E-commerce & Dropshipping Misleading — the headline number is real but unrepresentativ

By the Income Reality Check Editorial Team · Updated 2026-05-23 · 7 min read

Jordan Welch’s 30-day Claude AI dropshipping pitch: the missing math

Verdict: Misleading — the headline number is real but unrepresentative. Medvi’s $1.8B run rate exists, but so does its FDA warning, and the math on ads and fees doesn’t give a beginner anything like the same result.

In a video titled “How I’d Start a 1-Person Business With Claude AI in 30 Days,” YouTube creator Jordan Welch tells viewers he can copy the playbook behind Matthew Gallagher’s Medvi — the AI-built telehealth company tracking $1.8 billion in 2026 sales — and run a small dropshipping version of it. He shows $400 in sales on day two of his own test store, $100 of which he calls profit. The Medvi number is real. The Claude workflow is real. Almost everything around them isn’t a beginner business.

What the video actually claims

Welch frames the pitch in three steps. Use Claude (specifically the desktop app’s project workflow) with a prompt borrowed from another creator to “scan tens of thousands of data points” and surface a product. Feed your Shopify product page’s Liquid/JSON code into Claude so it rewrites the copy in your competitor’s voice. Then brief Claude on a competitor’s top-performing Facebook image ad, send the resulting prompt to Higgsfield (an AI image generator), and produce ads featuring AI-rendered people and pets.

He anchors all of this to Medvi. Matthew Gallagher, he says, “built what the media is calling the first $1.8 billion solo business” by selling GLP-1 weight-loss drugs through an AI-built website with AI-generated faces and “tens of thousands of AI generated ads of people that weren’t even real.” Welch’s proposal is smaller: “a couple hundred a day” by following the same blueprint with a senior-dog hip-support sling, which is the product Claude ranked first for him.

The case study is the day-two screenshot — $400 in sales, “over a hundred of that being profit” — plus a member of his mentorship program named Odalis who “just passed $100,000 in revenue.” He recommends a $30–$50/day ad test budget to start, with $100/day “if I had more money to invest.” A link to his AI Com Academy mentorship program sits in the description.

What the method actually requires

Strip away the AI talk and the math becomes mundane. A Shopify Basic plan in the U.S. runs $39/month with 2.9% + 30¢ per online card transaction, or $29/month on annual billing. Dropshipping profit margins typically sit at 15–20% after product cost, shipping, ads, and platform fees, which means a store grossing $1,000/month is keeping $150–$200 before its operator pays themselves for time.

The harder cost is ads. Meta CPMs for e-commerce ran $10.80–$23.10 across most of 2025–2026, with health-adjacent categories (pet wellness, supplements, beauty) trending toward the $25–$40 range during peak periods. At a generous 2% landing-page conversion rate and a 1% click-through rate — both optimistic for a brand-new store with zero retargeting list — every $40 of ad spend buys roughly 2,000 impressions, about 20 link clicks, and somewhere between zero and one sale.

To reliably clear $100/day in profit at a 20% margin, a dropshipping store has to gross $500/day, and the ad spend behind that number for a new account is usually well north of $150/day. That’s three to five times the budget Welch recommends as a starter. Can a beginner buy that math at $30/day? The honest answer is: occasionally yes, for a few good days during testing, almost never as a steady run rate.

That gap is where most experiments die. Industry compilations put dropshipping’s “consistent profitability” rate at somewhere between 1% and 20% of stores depending on how strictly you define it; the share of stores ever clearing $50,000/month in revenue is roughly 1.5%. A $400 day-two screenshot is what testing looks like before you know if you have a winner.

It is not a payday.

There’s a second cost the video skips entirely: regulatory risk on AI-generated marketing. The FTC’s final rule banning fake reviews and testimonials took effect October 21, 2024, and it explicitly prohibits creating, selling, buying, or disseminating fake reviews — including ones generated by AI to look like real customers — with civil penalties up to $51,744 per violation. The FTC’s broader AI guidance and updated endorsement rules go further: AI-generated images that read as customer testimonials or doctor endorsements must be clearly disclosed when they could mislead a reasonable viewer. For U.S. readers, that’s the regulator. U.K. sellers face the ASA’s CAP Code; EU sellers face the Unfair Commercial Practices Directive enforced by national consumer agencies; Australian sellers answer to the ACCC.

What about the Medvi blueprint itself?

This is the part the video should have addressed and didn’t. On February 20, 2026, the FDA sent Medvi a warning letter as part of a wave of about 30 letters to telehealth companies marketing compounded GLP-1 drugs. The agency cited misbranding — false suggestions that Medvi itself was the compounder, claims that its products were equivalent to FDA-approved Wegovy, Ozempic, Mounjaro, and Zepbound, and language that misrepresented the agency’s evaluation. Medvi was given 15 days to fix the violations or face possible seizure or injunction.

Reporting since then has stacked up additional concerns: AI-generated before-and-after photos, AI-generated physician personas used in social ads, and a class action filed in California in March 2026 alleging affiliate-spam tactics affecting roughly 100,000 people. Bloomberg has covered the broader push of GLP-1 telehealth marketing to people the drugs weren’t designed for, and 35 state attorneys general have flagged AI-generated physician profiles as a specific harm.

None of that disqualifies Claude as a tool, and none of it implies Welch is doing anything Medvi did. It does mean that when the video holds up Medvi as the model — “tens of thousands of AI generated ads of people that weren’t even real” — it’s pointing at tactics that are now actively under regulator review and active litigation. A dropshipper using the same playbook to sell a dog harness with AI-generated “happy owner” photos isn’t building Medvi at small scale; they’re stepping into a body of FTC enforcement that was written specifically with that workflow in mind.

Who actually wins this game

Three groups. Operators with prior e-commerce experience and meaningful ad budgets ($3,000+/month, not $30/day) who can run enough creative variations to find a winner. Creators who already have an audience, where the “store” is a monetization layer on top of a YouTube channel — which describes Welch himself, a channel approaching two million subscribers. And gray-zone arbitrageurs willing to push hard against advertising and product-claim rules until somebody complains.

What it isn’t is a fair game for the viewer with $200, no email list, no audience, and 30 days. The senior-dog harness Welch’s Claude prompt landed on is already being sold by multiple Amazon listings doing thousands of units a month and at least one Facebook-ad-heavy competitor with 200 active creatives. Beating them on cost requires a supplier edge; beating them on creative requires a brand edge; beating them on attention requires a paid-traffic edge. AI shortens the production loop. It doesn’t fund the auction.

What you’d realistically earn

Compiled industry data on dropshipping income is consistent and unflattering. Beginners typically earn $0–$2,000 in monthly revenue (not profit) in the first three to six months, with most stores never reaching steady profitability. Operators who do break through tend to land in a $1,000–$10,000/month revenue band after six to twelve months of sustained work — at 15–20% net margins, that’s perhaps $150–$2,000/month in take-home, before the operator’s own time is paid.

Compare that to Welch’s day-two demo. $400 in sales with $100 of “profit” is broadly consistent with industry data — for a single good day during a paid-traffic test. It isn’t extrapolatable to $100/day forever. Anyone who has run Meta ads knows that fatigue, account warm-up, and product seasonality eat through “winning” creatives in weeks, not years. The honest range for someone who follows the video’s playbook and grinds for six months is closer to $0–$500/month net than to “a couple hundred a day.”

Who this is (and isn’t) for

This works for someone who already has $2,000–$5,000 they can lose, 15–20 hours a week, prior experience running paid social, a supplier they can call, and the patience to spend three to six months testing before anything stabilizes. It also works as content for someone building a YouTube channel about dropshipping, where the funnel ends in a mentorship program priced around $1,997 — which is roughly what Welch’s AI Com Academy costs.

It doesn’t work for the viewer the video implies: a beginner with no audience, no ad budget, and the expectation that Claude does the heavy lifting. AI shortens the time from idea to launched store. It does not shorten the time from launched store to profitable store, which is where almost every aspiring dropshipper actually gets stuck.

What to remember

Treat the Medvi reference as a regulatory warning, not a roadmap. Treat the $400 day-two number as a testing snapshot, not a daily run rate. And treat any “AI-generated ads of people that weren’t even real” workflow as something a U.S. regulator already wrote a rule about. The Claude workflow Welch demonstrates is genuinely useful for product research and copy rewrites; the 30-day income promise riding on top of it is doing the heavy lifting in the title, not the math.

If you’re comparing Claude-AI side-hustle pitches, you might want to read our breakdown of an alternative Claude side-hustle pitch from 2026 and a week-long AI dropshipping experiment with raw numbers for comparison points before you spend a dollar.

View counts, mentorship prices, and platform pricing may have changed since publication.