Why we removed the cart from the store completely, and the right COD architecture

A decision that seemed strange to store owners at the time: we said we'd remove the cart from the site completely. The result was a CVR of 3.14% and 21.7% fewer lost orders in one month.

Let me explain the idea, because it has nothing to do with ads. It's about architecture.

The wrong question and the right question

Most stores ask: "How do I get more traffic to checkout?" The right question is: "The customer who has already decided to buy, where do we lose them along the way?"

On COD, the customer makes the buying decision while still on the product page. No wallet, no registration, no online payment. So why do we pass them through Add to Cart, a basket and a checkout page? Each of those extra steps isn't an "experience". It's a place where orders you've already paid ads for get lost.

What we did exactly

• We removed the cart altogether. Any buy button leads to one thing: a landing page, then the order form, and that's it. No "added to cart" page, and no empty cart icon at the top distracting the customer from the decision they came to make.
• The landing page closes the sale before the form. It answers every objection before the customer asks: what the product does, proof, reviews, licensing and FAQs. Whoever reaches the form arrives convinced, so the form became data entry and not another point of negotiation.
• The home page was built on the bundle and not the catalog. The first thing you meet is the main offer and a direct try button, not a product gallery. A customer coming from an ad for one specific product never needed to get lost in a catalog in the first place.

Each of these three changes serves one idea: every extra screen between the customer and the buying decision is a chance to leave, not a step that builds trust. In stores with online payment, a cart really does protect the customer (they review the order, add another product), but on single-product COD that benefit doesn't exist at all, so it goes from an advantage to just an obstacle.

The full numbers from the case

In one month: 3.64 million EGP in sales, and 4,220 orders at an average order of 862 EGP. Conversion rate rose to a CVR of 3.14% (up 5.7%), and most importantly, lost orders fell 21.7%. That's the cleanest money in the whole business, because those orders' ads were paid for and they were getting lost at the form.

And the number that confirms this wasn't a traffic fluke: visits that month were lower than the month before, yet the number of orders rose 2%. So the same visitors, or fewer, converted more, because the distance between them and the form got shorter. The full story with numbers is in the no-cart store case.

Lost orders aren't the same problem as RTO

It's important to make the difference clear here because people mix them up. Lost orders happen before the order is registered at all: the customer starts the form and abandons it halfway, for whatever reason, whether the internet cut out, a question confused them, or simple distraction. That's the part a no-cart architecture solves.

A completely different problem is RTO, when the order is registered, shipped, and then the customer refuses delivery. That happens long after the form is done, and its fix is in confirmation and delivery, not in site architecture. If you have both problems together, start with lost orders first, because they're easier to fix and the effect is immediate.

But this isn't a decision to generalize

If you have dozens of products and the customer buys from more than one category in the same order, the cart is essential and removing it is wrong. For example: a clothing store with shirts, pants and accessories, where the typical customer takes 3 different things in one order. Here the cart is what lets them gather their order before deciding, and removing it would cost you average order value without gaining conversion.

This architecture shines in one situation only: one hero product or a fixed bundle, a simple and quick buying decision, and cash on delivery with no online step that calls for a basket. If your store looks like this, review the elements of a product page on Shopify before you touch the architecture, because the problem is usually the page content and not the number of steps.

Is that 3.14% good or not?

A question that comes up a lot after this case: is 3.14% really high? The answer: it's within the healthy range for a specialized store in Egypt on cash on delivery, which runs between 3% and 7% according to the ecommerce benchmarks for Egypt and the Gulf. So this store isn't only in the range. It's at the end of it that a no-cart architecture specifically gets you to. Other stores with a stronger bundle or a different offer reached higher numbers, as shown in the natural conversion rate for an online store in Egypt and the Gulf.

What matters here is not to compare yourself with a store selling a big catalog with a cart and online payment. That's a completely different model and its normal numbers are different. The architecture in this article serves one model: one product or bundle, COD, and a simple buying decision.

The form is the real test

When you remove the cart, the form becomes the only link left between the customer and the order, so it has to be short and clear: name, phone, address, and that's it. Any extra field in the form now has a bigger effect, because it has no competitor or alternative left in the journey.

And before you apply any of this, you have to make sure your pixel is recording correctly on the new path. If the purchase event isn't recorded accurately from the new form, you'll decide to develop or drop it based on wrong data. And if the store is still in its first months, start with the first 5 things before your first ad before thinking about funnel architecture at all.

How to test the idea without overhauling the whole store

You don't have to remove the cart suddenly with no way back. The safe way is to start with just one product or bundle, your best seller, make it a standalone landing page with the new path (landing page → direct form), and point a separate ad at it. If the numbers (CVR, lost orders, cost per order) come out better than the old path over a long enough period, at least two weeks, extend the decision to the other similar bundles. If the whole store is one catalog with a clear hero product, you can make the decision for the whole site from the start.

The point to track throughout the test is the same one this case measured: not just conversion rate, but the number of orders that reach the form and don't abandon it. That's the number that actually shows whether the new architecture reduced friction, apart from any natural swing in traffic.

In short: an extra step doesn't mean a better experience. For a customer who has decided to buy, each second between you and the form costs you money. If you feel orders are getting lost somewhere between the ad and the form, let's do a free audit and see exactly where.