What seven figures of paid social taught us about publisher growth

MMG manages seven figures of annual paid-social spend for publisher clients: traffic acquisition at a scale where mistakes are expensive and lessons arrive quickly. Most of what we learned contradicts how marketing teams are taught to buy ads, because publisher economics are not e-commerce economics. These are the rules we now run everything by.

Optimize to revenue, not to engagement

Click-through rate and CPM are diagnostics, not objectives. A publisher buying traffic has exactly one question: what does a session cost, and what does a session earn? We optimize to cost per pageview tied to the revenue each pageview generates, a closed loop from spend to session to dollars. Campaigns that look mediocre on engagement metrics regularly win on that loop, and campaigns with beautiful CTRs regularly lose money. If the optimization target isn't denominated in revenue, the algorithm is being trained to buy the wrong thing.

Budget steers. Caps are ceilings.

The most common error we see is treating bid caps as a steering wheel and jerking them around to chase delivery. A cap is a ceiling on what you'll pay. Budget allocation is the actual steering wheel: winners get more in measured steps, losers get cut, and caps move slowly — think a cent at a time. Teams that invert this spend their weeks fighting the auction instead of compounding what works.

You are your own worst competitor

Run two campaigns promoting the same content to the same geography and you are bidding against yourself, paying a premium to show your own ad over your own ad. Overlap is invisible on a dashboard and expensive in an auction. The fix is structural: expand across geographies and content before stacking spend on the same audience twice, and check for the collision before scaling.

Creative decays on a schedule

Every ad has a half-life. Frequency climbs, novelty fades, and unit costs drift up on a clock you can roughly predict. The discipline is refreshing creative on cadence, before decay shows up in the numbers, instead of scrambling after performance collapses.

Auctions have weather

During prime shopping events, elections, and seasonal surges, auction prices inflate and delivery slows for reasons that have nothing to do with your campaign. Slow spend during a hot auction is often weather, not failure. The costly response is panic: raising caps into an inflated market locks in bad unit economics that persist after the surge passes. Patience is frequently the highest-ROI decision available.

The same discipline transfers directly to newsletter subscriber acquisition, where we now apply it for the properties we own: CPA by channel, lifetime value against a 3:1 floor, and scale thresholds that decide when a channel earns more budget. Paid growth is unforgiving arithmetic. The teams that win are the ones that refuse to look away from it.