Choosing an e-commerce platform · 1 guide

How to choose an ecommerce platform and migrate without losing sales

How to choose the platform for your store and migrate without losing sales.

Two laptops with e-commerce platform dashboards and a comparison checklist

To choose an ecommerce platform, list what your business needs (payments, tracking, page speed, design control, and order volume), add up the total cost including apps and fees, and test your finalists with a real purchase during the free trial. To migrate, export products and customers, map your old URLs, and only switch the domain after a test order.

This is the central guide in our content on platforms and migration. It's for anyone opening a first store and deciding where to build it, and for operators who already sell and feel their current platform has become the bottleneck. The criteria below are written for merchants who sell with paid traffic, because they feel the difference between platforms most at the end of every month.

What is an ecommerce platform?

An ecommerce platform is the software that runs your online store. It brings together, in one place, what used to take a separate system for each part:

  • Storefront: the pages shoppers see, from the home page to the product page.
  • Catalog: products, variants, prices, and inventory.
  • Checkout: cart, shipping, payment, and confirmation.
  • Orders: status, fulfillment, exchanges, and refunds.
  • Customers and marketing: accounts, email, discount codes, cart recovery.
  • Integrations: payment gateways, carriers, ERP, ad pixels.
  • Hosting and security: servers, SSL certificates, uptime.

Platforms differ in how much of that comes built in, how much depends on third-party apps, and how much control you get over each piece.

What are the types of ecommerce platforms?

TypeHow it worksUpsideDownside
Hosted (SaaS)You pay a subscription and the platform handles servers, security, and updates.Fast to start, no technical team required.You live within the plan's rules and limits.
Open source, self-hostedYou install the software on your own server and maintain everything.Full control and no platform subscription.Hosting, security, updates, and plugins become your job.
HeadlessThe commerce engine is separate from the storefront, which developers build.Front-end freedom and performance.Requires an ongoing engineering team.
AI store builderA hosted platform where AI builds and edits the store from a description.Test niches and offers in minutes with a small team.Everything the AI writes needs a review before it goes live.

Marketplaces aren't on the list because they aren't your store: they're someone else's storefront, with their rules. Plenty of brands run a marketplace and their own store side by side. Our hub on how to build an online store explains when each one makes sense.

Hosted or open source: how do you decide?

Choosing between hosted and open source is really a choice about who owns the infrastructure. With a hosted platform, the vendor keeps the store online, secure, and up to date, and you pay for that in the subscription. With open source, the software is free, but someone has to pay for hosting, apply security updates, keep plugins compatible with each other, and fix things when checkout goes down.

Three questions help you decide:

  • Do you have a developer on call? If not, every plugin update is a risk you can't measure.
  • What does an hour of downtime cost? On a campaign day, a store that's down burns the whole ad budget. Add that to the math.
  • Do you need something no hosted platform does? If the answer is no, the extra control of open source tends to cost more than it delivers.

For most stores that are just starting, or that sell with paid traffic and a small team, a hosted platform is the cheapest option overall, even with the subscription. Open source makes sense when the business already has a technical team and requirements outside the norm.

Which criteria matter when you sell with paid traffic?

Every platform has a feature list. What separates them is how they hold up in the daily routine of someone buying traffic. Use the table as a set of questions for each finalist.

CriterionQuestion to askRed flag
1. CheckoutDoes it support the payment methods my buyers use (cards, Apple Pay, Google Pay, PayPal), and do I know where each one takes the payment?A third-party checkout bolted on top, or surprises about which methods send shoppers to another site.
2. TrackingDoes it send purchase events from the server, alongside the pixel, with deduplication?Browser pixel only, or server-side tracking sold as a separate paid app.
3. SpeedDo pages arrive rendered and light on mobile?A slow product page on cellular because of app scripts.
4. Total costWhat does the plan cost once I add the apps I need?A low subscription that requires four or five paid apps.
5. DesignCan I change any section without a developer?A rigid theme where a simple change becomes a quote.
6. Recovery and CRMDoes it recover carts by email and messaging and keep the customer's history?Every channel in a different app, with no single customer view.
7. Cross-borderCan I sell in another language and currency, with gateways that operate in that country?Checkout in one currency only, or a gateway that doesn't operate where you sell.
8. LimitsDo the product, order, and store limits fit my growth plan?Per-order fees that grow faster than your margin.
9. DataCan I export products, customers, and orders to a spreadsheet anytime?Locked data, with no full export.
10. SupportCan I reach someone when checkout breaks on a Friday night?Forum-only or email replies that take days.

Criteria 1 and 2 outweigh the rest for anyone running ads. A checkout that loses sales and tracking that doesn't send the purchase back to the ad cost more, every month, than any difference in subscription price. The details are in our hubs on checkout and payments and tracking and attribution, and in guides like on-site checkout vs hosted checkout and server-side tracking.

How do you evaluate a platform's checkout?

Checkout is the page with the fewest visits and the most money on the line. Evaluate four things on each finalist:

  • Payment methods: cards, Apple Pay, Google Pay, and PayPal cover most US shoppers. If you sell into other markets, check their local methods too. If your buyers use a method the checkout doesn't accept, that sale doesn't exist.
  • Where the payment happens: on your domain or on another company's hosted page. Each option has trade-offs in trust, conversion, and tracking, so know which one you're getting for each gateway.
  • How many fields and steps: every extra field is a reason to leave. Address autocomplete, guest checkout, and a visible order summary help.
  • What happens after payment: the confirmation email, order status, recovery for shoppers who didn't finish, and an offer on the thank-you page.

Place the test purchase as a customer, on your phone, and count the taps to confirmation. Compare that number across platforms.

How do you evaluate tracking and speed?

Tracking is what lets the ad platform optimize for buyers. Ask whether the platform sends the purchase event from the server (the Conversions API on Meta, the Events API on TikTok) in addition to the browser pixel, and whether both events carry the same ID so the sale is counted once. If the answer is "install an app," add that app to the total cost.

Speed matters because nearly all paid traffic arrives on mobile. Google publishes its page experience targets as Core Web Vitals: the largest element on the page should render within 2.5 seconds (LCP), the page should respond to a tap within 200 milliseconds (INP), and the layout shouldn't jump while loading (CLS of 0.1 or less). Run PageSpeed Insights on the product page of a real store on each platform, not on the platform's own marketing site.

What questions should you ask before you sign up?

  • Is there a platform fee on each sale, on top of the payment processor's fee?
  • What happens if I go over the monthly order or product limit?
  • Can I export products, customers, and orders at any time, including after I cancel?
  • Does checkout work with the payment providers I already use, on my own account?
  • How does support work when checkout breaks outside business hours?
  • Which features I saw in the demo are on my plan, and which are "coming soon"?

Get the answers in writing. They become your comparison baseline and prevent surprises on the first invoice.

What is the total cost of an ecommerce platform?

Compare platforms on total cost of ownership, not the subscription. The math has four parts:

  1. The platform plan at the size of your business, not the entry tier.
  2. Apps for every function the plan doesn't cover: checkout, tracking, cart recovery, email, reviews, page builders.
  3. Per-sale fees: the payment processor, and on some platforms, a platform fee on every order.
  4. Cost of change: developer hours for every layout tweak and the time you lose waiting.

Build one row per platform with these four items for a typical month of orders. The full list of store costs, with a worked break-even example, is in how much it costs to start an online store.

How do you compare platforms during a free trial?

A free trial is for finding problems before you pay, not for touring the dashboard. Run the same script on every finalist:

  1. Build the same store. One real product, with the photos, price, and description you'll actually use.
  2. Change three things in the layout. A headline, the order of two sections, and the main image. Time it, and note whether you needed help.
  3. Connect the payment provider you'll use and make a small real purchase with each method. Then refund it.
  4. Install your pixel and check that the purchase event reaches your events manager exactly once.
  5. Open the product page on your phone, off Wi-Fi. If it feels slow to you, it's slow for your customers.
  6. Export your products and orders. Check that the file is complete.
  7. Send support a question and note how long the answer takes.
  8. Add up the total cost with the apps you'd need to install to get the same result.

One afternoon with this script tells you more about a platform than any comparison chart.

When is it time to switch platforms?

Switching takes work, so it only pays off when staying costs more. The most common signs:

  • Your app bills add up to more than the platform subscription, and keep growing.
  • Sales in your ads manager don't match your store, and the platform has no server-side tracking.
  • Checkout doesn't accept a payment method your buyers use.
  • Every page change needs a developer, and offer tests take weeks.
  • The product page has slowed down because of app scripts.
  • You want to sell in another country and the platform doesn't support that currency or a gateway there.

If two or more of these show up at once, run the total cost math from the previous section on your current platform and one alternative.

How do you migrate platforms without losing sales or SEO?

A good migration is invisible to customers: they land on the same domain, find the same products, and check out the same way. Follow this order.

  1. Take inventory. Export products, customers, and orders from your current platform. List every URL that gets traffic: home, collections, products, blog posts, and policies.
  2. Build the new store in parallel on a temporary address, without touching the domain.
  3. Import the catalog and customers and check price, variants, inventory, and weight on every product.
  4. Map your URLs. For each old URL, define the new one. Wherever the path changes, plan a 301 redirect and confirm with the new platform how to set it up. Google documents the process in site moves with URL changes.
  5. Rebuild the integrations: payment provider, shipping, transactional email, pixels, and the Conversions API. Write down your pixel IDs before you shut the old store off.
  6. Place test orders with each payment method and check the purchase event in your ads manager.
  7. Switch the domain during a low-traffic window by pointing DNS to the new store. Keep the old store running until propagation finishes.
  8. Monitor for two weeks: 404 errors in Search Console, conversion rate, purchase events, and customer complaints.

Don't pause your ads during the switch unless you have to. If the pixel and the domain stay the same, your campaigns keep learning on the new store.

Only tell repeat customers if something changes for them, like the account login or a payment method. If everything stays the same, the best migration is the one nobody notices. Keep the export from your old store for a few months: it's your backup of orders for returns and exchanges.

What are the most common migration mistakes?

  • Switching the domain before testing a purchase. Your first customer finds the bug for you.
  • Forgetting the old URLs. A product that moves without a redirect loses its Google ranking and breaks links in ads and emails.
  • Creating a new pixel. Your campaigns lose their learning history. Reuse the pixel you already have.
  • Migrating during peak season. Black Friday week and launch week are not migration weeks.
  • Copying old policies without reviewing them. Shipping times, costs, and return rules may have changed with the operation.
  • Leaving transactional emails on defaults. Order confirmations and shipping updates are the emails customers actually read. Rewrite them in your brand's voice and test each one before launch.

Most of these mistakes come from rushing the last step. Give the new store a few days running in parallel on its temporary address, with a handful of real orders, before you move the domain. The time you spend there is cheaper than a week of broken checkout.

What changes with an AI ecommerce platform?

An AI platform mainly changes the cost of change. Building a new store, testing a different product page, or creating a version of the store in another language stops being a project and becomes a request. In a migration, that means rebuilding the layout of your current store from its link and adjusting it by chat, instead of reconstructing a theme piece by piece.

The selection criteria stay the same as in the table: checkout, tracking, speed, and total cost. AI saves time; it doesn't replace review. To see how it works from brief to first order, read how to create an online store with AI.

Where does Horsee fit, and where doesn't it?

Horsee is an AI ecommerce platform built for merchants who sell with paid traffic. You create the store from a description, a reference link, your own HTML, or a template, and edit it by chatting or in code. Every plan includes checkout with global gateways (Stripe, PayPal, Ziina, Whop) plus Brazilian providers for local methods like Pix, server-side tracking for Meta and TikTok with deduplication, WhatsApp and email cart recovery, a CRM, and an AI photo studio.

It isn't the right choice for everyone. There's no permanent free plan: the free trial lasts 3 days, then you need a subscription. Every plan has limits on stores, products, and orders, so if your catalog runs into the thousands of SKUs, check the numbers first. And B2B operations with customer-specific price lists and minimum order quantities usually need a specialized platform.

If you're coming from another platform, products come in by spreadsheet, the layout can be rebuilt from your current store's link, and the domain is connected under Settings → Domains. If you use Shopify, Horsee can send its orders to your Shopify account. Step-by-step guides are in the Horsee Help Center, and the limits and prices for each plan are on the Horsee pricing page.

From the guide to a live store.

Build your store with AI in minutes, with checkout, payments and server-side tracking ready.