The platform decision, and why it is not about features
Every platform comparison you will read lists features. Almost none of them matter, because the platforms have converged and they all do the obvious things.
What actually decides it is who maintains the store afterwards. A hosted platform charges a monthly fee and handles security, updates and uptime, so a business without technical staff is buying the absence of a problem. A self-hosted store is cheaper monthly and hands you the plugin conflicts, the update schedule and the security patching.
The second factor is catalogue complexity. Fifteen products with two variants each is a different problem from four thousand SKUs with inventory syncing to a physical shop, and the second one rules out most of the easy answers.
The third is what has to connect: invoicing, inventory, a CRM, a shipping account. Integrations are where ecommerce projects overrun, and they should be listed before anyone quotes, not discovered in month two.
We do not have a default platform we push. The right answer changes with those three inputs, and an agency with one answer for everyone is describing what they like to build.
Payments and shipping are the Mexican part
This is where stores built from international tutorials break, because the assumptions do not transfer.
Card penetration is not universal, so a checkout that only takes cards excludes buyers who would have paid. Cash-based and transfer-based methods are still a meaningful share of Mexican ecommerce, and supporting them is a revenue decision rather than a technical one.
Months without interest are an expectation in higher-ticket categories, and whether your processor and your margin can absorb them changes what you can sell online.
Shipping has to be real, not a flat number invented at launch. Rates by zone, a delivery window the customer sees before paying, and tracking that does not require them to write to you. Surprise shipping cost at the final step is one of the most reliable causes of abandonment anywhere.
Invoicing matters more here than in most markets, because a business buyer who cannot get a CFDI will buy elsewhere. If your customers are businesses, this belongs in the first conversation.
Where the revenue actually leaks
Forced account creation before checkout. Guest checkout costs nothing to enable and removes a step that loses buyers who were ready.
Shipping cost revealed at the last step. If it cannot be shown earlier, show the rule earlier.
Product photography that does not answer the question. Buyers cannot hold the item, so the photograph is doing the job of the shelf. Scale, texture and what is in the box are worth more than a styled hero shot.
No abandoned-cart follow-up. It is the highest-return automation in ecommerce and it is frequently not installed.
Mobile checkout that was never completed on a phone by the person who built it. Test it on a phone, on mobile data, with a real card.
Tracking that reports sessions but not revenue by source, which makes every later marketing decision guesswork.