What Separates a Great Ad Campaign from an Average One?

Does Every Great Campaign Start with a Clear Hypothesis?

Yes. The panel agrees that the single biggest differentiator between great campaigns and average ones is whether the campaign has a clear hypothesis about who it’s reaching and what action it expects them to take. Paxton Gray (CEO, 97th Floor) sees this across dozens of agency clients: campaigns that start with “let’s run some ads and see what happens” always underperform campaigns that start with “we believe this audience will respond to this message because of this insight.” Derek Cohen (Head of Acquisition, Sunrun) adds that the hypothesis needs to be falsifiable — if you can’t tell whether your hypothesis was right or wrong after the campaign runs, you haven’t learned anything, and learning is the point.

How Many Creative Variations Should You Test?

Test at least 3-5 variations per ad set, and test across meaningful dimensions — hook, format, value proposition, and audience match — not just cosmetic differences. Rob Lamb (VP Marketing, Blueshift Cyber) makes a distinction between testing and randomness: swapping a blue background for a green background isn’t a test, it’s noise. A real test changes one variable that represents a genuine hypothesis about what the audience cares about. Paxton adds that the best agencies he’s seen test creative on four dimensions: the hook (first 3 seconds or first line), the format (video vs. static vs. carousel), the value proposition (what benefit is promised), and the audience match (same creative to different segments). Each dimension teaches you something different.

How Should You Structure an Ad Account to Isolate What’s Working?

Each campaign should have a single objective, a clearly defined audience, and enough budget to reach statistical significance within two weeks. Derek’s in-house perspective at Sunrun is that most underperforming accounts have a structural problem, not a creative problem. Campaigns with overlapping audiences cannibalize each other. Ad sets with too many variables make it impossible to attribute results. The fix is organizational discipline: one campaign per objective, non-overlapping audiences, and enough daily budget that each ad set gets at least 50 impressions per day. If an ad set can’t get 50 impressions, it shouldn’t exist.

How Do You Decide Which Campaigns to Fix First?

Rank campaigns by the ratio of spend to learning — campaigns that are spending the most while teaching you the least should be fixed first. The panel converges on a prioritization framework: a campaign spending $500/day with no clear signal about what’s working is more urgent than a campaign spending $50/day that’s also struggling. The high-spend, low-learning campaigns are actively wasting money. The fix for each campaign follows the three-axis grading: check strategy first (is the hypothesis clear?), then creative (is the test designed to teach?), then structure (can you isolate the signal?). The axis with the lowest score is where to start.

What Did the Live Campaign Build Contest Reveal?

The participants who built winning campaigns in Leo during the session shared a common trait: they started with audience insight, not creative execution. The session closed with a contest where participants built Meta campaigns in Leo and the panel picked a winner. The campaigns that impressed the panel weren’t the flashiest — they were the most strategically grounded. The winners had a specific hypothesis about their audience, chose creative that tested that hypothesis, and structured the campaign so they’d know whether they were right within a week. The panel’s feedback reinforced the session’s core theme: strategy beats execution every time, and good execution without strategy is just organized waste.