Signal engineering: move conversion signals toward revenue

Marcus Burke, Meta Ads consultant, explains why blended CPA is misleading, how creative format determines your audience targeting, and what signal engineering means for subscription apps in an AI-driven ad landscape.

Marcus breaks down his approach to working with Meta's algorithm rather than against it. He advocates for strategic ad set segmentation based on where different creative formats naturally deliver: static ads to Facebook feed, short-form video to Instagram Reels.

The conversation goes deep on the relationship between product design and ad optimization. Marcus explains how your subscription model, trial length, and paywall structure all affect the quality of signal you can send back to Meta. Sometimes optimizing for LTV conflicts with optimizing for ad signal, and growth teams need to navigate that tension intentionally.

What you'll learn:

• Why a $10 cost per trial can lose money while a $100 cost per trial can be profitable
• How to use creative format (static vs. video vs. playable) to control placement distribution
• Why "broad targeting" often results in narrow reach and high frequency on the same audience
• How to structure ad sets by expected delivery rather than demographic targeting
• What value rules are and how to use them for country, age, and gender optimization
• Why the conversion event you optimize for should determine your account architecture
• How to connect onboarding survey data with Meta demographic breakdowns
• Why cold social traffic requires a fundamentally different onboarding approach than search traffic
• What makes an effective "aha moment" before the paywall
• How multi-price point strategies enable broader audience targeting
• Why signal engineering is one of the last remaining levers for growth marketers

Key Takeaways:

• Blended CPA hides traffic quality problems. A $10 cost per trial from Instagram Reels represents a completely different audience than $10 from Facebook feed. Break down your metrics by placement to understand what you're actually buying.

• Creative equals targeting. Your media format determines where your ad delivers. Short-form vertical video goes to Reels; statics go to Facebook feed. This isn't a bug but a feature you can use to control your audience mix without hard targeting.

• Guide the algorithm, don't force it. Hard targeting gets expensive fast. Instead, use creative segmentation and value rules to nudge Meta toward your high-value audiences while keeping delivery efficient.

• Your conversion event determines your account structure. If you're optimizing for a shallow event like trial starts, you need more ad sets to compensate for the algorithm's lack of business knowledge. Moving closer to revenue lets you consolidate more.

• Onboarding should match your traffic source. Paid social users were just doom-scrolling and need to be entertained and re-sold on their problem. Search traffic already has intent. Design your onboarding accordingly.

• Create an aha moment before the paywall. Prove value in the first session through something tangible: a sample scan, a personalized analysis, an imported recipe. This converts better than promising value during a 7-day trial.

• Your pricing should match your creative strategy. Young audiences from UGC won't pay $70/year. Older Facebook feed audiences justify higher CPMs. Align your price points with who your ads are actually reaching.

Links & Resources

• Marcus Burke on LinkedIn: https://linkedin.com/in/marcusburke
• Growth Festival Presentation: https://www.linkedin.com/posts/marcusburke_postmedia-buying-strategies-scaling-meta-activity-7373668067580563456-iziy

Timestamps

00:00 – Intro clips
01:41 – Introduction and context on Marcus's Growth Festival presentation
02:12 – Why blended CPA is irrelevant and how it differs from blended ROAS
04:36 – How placement affects traffic quality: $100 vs $10 cost per trial
05:11 – Using creative format to control placement distribution
07:32 – Working with the algorithm vs. forcing targeting
10:19 – Why "broad targeting" doesn't mean broad reach
11:06 – Getting placement and demographic data from Meta
14:47 – Layering complexity: placements, demographics, user goals
20:09 – Signal engineering and moving closer to business value
22:30 – Account architecture: stop over-consolidating
26:49 – Should subscription apps test removing trials?
31:52 – Value rules: what they are and how to use them
36:30 – Onboarding for paid social: entertainment over efficiency
39:36 – Creating aha moments before the paywall
45:44 – Multi-price point strategies to capture the full demand curve
48:31 – Wrap-up and where to follow Marcus