How to exploit Claude to reverse engineer any $100K-$1M/mo ecom store's strategy

Ecom bros hear "reverse engineer a $1M store" and the first thing their brain does is go straight to robbery mode.

Copy the homepage. Rip the ads. Clone the product page. Paste the whole thing into Claude and pray it spits out a gameplan that prints money.

That's not reverse engineering buddy. You're just doing ctrl+C with delusions.

And it doesn't work. The store you're staring at right now is just the final exam.

You're looking at the answer sheet without understanding a single question that was on the test.

Behind that clean product page is months of failed products, burnt ad spend, pricing experiments that flopped, and audience pivots you'll never see.

The actual strategy is invisible. The storefront is just what survived.

So the main question is: if all the internal testing is hidden, what can you actually figure out from the outside?

Way more than you think. You just need to know where to look, what to collect, and how to get Claude to connect the dots for you.

That's what this article is. The full system for taking any ecom brand doing $100K-$1M per month and reconstructing how it works, what's likely driving their revenue, and which of those principles you can steal and test on your own store.

One important thing before we get into it: Claude is powerful but it's not a psychic. + recently it's even more ret*rded than ever.

So for this process to actually give you useful output instead of hallucinated nonsense, you need the right instructions loaded into a Claude Project with the right prompts at each stage.

I compiled the full skill files, prompts, and checklists into a document you can grab at the bottom of this article.

That's what makes the difference between Claude giving you generic advice and Claude actually thinking like an operator.

Let's get into it.

First rule: know what's real and what's a guess

Before you analyze anything, you need to accept something that'll save you from looking stupid.

There are things you can verify from the outside and things you absolutely cannot.

Mixing those up is how you end up copying a store that's secretly hemorrhaging cash.

What you can verify: The product costs $49. That ad launched in March. They use Recharge for subscriptions. These are observable facts you can see with your own eyes.

What you can infer: If an ad has been running for 6 months and they keep launching new variations of the same angle, that concept probably survived some internal review. You can't call it profitable, but you can say it lasted. There's a difference.

What you will never know from the outside: Their real CAC, ROAS, margins, refund rate, churn, and ad spend. Anyone telling you they calculated a competitor's exact revenue from SimilarWeb estimates is doing creative writing, not analysis.

Keep this in the back of your head through the whole process.

STEP 1: Pick a brand worth studying

Not every store deserves your time. You need a brand with enough public evidence to actually pick apart.

What makes a good target:

• Active ads on Meta Ad Library or TikTok Creative Center
• Real customer reviews on their site, Amazon, Trustpilot, Reddit, wherever people complain
• A clear product and offer you can map out
• Enough history that they've iterated and changed things over time

If the brand has 3 ads, 12 reviews, and a one-page store, move on. You're looking for someone who's been in the game long enough that their store reflects actual learnings.

The sweet spot is a DTC brand doing $100K-$1M/mo that's been running for at least 6 months. That gives you enough evidence to find real patterns.

Open a new Claude Project for this brand. Everything you collect goes in there.

STEP 2: Decode the customer (this is where 90% of people skip and it costs them everything)

Many of y'all start with the product. But the right way to go about it is starting with the person buying it.

That's because the product, the ads, the funnel, the pricing, none of it makes sense until you understand who this brand is actually serving and what's keeping them up at night.

Go collect everything:

• Product reviews (sort by most recent AND by the 2-3 star reviews because those are where the real expectations vs reality show up)
• Comments on their Facebook and Instagram ads (people literally tell you their objections in the comments for free)
• Reddit discussions about the brand or the category
• Trustpilot or any third-party review sites
• Their FAQ page (this tells you what questions come up so often they had to address them publicly)
• Testimonials on the site

Dump all of this into Claude and ask it to pull out:

• The repeated pain points in customers' actual words
• What outcome they expected when they bought
• What they tried before and why it failed them
• The objections and hesitations people mention before buying
• The exact phrases they use to describe the problem

Here's why this matters so much. The language real customers use is almost never the same language the brand puts on the product page.

That gap is literally a cheat code for many.

You now know how the market actually talks about their problem, which means your ads and copy can speak their language instead of guessing.

Claude will eat through 200 reviews and pull patterns faster than you could in a full day of reading. Those patterns are your market intelligence.

STEP 3: Decode the offer

Now you look at the store. But you're not window shopping. You're mapping the commercial decisions behind every element on the page.

Go through and document:

• Hero product and its price
• Bundles or quantity breaks (buy 2 save 10%, buy 3 save 20%)
• Subscription option and the discount they offer for subscribing
• Free shipping threshold
• Guarantee or return policy
• Upsells in the cart or at checkout
• Cross-sells on the product page
• Post-purchase offers if you can spot them

Feed this into Claude and ask what these decisions suggest about how they think about their economics.

This is where it gets fun because every pricing decision tells a story.

If they're pushing subscribe and save hard with 20% off and free shipping, they've already figured out that their LTV is strong enough to take a hit on the first order.

They're buying customers, not one-time sales.

If they have aggressive quantity breaks like buy 3 get 1 free, they're probably pushing AOV up because their ad costs only work at a higher cart value.

Completely different strategy.

You don't need their actual spreadsheet to understand the logic. The offer structure IS the spreadsheet, just in public.

STEP 4: Decode the ads

Everyone starts here and that's backwards.

You should arrive here after you already understand who the customer is and how the offer is structured. Because now the ads actually make sense in context.

Go to Meta Ad Library and search the brand. If they run TikTok ads, hit TikTok Creative Center too.

For each ad, note:

• The hook (first 3 seconds or first line)
• What problem it leads with
• The promise it makes
• How it explains why the product works
• Format (UGC, studio, image, carousel)
• The CTA
• How long it's been running

Now here's where Claude goes crazy.

Don't analyze ads one at a time like you're doing a school assignment. Feed Claude 10-20 ads from the same brand at once and ask it to cluster the patterns.

Which problems show up across multiple ads? Which hooks keep getting recycled? What format appears the most? Which angles have lasted the longest?

You're hunting for recurring themes, not individual winners.

If 8 out of 15 ads all lead with the same frustration but approach it from different angles, you just found a pain point the brand is repeatedly betting on.

If every ad that's lasted 3+ months features a woman 45+ talking to camera, that persona is doing work.

Quick thing to keep in mind: an ad running for 6 months doesn't automatically mean it's their cash cow. It means it survived.

Maybe the brand is lazy about turning stuff off. Maybe it's a retargeting ad spending $5 a day. Longevity is a signal not a verdict.

STEP 5: Map the funnel

Click their ads. Go through the entire buying experience like you're an actual customer with a credit card in hand.

Because the funnel tells you what beliefs they're trying to build and in what order.

At every single stage ask 3 things:

• What belief is this page trying to create?
• What objection is it handling?
• What action does it want me to take next?

Walk through it:

Where does the ad click land? Product page? Dedicated landing page? A quiz? An advertorial? That first page alone tells you a ton about their conversion strategy.

Product page structure. What's above the fold? How do they handle social proof? Where's the subscription option relative to one-time buy? What kind of images? Any videos? How long is the page?

Cart and checkout. Any upsells? Urgency elements? How many steps to complete purchase?

Post-purchase. One-click upsell offers? Account creation push? Survey? Email opt-in?

Feed all of this into Claude and have it map the belief chain.

Good funnels create a sequence: "I have this problem" flows into "this product solves it" flows into "I trust this brand" flows into "this price is fair" flows into "I should buy now."

If you can see the belief sequence, you understand why it converts. And you can build your own version for a completely different product using the same psychological progression.

STEP 6: Check the history

The store you see today is a snapshot. It's the result of everything they've tested and changed. The old versions tell you what they learned.

Hit up the Wayback Machine and pull up previous versions of the site.

Look for:

• Price changes (they probably tested and found a better point)
• Hero product changes (they might have killed underperformers)
• Positioning shifts (headline, value prop, angle evolution)
• Subscription added or removed (massive strategic signal)
• Landing page additions (new pages usually mean they're testing conversion optimization)
• Ad creative style changes over time

Feed the before and after into Claude and ask it to describe what changed. Not WHY it changed because you genuinely don't know. Just what's different.

The direction of changes tells you what they learned.

If they went from selling singles to pushing bundles, higher AOV probably made their ads profitable. If they added a quiz funnel, they're trying to pre-qualify traffic before the product page.

This is like reading someone's lab notebook. You're seeing the experiments they ran over months compressed into a few screenshots.

STEP 7: Model the economics

You cannot calculate a competitor's real profit from the outside. If someone tells you they did, they made it up. Period.

But you can test whether the model makes sense by running scenarios.

Take what you know:

• Product price and discounts
• Rough COGS estimate (reverse image search the product on 1688 or Alibaba to ballpark sourcing cost, treat that as a rough market price not their actual COGS)
• Subscription structure
• Shipping threshold

Now ask Claude to run scenarios:

"If this brand converts at 2% with a $65 AOV and 60% gross margins, what CAC would keep them profitable on the first order?"

"If 30% of customers subscribe and the average subscriber stays 8 months at 7% monthly churn, what does LTV look like?"

"What would have to be true for this business to work at their current price point?"

You're not pretending to know their numbers. You're asking Claude to help you figure out what would need to be true for the model to be viable.

That tells you if the opportunity is realistic and what assumptions you'd need to hit.

STEP 8: Merge the whole thing

This is where Claude connects every piece of evidence you've collected across all the previous steps.

Feed Claude your full evidence set and ask it to produce:

• The 3-5 strongest transferable mechanisms (not things to copy, but principles that seem to drive the business)
• The evidence behind each one with confidence levels
• What you genuinely cannot determine from the outside
• What you should NOT try to replicate (brand-specific elements that only work because of their history, audience size, or reputation)
• 3-5 original tests you could run on your own store based on these mechanisms

This is the final output. A set of hypotheses backed by real evidence that you can take and test on your own product, your own audience, your own business.

That's what real reverse engineering looks like.

Study the system. Extract the principles. Run your own experiments.

What this looks like in practice

Here's what a hypothetical teardown might look like so this isn't just theory.

You pick a DTC supplement brand targeting women 40-55. You run the process and here's what you find:

The customer evidence shows women repeatedly mentioning energy crashes and brain fog.

The 2-3 star reviews say "took a few weeks to notice anything" which reveals the expectation gap is about speed.

The offer is $49 one-time or $39 on subscription with free shipping. Subscribe and save is the default. They push a 3-bottle bundle at $99.

That tells you they're optimizing for both LTV and AOV.

The ads cluster around 2 angles across 18 creatives: "energy without caffeine" and "clear head by noon." Both use UGC from real women in the age range. The ads that lasted longest all lead with the brain fog angle.

The funnel goes from ad to a dedicated landing page (not the main store), heavy testimonials, then to checkout with a one-click upsell.

Wayback Machine shows they switched from their general product page to the dedicated landing page 5 months ago.

That page probably lifted conversion.

The mechanisms you extract: problem-specific landing pages per ad angle, subscription as default, UGC from actual target demographic, bundle structure for AOV, and brain fog as their strongest hook.

Your tests: apply those same principles to a different product in an adjacent space. (OR you could even find a different product that solves a similar problem and run it up)

The skill files that make Claude actually useful for this

Like I mentioned up top, Claude is only as good as the instructions you give it.

If you just paste a product URL and say "analyze this store," you'll get generic surface-level output that sounds smart but teaches you nothing.

The reason this system works is because each step has specific prompts that tell Claude exactly what to look for, what to compare, and what confidence level to assign.

I put together a complete kit with:

• Master instructions for your Claude Project
• Evidence collection checklists for each step
• Customer decoder prompts
• Offer and economics analyzer
• Ad pattern decoder
• Funnel mapper
• Strategy synthesizer
• Original test generator

You load these into one Claude Project and every brand you want to study follows the same process. Swap in new evidence and the system does the rest.

That full document is at the bottom of this article. Grab it.

P.S. A 14 yr old in our program is using AI + ecom to pull $100K+/mo.

Another is pulling $17k/mo organic with the same setup.

If you want to see how they're doing it step by step, I laid it all out here ($10k/mo guaranteed in 3 months if you actually follow through):

https://bit.ly/3RHQdTc

-AC

Reply "REVERSE" and I'll send you the full Claude skill files + prompts to run this system yourself.