The expensive problems in ecommerce rarely announce themselves. These are the ten that show up most often across thousands of real stores, and how to tell whether they’re on yours.
Most of the problems that cost online stores money don’t look like problems. There’s no error message, no crash, no angry email from a developer. The store loads, orders come in, and a handful of small technical failures quietly tax every visit: a missing line in a config file, a piece of structured data that was never added, an image that loads a fraction of a second too late.
We audited 80,000 ecommerce stores to find out which of these failures are the most common. The sample covers four platforms (Shopify, Shopware 6, Magento 2, and PrestaShop) at 20,000 stores each, audited in early 2026. It’s the English-language sample, and the full methodology and its limitations are set out in the benchmark report. The ten problems below are the ones that turned up again and again, no matter which platform a store was built on.
They’re ranked by how common they are, not how severe. Common is not the same as critical: a missing SPF record (number one) is a serious deliverability and security failure, while unminified code (number ten) is a minor drag. Every one of them is something an automated ecommerce audit can detect, so you can check your own store against the list at the end.
A result here means the scan detected a signal’s presence or absence across 80,000 live stores. It is not a measure of business impact on any single store.
The 10 at a glance
| Rank | Problem | Stores affected |
|---|---|---|
| 1 | No valid SPF email record | 34,755 |
| 2 | No return-policy schema | 33,148 |
| 3 | No XML sitemap | 32,962 |
| 4 | Images not lazy-loaded | 27,753 |
| 5 | No visible return-policy link | 27,219 |
| 6 | Misconfigured robots.txt | 25,999 |
| 7 | No AI-readable page structure | 25,683 |
| 8 | No shipping schema | 25,515 |
| 9 | No DMARC policy (self-hosted) | 23,909 |
| 10 | Unminified CSS/JavaScript | 21,422 |
Counts are across all 80,000 stores, except DMARC, which is measured on self-hosted platforms only.
1. Email anyone can send on your behalf
The most common failure in the entire benchmark has nothing to do with your storefront. It’s email. On 34,755 stores, the domain had no valid SPF record, the DNS setting that tells the world which servers are allowed to send mail in your name. Without it, anyone can send email that appears to come from your store, and mailbox providers, unsure whether your real messages are legitimate, start routing them to spam. Order confirmations and shipping updates that land in spam turn into support tickets, refund requests, and customers who don’t come back.
How to fix it: set up SPF and email authentication.
2. A return policy search engines and AI can’t read
AI shopping assistants and search engines increasingly rely on structured data, not just the visual content shoppers see. On 33,148 stores, the second most common failure we found anywhere, the return policy existed for humans but carried no schema markup, so machines couldn’t parse it. When a shopping assistant is asked “can I return this,” a store with no return-policy schema is not part of the answer.
How to fix it: add return-policy structured data.
3. No proper sitemap
Roughly 41% of the stores we scanned, 32,962 of them, lacked a properly configured XML sitemap, making it the most common search-visibility failure in the benchmark. A sitemap is how you hand a search engine a map of your catalog instead of making it find every page on its own. Without one, new products and price changes get discovered slowly, and large catalogs get crawled incompletely.
How to fix it: configure your XML sitemap.
4. Every image loading at once
On 27,753 stores, the most common speed problem we measured, images loaded eagerly: the browser pulled down every picture on the page, including the ones far below the fold, before the shopper could do anything. The result is a slower first impression and worse Core Web Vitals, which Google treats as a ranking factor and which also drive the rate at which visitors give up before the page settles.
How to fix it: enable lazy loading.
5. A return policy shoppers can’t find
Separate from the machine-readable version is the human one. On 27,219 stores, there was no clear front-end link to the return policy where it matters, near the product, the cart, the checkout. Returns are a trust question, and a shopper who can’t confirm how returns work before paying is a shopper more likely to close the tab.
How to fix it: add a visible return-policy link.
Recognizing a few of these already? A free audit checks your own store for all ten in a couple of minutes. Scan your store.
6. A robots.txt working against you
On 25,999 stores, the robots.txt file was misconfigured. It’s the instruction sheet that tells crawlers where to go and where to stay out, and the failure cuts both ways: blocking pages that should be indexed, or wasting crawl budget on pages that shouldn’t be. Either way, the wrong parts of the store end up in search.
How to fix it: fix your robots.txt.
7. Pages an AI can’t summarize
Large language models work by extracting the gist of a page. On 25,683 stores, the page gave them nothing clean to work with: no clear heading hierarchy, no summary, no machine-readable structure. A page an AI can’t summarize is a page it’s unlikely to cite or recommend, and that matters more every quarter as buyers ask an assistant instead of typing a query.
How to fix it: strengthen product content depth and structure.
8. Shipping details invisible to machines
Delivery cost and time are among the first things a buyer wants to know and among the most persuasive things you can show. On 25,515 stores, that information carried no structured markup, so it couldn’t appear in the rich results and AI answers where it would do the most good.
How to fix it: add shipping information and schema.
9. Email authentication left half-finished
SPF (problem number one) says who is allowed to send mail for you. DMARC says what should happen when someone fails that check. On 23,909 self-hosted stores there was no DMARC policy, which means even a correct SPF setup had no enforcement behind it. Hosted platforms handle parts of this for you, so this one weighs most on self-managed stores.
How to fix it: complete your DMARC policy.
10. Code shipped without minification
Despite modern build tools that minify automatically, 21,422 stores still served raw CSS and JavaScript, full of the spacing and comments meant for developers rather than browsers. It’s invisible to look at and straightforward to fix, and until it is fixed, every visitor downloads more than they need to, on every page.
How to fix it: cut render-blocking and unminified assets.
The pattern worth noticing
Sort the ten by what they actually break and a clear picture appears. Two are about email deliverability (1, 9). One is about whether shoppers can find the answer they need (5). Two are about whether search engines can crawl you (3, 6). Three are about whether AI assistants can read and recommend you (2, 7, 8). Only two are about raw speed (4, 10).
The surprising finding from 80,000 audits is that most ecommerce problems are not visible to shoppers at all. They are invisible to search engines, email providers, AI assistants, and browsers, the systems that increasingly decide whether customers find your store, trust it, and finish a purchase. None of them throw an error, which is exactly why they are so common: nothing forces anyone to fix them.
And prevalence is not a diagnosis. The fact that 41% of stores lack a sitemap tells you nothing about yours specifically. The only way to know which of these ten apply to your store is to check it, which takes a couple of minutes.
Want platform-specific findings?
The benchmark breaks down by platform too: