Read the lane left to right: 360,000 → 144,000 → 36,000 → 16,200 → 9,720. Each arrow carries the gate rate that produced the next count.
Check the scorecard line: 9,720 orders, 2.70% session CR, $660,960 GMV. The machine at rest is exactly the month you have been decomposing since b1.
Say out loud which team owns each arrow. If you cannot name an owner, that gate is nobody's job - and it will never move.
Run the four fixes - the ladder ★ 15 min · everyone builds
Now turn the dials one at a time, in order, and check the machine against the expected ladder below. Every rung is a real recomputation - the simulator multiplies the counts through live.
| Rung | Fixes on | Orders | Session CR | GMV | vs baseline |
|---|---|---|---|---|---|
| 0 | none (baseline) | 9,720 | 2.70% | $660,960 | - |
| 1 | Smarter search & nav | 10,449 | 2.90% | $710,532 | +7.5% |
| 2 | + Richer PDP | 11,703 | 3.25% | $795,796 | +20.4% |
| 3 | + Persistent cart | 13,003 | 3.61% | $884,218 | +33.8% |
| 4 | + One-page checkout | 14,304 | 3.97% | $972,639 | +47.2% |
Toggle Smarter search & nav. PDP sessions jump to 154,800 and every later stage inherits the gain - orders 10,449, CR 2.90%. One dial, whole machine.
Add Richer PDP. Now two gates are coral: orders 11,703, GMV +20.4%. Notice 7.5% and 12% compounding past 19.5%.
Add the remaining two fixes and land on the full ladder: 14,304 orders, 3.97% CR, $972,639 GMV - +47.2% from four modest, testable deltas.
Now press Reset and try the fixes in a different order. The end state is identical - multiplication commutes - but the per-fix credited lift you would report differs. Remember that when teams fight over credit.
Last drill: turn on only ONE fix of your choice and write its one-line business case: "moving [gate] from X% to Y% is worth [orders] orders / [$] GMV per month."
This table is how journey roadmaps get funded. A UX team asking for "checkout redesign budget" gets questions; a team showing "completion 60% to 66% = +$88K GMV per month on current traffic" gets a decision. The simulator's honesty rail still applies - the effect sizes are estimates until an A/B test confirms them (that craft lives in learn-experimentation-with-phoebe) - but the arithmetic of the claim is exactly this.
Self-studyWhich fix first? The preview of a64 min read▶
All four rungs assume you can do everything at once. Real teams pick. Two useful tie-breakers, both computable from this machine:
- Reach - how many sessions touch the gate. Gate 1 touches all 360,000 sessions; gate 4 touches only 16,200. A small % lift on a huge base can beat a big % lift on a tiny one.
- Distance to benchmark - Mango Lane's completion (60%) is already strong versus the ~45% Shopify average (Littledata, 2025), while its PDP penetration has no such cushion. Fix the gate that lags its benchmark, not the one that is easiest to imagine.
a6 turns this into a full prioritization ritual (reach x severity x confidence); b7 uses the same machine backwards to diagnose a leak.
Try it yourself - this week ◐ 20-30 min total
- Rebuild the baseline flow for your own store: sessions, your four gate rates, orders. If a gate rate is unknown, that gap is finding number one.
- Write your top three journey fixes as gate deltas (gate, from%, to%, mechanism). Toss any "fix" that cannot be written that way.
- Compute each fix's monthly GMV value with the simulator (paste your numbers into the flow JSON mentally or on paper) and rank them.
- Check one gate against a real benchmark before proposing to fix it - a5 of the metrics course has the honest tables.
Sources this session draws on
The simulator is a teaching machine; the gates and fixes it models are standard, documented patterns. This page draws on:
Three questions before you go 🎯 ◐ 90 seconds
1 · PDP penetration improves 7.5% and add-to-cart improves 12%. Orders rise by...
Gate improvements compound multiplicatively: 1.075 x 1.12 = 1.204. For small moves the sum is a decent approximation; for stacked fixes the compounding is real money.
2 · What makes "improve search" unacceptable on a journey roadmap?
A fix must be a gate delta: "PDP penetration 40% to 43% by fixing zero-result searches." That form makes it valuable (the simulator prices it) and testable (an experiment can confirm it).
3 · Completion (60%) already beats the ~45% benchmark; PDP penetration has no such cushion. Which fix deserves scrutiny first?
Distance-to-benchmark plus reach are the two computable tie-breakers. A gate already beating its benchmark has less headroom; a gate touching all 360K sessions has more leverage per point.