kingtommy dot com

Campaign tools

Offers Overview: how many creatives each offer has, and how much weight they carry. Too few creatives is a production problem; plenty of creatives with little weight is a performance problem. Optimizations: the scheduled job's changes, creative by creative, with each one's earnings per thousand impressions over the last two days and its old and new weight. The Optimization History heatmap: one row per creative (O is its offer, C the creative) from April to June. When one row shifts, it's that creative; when many shift at once, the cause is bigger. Poor Offers Overview: rating, weight, impressions, clicks, revenue, conversions, CTR and eCPM for each underperforming creative, lowest eCPM first. Verticals Overview: creatives and offers for each vertical in each offer pool, to show where inventory is thin. The Campaigns editor: every offer and its creatives, each with a preview of the notification and its weight.

For PushPros · These tools were for PushPros staff only, so the screens show made-up data.

A collection of campaign management tools for the PushPros ad content network. Showed why an offer or vertical wasn't earning, and moved traffic toward the higher performing creatives.

PushNetwork was a predictive ad delivery service that served advertisers' offers across a network of publisher properties as web ads, email, and push notifications, peaking at over half a billion impressions daily.

Some offers never gained traction, and the cause wasn't obvious. An offer only reaches users once the production team builds ads for it, to spec for each property, and adds them to campaigns. Offers Overview diagnosed a quiet offer at a glance: a performance problem, with impressions but no clicks or conversions, or a production problem, with too few ads built for it. It recovered revenue for advertisers that had fallen through the cracks. Verticals Overview applied the same analysis to entire verticals, to identify where inventory was thin.

Weighting drove performance. Every creative carried a weight that set its share of traffic, and a scheduled optimization job shifted weight toward the creatives with the highest eCPM and away from the rest. As a result, an offer could have plenty of creatives and still reach only a sliver of the network. The Optimizations screen shows each change, creative by creative.

The Optimization History heatmap shows when performance shifted: sometimes a single creative, sometimes an entire advertiser, and sometimes the audience itself, which was the hardest to isolate. The heatmap helped the team identify which.

I led PushNetwork's architecture and design.