In this webinar replay from last Thursday, you'll learn:

Everyone who watches get a complementary $497.00 AEO audit of their website too. 

Watch the replay on Youtube here

AI video for paid ads is the biggest make-money lever in AI right now, and the one with the loudest hype and the most expensive mistakes.

The numbers you can trust:

→ Meta's AI-powered ad tools just hit a $10 billion annualized run-rate, growing roughly 3x faster than its ads business overall, with new ranking models driving measurable lifts in clicks and conversions (per Meta's own newsroom).

→ RevenueCat ran the head-to-head everyone should study. An AI digital-twin avatar ad against a human creator: the avatar hit 87% of the human's conversion rate at $20 per avatar versus $500 per creator, with a 31% lower cost-per-acquisition.

A Brazilian-Portuguese dubbed version came in at $8 CPA, 4.5% CTR, and 2.1x ROAS. Read that carefully: AI got ~90% of the result at ~4% of the cost. It didn't beat the human on quality. It won on economics.

→ The only independent academic study (Taboola with Columbia, Harvard, TUM, and CMU) found AI ads averaged a 0.76% click-through rate versus 0.65% for human ones. Real, but modest.

And the numbers you should ignore: those "AI ad got 45% higher conversion, 3.4 ROAS" stats floating around are almost all anonymous agency blog claims with no named brand. Modest and real beats spectacular and fabricated. Plan around the modest number.

The rule that makes this all work: AI wins the test, humans win the close. AI UGC dominates on volume, speed, and cost-per-variation. You can generate dozens of ad angles for the price of one human shoot.

But in trust-dependent categories (finance, health, beauty), real humans still convert better at the bottom of the funnel. One first-party $100K test: human UGC won click-through 2.4% to 1.9%, AI won on cost-driven ROAS, and the hybrid beat both.

Steal this prompt for step 1:

And track the number that matters. Cheap AI video lies to you if you count generations instead of finished, usable, converting videos:

Cost per finished video = (base generations + expected regenerations) × per-generation cost + editing/stitching time × your hourly rate ÷ videos that ship

Budget 15–25% for regenerations. That "$1.50 clip" is usually a $5–8 finished video once you count the misses and the assembly. Still far cheaper than a human shoot, but now you're not lying to yourself.

Bottom line: use AI as the wide top of the funnel and the cheap testing engine. Don't ask it to be the trusted face that closes a big sale. It's not there yet, and the data says your customers can tell.

The first piece just landed in the ad console: a new "Targeting strategy" beta inside manual Sponsored Products campaigns. One checkbox, easy to scroll right past.

Don't.

Flip it on and Amazon's AI adds up to 4 new keywords and product targets every 12–24 hours, all at or below your current bids, based on your existing targets and what's worked for similar products.

It also pauses targets. And that's where Amazon shows its hand, because these are the exact rules its algorithm uses to decide a target isn't working:

→ No impressions in 3 days
→ No clicks in 7 days
→ Over 15% of daily budget spent without converting
→ 20+ clicks with no conversion in 14 days

Screenshot those. When everything runs on AI, this is the logic deciding your spend.

Why Amazon is doing this — and where it's heading:

→ Google already proved it. Performance Max showed that automation grows ad revenue faster than any bidding UI. Amazon is behind on that curve and knows it.

→ The incentive is simple. Every keyword the AI adds is another auction Amazon gets to monetize. "At or below your current bids" caps your risk. It does nothing to cap their coverage.

→ You're the training data. "Performed well for similar products" means every campaign that opts in teaches the algorithm — Amazon learns from all of us at once.

The playbook from here writes itself: opt-in beta first, visible guardrails to build trust, then it becomes the default. Then targeting merges with the ROAS-based bidding already sitting in the console. Then you hand Amazon a budget and a goal, and it does the rest.

Manual PPC isn't dying today. But Amazon just told us how it ends.

Credit: BDSS Dream 100 member Mansour Norouzi, partner & director of advertising at Incrementum Digital, who posted this inside the Billion Dollar Sellers Club (BDSC).

If you were hoping the Supreme Court striking down Trump's "liberation day" tariffs meant your landed costs were coming down, that window just closed.

Right as the clock ran out Friday on the temporary 10% global tariff (imposed under Section 122, which legally maxes out at 150 days), the US rolled out its replacement: fresh tariffs on 60 trade partners, this time citing forced labor violations under Section 301 of the Trade Act of 1974.

The group covered, including China, Canada, and the EU, accounts for 99.4% of US imports. If you import, you're in it.

The new math:

→ Duties range from 10% to 12.5% so depending on your sourcing country, this is either a wash versus the expiring 10% or a bump on top of it.

→ Exemptions exist: oil and gas, fertilizer, many goods not produced in the US, and products that qualify for duty-free status under USMCA. Steel and aluminum keep their existing separate tariffs but don't get this stacked on top.

→ Good behavior gets rewarded. Countries that tightened forced labor enforcement after the tariffs were proposed last month got lower rates. India was penciled in at 12.5% and landed at 10%. Expect more countries to negotiate down.

Why "forced labor"? Because Section 301 is the legal tool that survived. The Supreme Court ruled IEEPA didn't authorize tariffs (and forced the administration to refund importers who paid them). Section 122 expired Friday. Section 301 targets "unfair trade practices" and forced labor qualifies. The legal justification changed. The bill you pay at the port did not.

The reaction was immediate: Australia called the tariffs "completely unjustified," Brazil is heading to the WTO and threatening retaliatory tariffs, and a group of US small businesses filed suit the same day to block them.

Meanwhile Bloomberg describes a "triple whammy" of tariffs, energy costs, and AI spending aggravating inflation with midterms on November 3.

What this means for you:

→ Stop waiting for relief. This administration has now rebuilt its tariff wall three times under three different legal authorities. Whatever the courts do to this version, assume a fourth exists. Price your 2027 POs with tariffs baked in.

→ Two more waves are already queued. The USTR has an open probe into whether 16 countries — 70% of US imports — have overproduced goods and pushed down global prices. That's aimed squarely at the low-cost manufacturing economics most sellers source from. A separate wave targeting the EU over its fines on US tech giants is also planned.

→ Watch your sourcing country's status. Rates moved between the proposal and the final list based on enforcement steps. If your factory's country drops from 12.5% to 10%, or gets hit in the overproduction wave, that's margin appearing and disappearing without you touching anything.

The tariff era isn't a storm to wait out anymore. It's the climate.

While everyone obsesses over AI shopping assistants, Marketplace Pulse says Amazon Business just hit $60 billion in annualized gross sales, up from $35B in 2023 and $25B in 2021. As a standalone marketplace, it would rank among the largest e-commerce platforms anywhere, and it's closing in on Amazon's $68.6B ad business.

The compounding has been machine-like: roughly 18% a year since 2021, doubling the 9% Amazon's overall GMV grew last year and lapping the 5-8% pace mature U.S. e-commerce has settled into.

Here's the part that matters for you: Amazon says more than half of Amazon Business sales come from third-party sellers. That's something in the region of $30 billion flowing to sellers, and most sellers aren't even playing.

Why the B2B side is wide open:

→ All the recent investment went to buyers, not sellers. Amazon just rolled out an AI account assistant, Savings Insights for bulk-discount analysis, spend anomaly monitoring, dedicated delivery trucks for loading docks in 13 states, and a Prime Business bundle with CrowdStrike, Gusto, and QuickBooks.

The seller-side tools like business-only pricing, tiered quantity discounts and custom quote requests have barely changed in years. The demand side got rebuilt; the supply side is still waiting for competition.

→ The demand shows up through listings you already maintain. Higher average order values, from buyers that include hospitals, universities, governments, and 97 of the Fortune 100. No new catalog required.

→ It's growth you don't have to buy at auction. With 49% of sellers naming marketplace fees as their top margin concern and 46% citing advertising, incremental sales captured through quantity discounts on existing inventory beat incremental sales purchased through ever-pricier ad clicks.

The structural reason growth hasn't exploded faster: B2B procurement runs on approval chains, POs, payment terms, and negotiated contract pricing, none of which the consumer marketplace ever had to solve. Amazon has spent eleven years digitizing the easy end of a market that's multiples bigger than consumer spending.

At the current trajectory, Amazon Business crosses $100 billion before the end of the decade. The sellers who set up their B2B pricing now get to ride it there.

✌🏼 See you again Thursday …

The answer to today’s STUMP BEZOS is
$49.2B in tariffs was refunded in June.