CompSource (compsource.com) is a long-established online retailer with a catalog of over a million products, running on a platform built two decades ago (classic ASP, IIS, and SQL Server), with orders flowing around the clock. Proven, profitable, and increasingly fragile. This case study is published with the client's permission.
By the time we were engaged, the problems had stacked up:
The conventional advice for a platform this age is a full rewrite: expensive, slow, and risky. The client asked us for a better answer.
Performance rescue. We instrumented the platform end to end and traced the slowdowns to their true source instead of treating symptoms. Targeted database work, rehearsed and validated against a copy of production before it touched the live store, brought peak page loads from around 30 seconds back to under a second, and took a substantial bite out of database load in the first pass alone.
Stability engineering. Rather than rebooting through each mystery failure, we gathered evidence across weeks of incidents, traced them to a single underlying condition, and addressed it at the root. We backed that up with monitoring built on one principle: a job does not get to call itself healthy, it has to prove it on every run, with output we can check.
Bot and scraper defense. Most of the site's bandwidth turned out to be industrial-scale scraping of the catalog, traffic deliberately engineered to evade conventional blocking. We shut it down at the network edge, verifying first that checkout and payment paths could not be affected, then confirming afterward that search-engine crawlers had not been caught in it. Bandwidth dropped by roughly 80% within minutes, and database load fell at the same time.
Accessibility remediation. We ran a complete ADA and WCAG 2.1/2.2 Level AA remediation program on the live storefront: audited systematically, fixed at the source rather than patched page by page, and re-verified after every phase. The program ran to completion, and the fixes are structural, so they hold as the site continues to change.
Search recovery. Working from the search engines' own reporting rather than assumptions, we identified what was starving the site's crawl and corrected it, validating each change before deployment and measuring each result in its own before-and-after window. The indexing decline reversed within weeks, with the largest categories of wasted crawl shrinking by six figures.
Product content at scale. Most of the catalog's pages carried empty or duplicated search listings, something the search engines had flagged themselves. We built an AI content pipeline with hard quality controls: every description is generated from the client's own catalog data, has to pass automated fact-checking before it can publish, and can be withdrawn instantly. Roughly 794,000 unique product descriptions are now live, covering about 99% of searchable products, with hand-written copy always taking precedence.
Pipeline modernization. We audited the client's catalog data pipeline, over 200 data-processing packages, and surfaced failures that had been passing silently for months and in some cases years. The critical paths are now monitored, and the client holds a measured, phased modernization roadmap to work from instead of guesswork.
We measure before we act. We rehearse significant changes against copies of production. Every change ships with a tested way back. Revenue paths get protected before anything gets blocked. And nothing is called fixed until the numbers show it.
That is how compsource.com, a twenty-year-old platform carrying over a million products, got a new lease on life, without the cost, delay, and risk of a rewrite.
Running a legacy platform that is showing its age? Talk to us before you pay for a rewrite.
