Magic City Telecom — Audience Analytics

Magic City Telecom was spending across a wide marketing footprint without a clear read on which audiences and regions paid back the spend. Four specialist teams — Audience, Spatial, Creative, and Inventory analysts — were assembled to turn raw campaign data into one coordinated recommendation: cut what doesn't convert, double down on what does.

On the Audience Analyst team, I analyzed audience data to align campaign goals with real behavioral signals, ran A/B tests to optimize ad spend and bring down cost-per-acquisition, and translated the numbers into recommendations: which segments to scale, and which placements to cut. I built the personas the creative team used to target messaging.

• Five states carried the campaign — Illinois, California, New York, Florida, and Tennessee. Reallocating budget toward them and away from underperformers pointed to roughly $500,000 in potential savings.
• Mobile-friendly creative won. Prioritizing mobile-first ad formats lifted conversion rates by an average of 8.54% — younger segments clicked more, at a lower cost per click.
• A/B testing cut wasted spend. Testing ad variants let us identify low-CPA activities to scale and high-CPA ones to fix — lowering the average cost per acquisition.
• Placement quality mattered as much as audience. Reviewing exchanges, devices, creatives, and viewability across top- and bottom-performing segments surfaced where impressions were being wasted.

The team's proposed allocation — weighted toward the channels the data said actually moved conversions.

Capstone project, COOP Careers Data Analytics Fellowship · figures are the team's modeled estimates from campaign data.