2HMG Performance Case Study

DEVELOPMENT

Back Up to Speed: Engineering a High-Volume Meal-Prep Store Past Its Own Ceiling

1 2 3 4 5 6

Case and Challenge:

MegaFit Meals is an established meal-prep brand with a real, hard-working store behind it — a substantial customer base, a meaningful book of active subscriptions, and traffic that spikes hard whenever an influencer posts a drop. This is not a hobby shop. It is a serious commercial operation that has grown for years.

The challenge was one of headroom, not effort. Success had a cost that nobody had set out to pay: over years of growth, the store had quietly accumulated the weight of every plugin, integration, and data table it had ever touched. By the summer of 2025 the site had become the bottleneck on its own momentum. It was running at a small fraction of the capacity it should have had, and that ceiling sat underneath everything the marketing team tried to do. More ad spend, more launches, more influencer drops — none of it could compound, because the store simply couldn’t carry the traffic. When a big drop landed, the site was something to protect from the surge rather than push hard against.

An independent third-party audit in the summer of 2025 put it plainly: the store was running at a small share of what it was capable of. That gap — between what the store was doing and what the same store could do once the drag was engineered out — was the opportunity we were brought in to capture.

Client: MegaFit Meals

Project Type: WooCommerce performance engineering — ongoing

More Info: A multi-front engineering engagement that began in the summer of 2025: database and plugin hygiene, a rebuilt caching architecture, front-end and theme-level performance work, in-house autoload monitoring, and a continuous weekly performance review.

More Info:

What They Need From Us

1. Stripped the Store Back to What It Actually Needed: Years of growth had left a thick layer of unused and redundant code and data sitting in the store’s path on every single request. We removed a large body of plugin code the store no longer used — a legacy image optimizer, an old tax integration, a popup tool, a mail-logging plugin dragging a huge capture table behind it, an error-tracing tool that wrapped nearly every operation in instrumentation, a heavy security plugin, a duplicate customer-sync feature buried inside the loyalty system, and several smaller utilities — and brought the active plugin count down sharply. We capped or dropped runaway log tables, purged completed scheduler logs, swept on-disk logs left behind by retired tooling, archived old order data, and reclaimed the freed space. The database had been carrying tens of gigabytes of bloat, much of it in one runaway table; we drove it down toward a fraction of its original size. All of it landed with zero functional regressions.
2. Rebuilt the Caching Architecture Cleanly: The store had two caching layers fighting each other over response headers and invalidation, which caused stale-content incidents and constant double overhead. We retired the conflicting layer entirely and rebuilt caching from a clean footing: per-template rules that switch off scripts where they aren’t needed, asset optimization, position-aware lazy loading, deferred and delayed JavaScript with a strict checkout safe-list, preload and prefetch on the hero image and primary font, throttled background processing, and a hardened login path. We also replaced the heavy security plugin with a lighter approach and swept out everything it left behind.
3. Hardened the Moments Customers Actually Feel: We wrote custom store-level code aimed squarely at the slowest real-customer experiences. On the shop page we ran a largest-contentful-paint pass — getting the hero image requested early, connecting ahead of time to the origins the page depends on, and caching product data in short-lived, explicitly-invalidated buckets — which cut the load delay and halved the database work behind each product badge. We added two-tier caching to the quick-view experience, keyed by role, locale, currency, and version, so repeat views skip the heavy work entirely. We also cleared the critical path of everything that didn’t belong on it: marketing trackers that had collected in the page head were moved down and set to start up later, so the storefront no longer waits on them to paint; a product-page timer that fired constantly to watch for a change was replaced with an event that fires only when something actually changes, and scroll handling was tied to the browser’s own paint cycle, cutting wasted work and the jank that came with it, especially on mobile; a sluggish one-second delay on the quantity controls was tightened to feel instant; and below-the-fold images were set to load only as a shopper reaches them, in a lighter modern format with safe fallbacks.
4. Brought Discipline to the Theme: Working safely in a child theme over the store’s existing parent theme, we added preload directives for the hero image on every hero-driven landing page, replaced the theme’s blanket “load everything at once” behavior with position-aware loading so the true lead image gets priority, and deferred a set of heavy scripts. We also found and removed a callback that had been firing on every product, order, and subscription save — quietly making admin saves take several seconds each. Because reliability is part of performance, we fixed what was breaking behind the scenes too: a packing report that buckled on large date ranges, because it loaded full orders one at a time, was rewritten to ask the database directly, and a heavy loyalty-points query got a targeted index. Together these gave the team back a fast, dependable back office that holds up under the same spikes that hit the storefront.
5. Built Our Own Monitoring Where the Platform Was Blind: WordPress autoload bloat is a silent performance killer, and the platform gives no live signal for it. So we built and shipped a small in-house tool that tracks autoload health continuously — a live dashboard of the autoloaded option count, total size, and the top offenders, plus a daily metrics feed. Now a single bad install shows up on the chart the moment it happens, instead of surfacing weeks later through slow pages. When the platform doesn’t expose a critical signal, we build the smallest thing that does.
6. Made Performance a Standing Discipline, Not a Fire Drill: We stood up a weekly performance review tied to a live dashboard that runs daily tests on the pages that matter — home, shop, product, and the order-builder flow — rather than a generic homepage score. It tracks page-speed metrics next to business signals on the same view, so the funnel reads as one picture. Anomalies are filed and triaged fast: a layout-shift regression on the homepage was caught and resolved within a week of appearing. Performance is now something we watch every day, not something we chase after an incident.

Results and Accomplishments

Capacity Ceiling Lifted: We took the store off the hard capacity ceiling that the independent audit had flagged, freeing it to do what the business actually needs it to do.
Critical Pages Dramatically Faster: We cut the shop page’s load to a fraction of what it had been — from a slow, multi-second backend response to a fast end-to-end load, measured in production rather than staging.
Massive Cleanup, Zero Regressions: We stripped tens of millions of rows of accumulated bloat, dozens of unused tables, and a stack of unused plugins out of the store’s path, driving the database down toward a fraction of its original size — without a single functional regression.
Holds Under Load: Cart and add-to-cart flows that had been breaking under simulated launch-day pressure now hold steady, so a big drop is something to push hard against rather than protect.
A Lighter, Steadier Storefront: With marketing scripts off the critical path, constant polling replaced by event-driven work, scroll tied to the paint cycle, snappier cart controls, and images that load only when a shopper can see them, pages paint sooner and interactions feel instant — most noticeably on mobile. The back office holds up too: reports and queries that used to buckle under big spikes now run cleanly.
Performance Tied to the Business: Page speed on the critical pages is now tracked weekly against real business signals on one shared view. In the early measurement period, performance gains and business numbers moved together — a relationship we now monitor continuously rather than guess at.

Conclusion

Through our partnership with MegaFit Meals, we helped turn a store that had been capping its own growth into one engineered to carry it. We didn’t hand over an audit and walk away; we re-engineered the store under the hood, built our own monitoring where the platform fell short, and stood up a discipline that keeps it fast as the business keeps growing. The store is no longer the thing the team has to work around — it’s ground they can build on.

The Feedback:

“From day one, 2H Media Group has been an indispensable and trustworthy partner in our journey at MegaFit Meals. Over the past decade, we’ve watched our vision evolve from an ambitious idea into a national powerhouse, and 2H Media has been there every step of the way. Their ability to craft a strategy and, more importantly, execute it with precision, has taken us beyond what we ever imagined. Not only are they driven and results-oriented, but they are also incredibly easy to work with, consistently making the complex feel simple. They understandd our goals and pushed us to achieve them at the highest level. Their dedication and reliability make them the perfect ally in our relentless pursuit of success.”