My product sells on Amazon but Meta ads to my own site won’t scale. Is that an ads problem or a pricing problem?

Low-ticket D2C founders keep hitting the same wall: the marketplace moves units, the own-site campaigns start well and then decay, and raising the budget makes it worse. This walks through the arithmetic that decides whether Meta can work for you at all, where the funnel actually leaks, and when a specialist is worth hiring.

Quick answer if you’re skimming

• If your average order value is under about ₹400, Meta on cold traffic will usually lose money on the first order. The arithmetic is the constraint, not the targeting.
• Work out contribution per order before you touch the ad account: price minus COGS minus shipping minus packing minus gateway. Whatever is left is your entire CAC budget.
• Marketplaces subsidise this for you. Amazon already paid to get the shopper there; your own site has to buy every single visit.
• Fix AOV first — multipacks, a free-shipping threshold, one checkout add-on. A three-pack can turn a loss-making order profitable without touching CPM.
• Campaigns that start strong then decay are usually a small addressable audience being re-served. Check frequency before you rebuild creatives.

Why does the same product sell on Amazon but not on my own site?

• A marketplace charges you commission on a sale that already happened. You pay after the fact, out of margin.
• Your own site charges you for attention before the sale. You pay Meta whether or not anyone buys.
• A marketplace shopper is already in buying mode. A Meta scroller is not — you are paying to create the intent, which is far more expensive than capturing it.
• So the same product at the same price and the same margin can be profitable on a marketplace and loss-making on your own site at the same time. Both results are correct.
• That is not a reason to abandon your site. It is a reason to sell something different there: bigger baskets, bundles, exclusive variants, subscriptions.

How do I work out whether I can afford Meta at all?

• One page of arithmetic, before any campaign decision. Worked example on a ₹249 impulse product:

• That ₹99 is your entire customer-acquisition ceiling — and that is to break even, not to make anything.
• A ₹249 impulse buy on cold Meta traffic in India rarely acquires for under about ₹150. At a ₹160 CAC you are paying ₹61 for every order you win.
• Your numbers will differ from these. The line that surprises most founders is shipping, because it usually sits in a different spreadsheet from the ad spend.
• Run this before you raise the budget, not after. Scaling a negative contribution margin just loses money faster.

Everyone says raise the budget. Why does my campaign die every time I do?

• The pattern founders describe is consistent: strong first few days, steady decay, never recovers, repeat for months.
• There are two very different causes and they need opposite fixes, so diagnose before you act.
• Small addressable audience: at a few thousand rupees a day on a single-product niche, you exhaust the people who would ever buy. Frequency climbs, CPA follows. New creative will not fix this.
• Actual creative fatigue: click-through rate falls while reach is still growing. Then new creative does help.
• Editing budgets mid-flight does reset learning, but that is a few days of noise — it does not explain two months of decline.
• Rule of thumb: if frequency is above about 3 on a cold prospecting campaign inside a week, the audience is your ceiling, not the ads.

Should I hire a Meta specialist, an agency, or neither?

• Not yet, if you have not compared marketplace conversion rate and AOV against own-site conversion rate and AOV side by side. No specialist can fix a ₹99 ceiling.
• Founders in these threads repeatedly recommend a small paid funnel audit before any hire — a few hours of someone’s time rather than a retainer.
• Early on, a freelancer or a small agency per channel is usually cheaper than building a team.
• Interview on numbers, not channels. Ask for a store where they owned revenue, and for AOV, CAC and repeat rate both when they joined and when they left. Then ask them to diagnose “Meta ROAS dropped 40%” on the spot.
• Be careful with the three-to-six-year CV where every number came from inside a system somebody else built.
• Sequencing disagreement worth knowing about: some operators argue you should hire for catalogue, pricing and UX first and ads much later. On a low-AOV store, they are usually right.

What actually raises AOV on a low-ticket product?

• Make the multipack the default, not an option. The three-pack is the hero; the single unit is the downsell.
• Set a free-shipping threshold just above the multipack price. Shipping is the largest hidden line on cheap goods, so make the customer’s cheapest path the one that also helps you.
• One checkout add-on, not five. Five options is a decision; one is a nudge.
• Exclusive variants on your own site only — a flavour, size or colour the marketplace does not carry. One buyer in these threads said plainly that they will absorb a little extra cost for something they cannot get elsewhere, but not much.
• Do not discount to lift AOV. You are already short of contribution.

My traffic is fine but nobody adds to cart. Where’s the leak?

• Real funnel numbers a founder posted this week: 3% of visitors added to bag, 71% of those started checkout, 10% of those paid — 0.2% conversion overall.
• Read that from the top. The add-to-cart rate is the problem; checkout is converting fine. The product page is failing before the cart exists.
• Commenters’ diagnosis was uniformly pre-ad: no reviews, no named human behind the brand, no on-body photos per size with model height and size, tech specs where a story should be, and a mobile carousel that hid the other products entirely.
• Split add-to-cart rate by mobile versus desktop before anything else. Indian D2C traffic is overwhelmingly mobile and the two figures often differ by a factor of two.
• Change one thing at a time. Six product-page changes at once and you will never know which one worked.

Should I just stay on the marketplace?

• It is a defensible position and some operators argue exactly that: marketplace for volume, own site as overflow only.
• The counter-argument is that you never own the customer and you cannot raise AOV through someone else’s checkout.
• The honest middle is to run both and stop expecting the same price point to work in both places.
• On moving marketplace buyers to your own site: a discount code or a “shop direct and save” insert is what gets accounts flagged, because it diverts a sale off the marketplace. Warranty registration, care instructions or a QR code to a loyalty list carry no monetary incentive and generally pass review.
• Expect the crossover to be modest. One operator’s blunt point: most of those buyers came for the marketplace, not for you.

How do I test this without spending more?

• Build the multipack. It costs a SKU and a photograph.
• Set the free-shipping threshold above it.
• Hold ad spend flat for two weeks. You are testing the store, not the auction.
• Compare marketplace AOV against own-site AOV at the end of it, then redo the contribution arithmetic.
• If the ceiling moved, scale. If it did not, the product needs a different price architecture before it needs a bigger budget.

What I’d actually do

• Work out contribution per order on a sheet of paper today, before touching the ad account. If it is under about ₹100, no amount of targeting will rescue the campaign.
• Ship a three-pack and a free-shipping threshold this week with spend held flat, then compare marketplace and own-site AOV.
• Buy a one-off funnel audit before you buy a retainer, and interview whoever you hire on AOV, CAC and repeat rate rather than on which channels they have run.