META Ads: A Practical Guide

Campaign Structure & Budget Consolidation

Use a single ad set unless you're generating at least 50 conversions per week. Meta's optimization algorithm needs sufficient conversion volume to work effectively—spreading budget across multiple ad sets or campaigns without hitting this threshold starves the algorithm of data and can cause outright campaign failure (via @benheathmarketing).
The same 50 conversions/week threshold applies to multiple campaigns—consolidate until you hit the minimum before fragmenting further.
Budget fragmentation across too many campaigns/ad sets—not bidding or targeting—is the real bottleneck in most underperforming accounts. Consolidate budget to give Meta's optimization engine sufficient signal (via @nathan.perdriau).
Don't duplicate winning ad sets expecting to replicate results—duplication splits signal and reduces overall performance instead.
Treat Advantage+ Shopping Campaigns as a delivery setting, not a complete strategy that requires additional strategic layering on top.
Use broad targeting with strong creative rather than narrow audiences—this is now honest strategy, not lazy execution, since creative quality does the targeting work.
If frequency is high on cold audiences, the problem is usually insufficient concept variety, not audience size—expand creative concepts rather than shrinking/expanding targeting.

Switch from Meta's default 1-day view attribution to 7-day click attribution. The 1-day view default incorrectly credits non-clickers, causing the algorithm to optimize for people who'd buy anyway rather than acquiring new customers.
Exclude view-through (1-day view) data entirely from reporting and optimization decisions—it inflates ROAS with dishonest numbers.
Expect reported ROAS to drop when you make this switch—that's a sign the fix is working, not a red flag. The algorithm will now optimize for users who actually clicked and purchased.
Don't kill ads based on last-click ROAS alone—this collapses campaign optimization. Use the broader 7-day click attribution window before making cut decisions.

Most brands only run 1-2 ad types, leaving major growth untapped. Run all 5 types to cover the full customer journey (via @adswithsimon):

Explain (product education): UGC how-tos, founder explainers, or text-heavy statics for audiences with zero prior product knowledge.
Prove It Works (social proof): Target mid/bottom funnel with one specific, verified claim instead of 5 vague testimonials—use UGC reviews, before-afters, statistics, or review testimonials.
Push the Offer (conversion): Lead with price/bundle/headline for bottom-funnel audiences with existing intent; offer statics, listicle breakdowns, or any creative with an offer overlay.
Break the Pattern (attention/anti-fatigue): Disruptive content—memes, text-heavy ads, "ugly" ads, yapping, podcast clips—that stands out in feed. Works on cold audiences for attention and on warm audiences to prevent creative fatigue. Bold doesn't require loud; blank space or raw photography can work.
Build the Brand (recall): Founder yapping, behind-the-scenes content, or brand films that build recall rather than drive immediate conversions.

Every concept has three core components, in this order: persona (who), angle (the unique perspective/problem), and offer (what you're giving them). Everything else—format, font, background—is decoration.
Lock in persona, angle, and offer before opening any design tool. Format (before/after, listicle, native static, advertorial) is just the delivery mechanism—it should follow naturally once the concept is set, not lead it.
A weak concept cannot be saved by a great hook. Since creative now does the targeting work, concept strength directly determines targeting success.
Stop treating minor edits (font changes, background colors, frame variations) as real tests—they dilute signal and cannibalize each other's performance. Test fundamentally different angles and personas instead; the delta between two versions of the same idea produces noise, not insight.
Don't create 20 variations of one ad concept and call it testing—treat that as one ad printed multiple times, which fragments signal rather than improving performance.

Roughly 20% of your concepts will generate 80% of your results. Prioritize finding and doubling down on those winners rather than distributing budget equally across all concepts.
Optimize for average winner spend capacity, not hit rate. A 30% hit rate is meaningless if winning ads plateau at $100/day—that's a shallow ceiling on scalability.
High hit rate + shallow winners = you're good at execution but lack strong core concepts. Deep winners (ads that sustain much higher spend) prove real concept strength.
Test whether your best performers can handle 2x, 3x, or more budget before performance breaks—this reveals true concept quality, not the percentage of ads that "work."
Weight budget allocation by each winner's spend capacity, not by equal distribution across all winners.

The most common static ad mistake is sitting in the middle of the spectrum between maximum clarity and deliberate curiosity gap—resulting in underperforming, forgettable creative.
Maximum clarity: viewer knows the product, angle, offer, and audience within 3-4 seconds.
Deliberate curiosity gap: native/advertorial/listicle style where the viewer doesn't realize it's an ad, pulling clicks through curiosity.
Stop trying to blend both approaches. Hybrid ads—vague enough to feel native but clear enough to feel like ads—underperform as neither extreme.
Commit fully to one direction rather than building middle-ground creative that lacks the strength of either pole. This principle applies beyond statics to any creative with competing variables.