Rebuilt a rented, unmeasurable web presence into an owned funnel with attribution on every channel, then built the content engine that feeds it.
The business ran on a rented platform. The site was a brochure, the lead database lived inside a vendor's CRM, and no channel carried a cost per lead. Budget decisions were made on impressions and instinct.
Roughly half of recorded traffic was bot noise, which meant even the numbers people did trust were wrong.
266 paid leads at a $20.36 blended cost per lead. Organic search grew 14x in the month after the rebuild. A placement reallocation returned 69% more leads on identical budget.
The placement with the best click-through rate had the worst cost per lead, and the one with the worst click-through rate was nearly the cheapest. Optimising to clicks would have moved budget in exactly the wrong direction. Cost per lead is the only target that survives contact with reality.
Here is the whole account, in four steps.
244,418 impressions. 5,026 clicks. 266 leads at a $20.36 blended cost. Most reporting stops right about here, and that is the problem.
This is the view a volume dashboard gives you. Facebook Reels lands top three on 36 leads, so it earns more budget. Hold that thought.
The order almost inverts. Facebook Reels falls to seventh. Facebook Search climbs from seventh to third. Same account, same money, opposite answer.
Facebook Feed had the best click-through rate at 4.39% and the worst lead in the account at $36.38. Moving spend off it and onto what was already producing did the rest. No new creative, no new audiences, no extra spend.
Roughly two thirds of those leads were renters rather than buyers, so true cost per qualified buyer ran three to four times the headline figure. The blended CPL is an honest cost per registration and I quote it as one. Shifting that mix is a creative and targeting problem, not a form-fields problem.