If it's a problem you can solve without me, I'll tell you that too.
Four cases
Proof, in writing.
Same three beats every time. Hypothesis. Test. Verdict, including when it went against me.
Field note 01Supplements & wellness
Four years of losses. Growth was making it worse.
By the time I got the call the brand had lived a whole arc. Revenue ran up to $9.9M, slipped to $9.1M, and the losses grew with the scale: $4.2M gone in the biggest year. By late 2022 revenue had nearly halved to $5.1M, still losing money. Every fix before then had been a growth fix. More spend, more launches, more channels. The topline moved, and the P&L quietly recorded what all that motion produced: a bigger loss. Marketing owned ROAS. Ops owned shipping. Finance owned the autopsy. Contribution margin per order belonged to no one.
Hypothesis
The economics are broken at any size. Growth multiplies them, whichever direction they point.
Test
Freeze growth. Rebuild contribution margin per order at $5.5M, on purpose. Cut the revenue we were paying to lose, knowing the topline would feel it.
Verdict
$5.5M → $30M
2023 closed at $5.5M, barely bigger than the year before, and swung from a $1.6M loss to a $300K profit, a small number that settled the argument. Then $18.8M, then $30M at roughly 10% net. Same playbook, with volume poured into economics that deserved it.
Net salesLossProfit
Take it home. Get contribution margin per order, fully loaded, before any scale plans. Scale only what is already profitable at small volume.
Field note 02Sustainable apparel
The bundle builder felt obvious. The data said skip it.
Average order value is the lever everyone agrees on, and the bundle builder is the prom king of AOV tactics. Let shoppers assemble a set, stack a small discount, watch order size climb. We believed it too. The brand sold basics people genuinely buy three at a time, so a builder felt like a layup. Conversion dropped 23% with it live, and we watched it hold. The builder showed a bigger number, asked for more decisions, and added steps before anyone had committed. We were taxing the first purchase to chase a bigger one.
Hypothesis
Where you ask matters more than what you ask. Protect the first yes, then sell the second yes after the money is in.
Test
Same idea, other side of the payment: one post-purchase add-on, one tap, charged to the card on file, packed in the same box. Same products, same discount.
Verdict
−23% → +30%
The builder cost 23% of conversion. The post-purchase offer lifted AOV 30% at a 20% take rate, with no extra shipping weight and no discount doing the heavy lifting. Revenue that arrives without new costs lands almost entirely as contribution.
Take it home. Judge every AOV tactic on contribution per session, not on AOV alone. The default home for “buy more” is after payment, where it can't cost you the sale.
Field note 03Premium menswear
A return means the customer isn’t done yet.
Returns are a cost center. Shipping both ways, processing, the occasional write-off. So the playbook writes itself: tighten the policy, add friction, drive the rate down, celebrate the savings. That math is correct per event. It answers the wrong question. The brand's in-store reps were good. They loaded first-time buyers up: this fit in two colors, the next size as a backup, a third piece to round it out. First orders regularly ran past three items, and something always came back.
Hypothesis
Returners aren't churning. They're fit-seeking, which is what committing looks like.
Test
Stop counting the cost per return. Cut lifetime value by who returned and who didn't.
Verdict
+20% LTV
Customers who made a return were worth 20% more over twelve months than customers who never sent anything back. A return that ends in an exchange is a second considered interaction: they handled the product, made a judgment, and chose to stay.
Take it home. Make the exchange the default path. Never punish returners in segmentation. Judge policy on cohort LTV, not per-event cost. Optimize returns to zero and you optimize away your best customers.
Field note 04Checkout economics
We turned on Apple Pay. First-order AOV went down.
Every CRO checklist says turn on express wallets, and none of it is wrong. Fewer fields, faster checkout, less typing on mobile. We expected a clean win. What the checklists don't mention is what the faster path skips on the way through. Shopify's post-purchase page, the screen where a one-click add-on shows after payment, does not render for wallet orders. The moment Apple Pay went live, every customer who used it became invisible to the offer doing our AOV work. We didn't remove the upsell. We installed a bypass around it.
Hypothesis
The faster path is skipping something worth more than the friction it removes.
Test
A clean cohort: new one-time customers over seven weeks, subscriptions and BFCM codes excluded. Split AOV, repeat rate and LTV by payment method.
Verdict
−11% AOV
Apple Pay took 56% of new checkouts. Those orders ran 11% lighter, repeated 8% less, and carried 10% lower LTV. For it to pay for itself, its conversion lift had to beat all of that. We could never isolate one. We turned it off until the math changes.
Take it home. Split first-order AOV and post-purchase take rate by payment method. Watch eligibility, not just take rate. Our offer stayed the same; fewer customers ever saw it. Re-run the math quarterly.
Next step
Text me.
Tell me what's not working. Short is fine. Most conversations don't turn into engagements. You'll leave with something useful either way.
I willEvery week, in writing.