Social advertising, used like an operator

If revenue is the goal, run conversions with a properly fired purchase event. Engagement objectives optimize for cheap engagement — and almost never for sales.

We've seen accounts spend $400k on 'engagement' objectives optimizing perfectly for cheap likes — and producing almost no revenue. The platforms do exactly what you tell them to. If the goal is purchases, run a purchase-optimized campaign with a properly fired conversion event, even if the early CPMs look more expensive. The expensive impressions are the ones that actually buy.

Three to five new concepts a week, two angles per concept, two formats per angle. Most accounts fail because creative volume drops the moment a winner appears.

The accounts that compound treat creative testing like a software release schedule: 3-5 new concepts shipped every Monday, no exceptions. Most accounts fail because they ship a winner, stop testing, and ride that winner until it fatigues — then panic. Rhythm beats heroics in paid social.

Test one element at a time — hook, offer, format, or audience. Bundled tests give bundled answers, which means no answers.

We audited an account that had been running 'tests' for six months with no clear winners. Looking at the test setup, every variant changed the hook, the offer, the format and the audience simultaneously. There was no way to attribute the lifts. One variable per test, two weeks per test — boring but conclusive.

Retargeting flatters the dashboard but rarely creates new demand. Hold the prospecting budget steady even when retargeting ROAS looks better.

Retargeting ROAS almost always looks better than prospecting ROAS, which tempts teams to shift budget toward retargeting. The trap: retargeting is harvesting demand that prospecting created. Cut prospecting and the retargeting pool dries up in 30 days. Keep them on separate budget lines with separate targets, judged independently.

Use platform reporting for in-flight optimization, but make spend decisions against post-purchase data, blended CAC and contribution margin.

Platform dashboards live in walled gardens that overstate their own contribution. Decision-grade reporting lives in the data warehouse where post-purchase, payback period and contribution margin live. Use the platform for in-flight tactical decisions; use the warehouse for budget allocation. Confusing the two is how brands end up spending big on the channel with the prettiest dashboard rather than the highest actual return.

Set a frequency cap or CPM threshold that triggers a refresh. Removes the guesswork and keeps the account out of the death spiral.

Set explicit triggers: frequency above 3.5, or CPM up 30% from baseline, or CTR down 20% week-over-week. When the trigger fires, refresh — no debate. Without explicit triggers, refreshes happen either too late (after the account spirals) or too early (killing winners that still had two more weeks of life).