Tyson Steel Building is a Doerun GA industrial steel building manufacturer running Google Ads on a $750/mo budget. The brief was straightforward: "we are spending the budget every month and we cannot tell whether it is working." The audit told us why.
The headline finding: Tyson Steel's Google Ads conversion tracking was misconfigured in a way that made the account look like it was performing twice as well as it actually was. The primary conversion action was set to a generic page-load event that fired on every visit, not on actual form submissions. Once we fixed the tracking, the real performance picture emerged: ~10 qualified leads per month at a real cost-per-acquisition that the budget could sustain, but with significant remaining optimization runway on landing pages and bidding strategy. This case study is about what an honest audit finds when no one has looked under the hood in 18 months.
The primary conversion action was tracking a generic "thank you page visited" event that fired on every page load of the thank-you template — including when users hit the page directly from a bookmark or refresh. Real form submissions counted, but so did dozens of fake conversions per week. The reported CPA looked like $35; the real CPA was $75.
We hit a technical wall trying to fix the conversion action programmatically — Google's API mutate request returns success but the value does not persist. Confirmed April 13, 2026 on this exact account. The fix required logging into the Google Ads UI and toggling Primary vs Secondary status manually. This is documented in our Google Ads API quirks list and informed every subsequent client engagement.
Because the account was using Maximize Conversions bidding tied to the broken conversion action, the algorithm was learning to find traffic that landed on the thank-you page for any reason — not traffic that actually filled out the lead form. Months of "smart bidding optimization" had been pointed at the wrong target. Once we fixed the conversion action, we restarted the bidding strategy's learning period and the real qualified-lead flow stabilized over 4-6 weeks.
Industrial steel building inquiries do not happen at 3 AM. Tightened to business hours plus a 2-hour evening buffer.
Google recommends 2+ RSAs per ad group for asset rotation and creative testing. Tyson Steel had one RSA, getting no comparative data on what messaging actually resonated. Added a second RSA with distinct positioning in each ad group.
Search terms report showed budget spent on "how to build a steel building yourself," "shipping container homes," and various job-seeker queries. Built a 28-keyword negative list filtering DIY, off-vertical, and recruitment queries.
This is the part of case studies most agencies do not write. Here is what changed and what did not.
What changed measurably:
What did not change yet (in the optimization queue):
Three lessons from the Tyson Steel engagement that we now apply to every new paid media client:
Send us your Google Ads MCC link with read-only access. We respond within one business day with an honest assessment.
