5 Things Your WooCommerce Agency Never Told You

This is not an anti-agency post. Good WooCommerce agencies do excellent work. But the agency model has structural incentives that do not always align with your interests as a store owner — and understanding them helps you ask better questions and make better decisions when you hire.

Contents

• Thing 1: Your plugin count is a ticking clock
• Thing 2: Child theme customisations are technical debt by design
• Thing 3: Page speed scores are vanity metrics in the wrong context
• Thing 4: The hosting tier was sized for the demo, not the checkout
• Thing 5: You may not actually own your plugin licences or your hosting account

Thing 1: Your Plugin Count is a Ticking Clock

Every plugin you add is a dependency. Each one has its own update schedule, its own potential conflicts with other plugins, and its own performance overhead. A store with 50 plugins is not twice as complex as a store with 25 — it is exponentially more fragile. When an update breaks something (and it will), diagnosing which plugin interaction caused it takes hours. Agencies bill for those hours. The incentive to reduce plugins does not always exist.

Two independent 2026 surveys of live WooCommerce stores put a number on this. Metorik’s analysis of 6,000+ stores found an average of 58 active plugins, climbing from 51 on stores under $50k in 2025 revenue to 66 on stores over $1m — one store in the sample ran 237. Studio Wombat’s separate survey of 5,000 stores found a lower average of 30 (21 free, 7 paid, 2 other). The two numbers disagree on methodology, not on direction: plugin count rises with store size in both datasets, because each new capability — subscriptions, loyalty points, freight quoting — usually arrives as another plugin rather than a replacement for one already installed.

The same Metorik sample found 12% of stores still running PHP 7.4, which stopped receiving security patches in November 2022. You can check your own count right now without asking the agency: Tools → Site Health → Info → WordPress lists every active plugin on your install.

Thing 2: Child Theme Customisations Are Technical Debt by Design

When an agency adds custom functionality by copying a WooCommerce template file into a child theme and editing it, they are creating code that WooCommerce’s own compatibility promise does not cover. WooCommerce’s deprecation policy protects hooks and filters deliberately: a deprecated function keeps working, with a warning notice, for “usually over a year or several major releases” before removal. A template file you copied and edited gets no such promise — the moment WooCommerce ships a new version of that file (in a minor release, not only a major one), your copy silently stops receiving the fix, feature, or security patch that shipped with it, and nothing tells you unless you go looking.

WooCommerce does tell you, if you check: WooCommerce → Status → System Status lists every template your theme has overridden, flags the ones whose version number no longer matches core, and names the file and path. Few site owners have ever opened that page. A properly namespaced plugin using hooks and filters — the pattern WooCommerce’s own developer documentation recommends — avoids the problem entirely, but it takes longer to build and the client usually cannot tell the difference at launch. Cart and checkout customisations are the highest-stakes version of this trade-off; migrating off the classic shortcode-based cart and checkout is exactly this kind of deadline, and a copied template is exactly the mechanism that breaks silently when it lands.

Thing 3: Page Speed Scores Are Vanity Metrics in the Wrong Context

A 90+ PageSpeed score in a staging environment with no products, no logged-in users, and no cart activity tells you almost nothing about how your store performs under real load. Google’s own Core Web Vitals — LCP, CLS, and INP — are measured at the 75th percentile of real user visits, not a single lab run against an empty homepage. Many agencies deliver stunning PageSpeed scores on launch day by testing the one page type that never gets slow: the static homepage.

Test a product page with items already in the cart, or the checkout page itself, and the number usually drops — because, as the next section covers, those pages cannot be served from cache the way the homepage can. Six months later, with 5,000 products and real traffic, the store is slow and the agency is gone.

Thing 4: The Hosting Tier Was Sized for the Demo, Not the Checkout

Most agencies quote hosting by storage and bandwidth. The number that actually determines whether your store falls over under load is the PHP worker count — the number of PHP requests the server can process at the exact same instant. Kinsta’s own documentation puts its cheapest single-site plan at 2 PHP threads, rising to 4 on the next tier and up to 16 on its most expensive single-site offering; other managed hosts publish comparable entry-level numbers.

Every worker slot is occupied for the full duration of one request. A tenth visitor arriving while all workers are busy does not get a slower response — it queues, and if no worker frees up within roughly 60 seconds, it fails outright with a gateway timeout.

The part agencies rarely explain: this ceiling has nothing to do with total site traffic, because most of your traffic never touches a worker. WooCommerce’s own developer documentation is explicit that cart, checkout, and my-account pages “need to stay dynamic since they display information specific to the current customer and their cart” — they are excluded from full-page caching by design, on every caching plugin WooCommerce documents. A product listing page can serve thousands of concurrent visitors from a cache with zero PHP workers touched.

A checkout page cannot. So the number that matters is not “how many visitors can this hosting handle” — it is “how many people can be paying at the same second,” and that number is set by the cheapest line item on the hosting quote, not by how well the store is built.

Thing 5: You May Not Actually Own Your Plugin Licences or Your Hosting Account

Premium WooCommerce plugins — payment gateways, subscriptions, freight calculators — sell through a licence key tied to one account, and that key is what unlocks automatic updates and direct vendor support. Some agencies purchase and register these under the agency’s own account rather than the client’s, sometimes bundled into an agency-wide membership that covers many client sites at once. It is faster to set up and easier for the agency to manage across a book of clients. The trade-off, documented plainly by WordPress licensing guides written for agencies themselves: because the software is licensed under the agency’s name, the client cannot get direct support from the plugin developer, and if the agency relationship ends, the agency is left choosing between continuing to renew a licence for a client who no longer pays them or telling that client the licensed software is being taken away.

The same pattern shows up one layer down, in who holds the hosting account and the domain registrar login. None of this is illegal or even unusual — it is a reasonable default when nobody asks for anything different. It just means the honest question to ask before signing is not “what does this cost,” but “if I walked away tomorrow, what would I be able to take with me.”

None of this is an argument against hiring help — it is an argument for hiring help that shows its work and hands you the keys. The same “who owns what, and what happens if you don’t own it” question applies just as directly to whoever eventually connects this store to inventory, accounting, or fulfilment systems, and the parallel checklist for choosing an integration partner and the honest accounting in the real cost of a bad implementation cover it from that side. When that overlap is closer to home, see the integration and implementation work we do.

Frequently asked questions

Related guides

• How to Choose a NetSuite Integration Partner (Without Burning $40k)
• Shopify Is Quietly Becoming an ERP: What That Means for Merchants and Integration Partners
• Amazon vs Your Own Store: Most Fee Comparisons Double-Count