Stop Trusting Blended Account Margin

Hey Readers 🥰

Welcome to today's edition, bringing the latest growth stories fresh to your inbox.

If your pal sent this to you, then subscribe to be the savviest marketer in the room😉

📦 The Margin Math Nobody Runs Before Scaling

Looking to scale on Amazon without quietly losing money on every sale? The brands winning this year are running better P&Ls, not bigger budgets. Here is what separates operators who scale on purpose from those who drift.

1️⃣ Know Your Real Fee Stack
Referral fees, fulfillment fees, inbound placement fees, and the low-inventory-level fee now stack up to 40 to 50 percent of revenue. The low-inventory fee applies per variant, and inbound fees drop to zero once you split shipments across five or more fulfillment centers.

2️⃣ Build a Per SKU Margin Sheet
Blended account margin hides SKUs that lose money on every ad sale. Calculate contribution margin per SKU after fees, landed cost, and ad spend, then only scale the SKUs that stay profitable.

3️⃣ Fix Your Pack Architecture First
Multipacks and bundles improve AOV against a mostly fixed fee base. Subscribe and Save builds repeat revenue that costs nothing to reacquire. Design your unit of sale before you design campaigns.

4️⃣ Optimize Listings for Rufus and A10
Listings need to answer who a product is for and when it gets used, not just stuff keywords. External traffic now influences ranking, and the Brand Referral Bonus rebates a portion of sales driven from outside Amazon.

5️⃣ Budget Ads Like a Portfolio
Split spend into launch, harvest, and defense tiers instead of one blended ROAS number. Test branded spend holdouts and use AMC to see what is actually incremental.

The Takeaway
Growth on Amazon in 2026 is decided by systems that tell you the truth, your fee stack, your listings, and your ad account. Check the math before you scale the spend.

📊 Meta Opens Up Two New Models

Meta just made a major move in open source AI. Not only did the company release a new 30 billion parameter open weight model, but plans were also announced to release the weights for a second, more advanced model soon. Here is what happened and why it matters.

1️⃣ Muse Glimmer Goes Live Today
Meta released Muse Glimmer, a 30 billion parameter dense model built to run locally, and made it available immediately on Hugging Face along with an official blog post detailing the release.

2️⃣ Muse Spark 1.2 Weights Are Coming Soon
Meta also confirmed it will open the weights for Muse Spark 1.2, one of its most capable foundation models. It currently ranks among the strongest models on the Artificial Analysis Intelligence Index, sitting close to leading proprietary models from other labs.

3️⃣ Leadership Frames It As a Bigger Mission
Meta’s leadership tied the release to a broader vision, stating that everyone should have access to personal superintelligence and that the goal is a personal AI agent working continuously to improve health, relationships, career, and finances.

4️⃣ Why This Matters for Open Source
Two releases in one day from a major lab signals real momentum for open weight models. It also reinforces that high performing models are no longer limited to closed, proprietary systems.

The Takeaway
This is a big day for open source AI. Meta releasing Muse Glimmer immediately, with Muse Spark 1.2 weights on the way, shows the industry is genuinely moving toward broader access rather than keeping the most capable models locked behind closed doors.

We'd love to hear your feedback on today's issue! Simply reply to this email and share your thoughts on how we can improve our content and format.

Have a great day, and we'll be back again with more such content 😍