The Master Brief: My Complete Approach to Meta Ads

This post isn’t like anything I’ve published before, so let me explain what it is and how it was created.

The Master Brief is a consolidated overview of my current recommended approach to Meta ads. It covers targeting, value rules, campaign structure, attribution, creative, tracking, and more. It’s the closest thing to a single document that explains how I think advertisers should approach this platform right now.

But here’s the part that requires transparency: I didn’t write it.

I created a tool (with some help from Claude Code) that turns my blog posts, short-form video posts, and podcast transcripts into ad briefs. I review, edit, and approve each brief, and they’ve become a core resource inside of my PHC – Elite membership community.

The Master Brief merges all 24 briefs into one document, removes the overlap between them, and organizes my complete approach by theme. Instead of reading two dozen briefs to piece together the full picture of my recommendations, you can read this one resource. Like the briefs, it’s written in the third person.

So while I didn’t type the words below, everything in this document reflects my positions, my recommendations, and my 15 years of experience with this platform. I reviewed and approved every section before publishing. If something in here is wrong, you can blame me.

Why give it away? Because I want you to see what this resource is.

The version below is a snapshot. Meta evolves, and my recommendations will evolve with it. When that happens, the member version of the Master Brief is updated to keep it current. This copy won’t change. Keep that in mind if you’re reading it months from now.

Here it is.

Prioritize Algorithmic Delivery and Embrace the Loss of Targeting Control

Jon’s foundational belief is that Meta’s algorithm is literal: it does exactly what you ask. Optimize for a purchase and Meta shows your ads to people most likely to buy; optimize for cheap clicks and you’ll get cheap, low-quality clicks.

Understanding this eliminates most of the perceived need for manual control.

Targeting has changed more than any other aspect of the platform. Advertisers still operating like it’s 2018 or 2021 are working against Meta and hurting their own results. Meta now has enormous historical and real-time data that far surpasses any manual input.

The key change: most targeting inputs (age, gender, detailed targeting, lookalikes, and custom audiences) are now only suggestions by default with Advantage+ Audience on. For conversion and lead goals, detailed targeting and lookalikes can’t be restricted at all. Jon has found no evidence that suggestions meaningfully affect delivery, so he doesn’t bother with them.

The biggest mistake is envisioning an ideal customer and restricting by age or gender to match. This assumes Meta will waste budget on people who won’t convert. That gets it backwards.

When optimizing for a purchase, Meta wants results too, so it won’t spend meaningfully on groups that don’t convert. Restricting only limits the algorithm and drives up costs.

Detailed targeting and lookalike audiences, once central to strategy, are largely obsolete. Building separate ad sets to test different interests or lookalikes usually just reaches the same people. Any difference in results is typically randomness, a classic correlation-versus-causation trap.

Jon, once known as “the targeting guy,” concedes suggestions might hypothetically help brand-new accounts where Meta lacks data, but treats them as unimportant.

Not everything is a suggestion. Audience Controls are hard limits Meta enforces even with Advantage+ on:

• Location. Imperfect rather than ignored: the only option is “living in or recently in,” and city/region targeting pulls in nearby areas via radius and expansion, so country-level targeting is most reliable.
• An age minimum (up to 25).
• Custom audience exclusions. Not ignored either. When someone slips through, it’s a completeness problem with the underlying audience rather than Meta defying the setting.

Jon’s hands-off approach: he doesn’t use detailed targeting, lookalikes, or age/gender restrictions, and he doesn’t bother with suggestions. For location, he targets the countries he wants and combines similarly priced countries into one ad set unless splitting drives meaningful volume.

He restricts only for legal requirements (age-restricted goods) or a proven performance problem where he has information Meta doesn’t.

Use Value Rules Instead of Restrictions

When you have information Meta doesn’t, like demographic-specific lead quality or customer lifetime value, Jon prefers a value rule over a restriction.

Value rules let you bid more or less based on criteria like age, gender, location, placement, mobile OS, conversion location, and now custom audiences. He calls them the future of delivery control.

The logic is simple: Meta optimizes for the cheapest version of the action you request, but cheap results aren’t always good results. Value rules let you guide budget distribution without disabling Advantage+ or restricting your audience. That avoids the harm caused by hard restrictions.

Jon’s go-to example comes from lead generation. He repeatedly found Meta concentrating budget on people 65+, producing cheap, low-quality leads.

His old fix was restricting by age, but that simply moved budget to 55–64 and eliminated valuable customers. Instead, he now bids 90% less on 65+ and 20% less on 55–64, dropping spend on 65+ from 45% to under 2% while keeping everyone in the mix. The same logic applies to gender, placements, and other criteria.

A newer, more robust version lets you bid based on custom audiences by labeling them: qualified leads, disqualified leads, high-value customers, cart abandoners, at-risk, disengaged, and more. You can then bid up on high-value groups and down on low-value ones. Jon sees promise here but warns it requires significant effort to label audiences properly.

Don’t use value rules to micromanage delivery. Meta openly encourages their use, adds alerts, and keeps expanding the feature, a signal that they aren’t going away.

Simplify Campaign Structure and Consolidate Budget

Jon recommends the fewest possible campaigns, ad sets, and setting customizations. The most efficient starting point is one campaign and one ad set, optimized for a conversion, with budget consolidated into a single business goal.

Every extra campaign and ad set waters down your budget. In a vacuum, $100 in one ad set is more efficient than $100 split across five. Splitting also invites auction overlap and audience fragmentation, making it harder for any ad set to reach the roughly 50 optimized events per week needed to exit the learning phase.

Advertisers overcomplicate because they misunderstand how Meta works. They think they need to control targeting, segment audiences, and build separate campaigns for testing, remarketing, or top-of-funnel. Much of this solves imagined problems while creating new ones.

Emotional attachment to specific ads often drives this, leading advertisers to create separate ad sets to force delivery of an ad they don’t want to risk.

On budget: there’s no universal “right” number. It depends on your goal and an honest expected cost per optimized action. If you sell a $100 product, conservatively assume it costs $100 to get a sale.

For beginners, $10–$20/day makes little impact but is still worthwhile for gaining experience. Spend what you’re comfortable with, then scale as results come in.

Simplicity is the starting point, not an absolute rule. Legitimate reasons to add complexity include:

• Separate business goals (a Leads campaign and a Sales campaign)
• Distinct products, product lines, or physical locations with different messaging
• Seasonal products

But only when you have the budget to drive meaningful, efficient results in each campaign or ad set. Jon specifically warns against splitting for creative testing or audience segmentation, which are outdated approaches.

On budget controls: Jon generally resists ad set spending limits, preferring to let Meta push budget where it works, and applies a minimum or maximum only if optimization struggles. He also cautions that setting bid controls (Cost Per Result Goal, Bid Cap, ROAS Goal) too aggressively can starve delivery, since you’ve removed Meta’s requirement to spend the full budget.

When results struggle, his advice is to simplify first, then focus on the ads instead of campaign structure.

Every Change Must Solve a Proven, Data-Backed Problem

Before adding any complexity, restriction, or customization, Jon wants one question ringing in your head:

Every separate campaign, extra ad set, demographic restriction, or disabled enhancement should exist to solve a proven problem rather than a feeling, vibe, assumption, or confirmation bias.

He recommends auditing your current campaigns. Identify your single most important one, then question every ad set, targeting restriction, placement removal, and disabled enhancement:

• What problem was each meant to solve?
• Did it solve it?
• Did a better solution exist?

Push back on your own reasoning and don’t let confirmation bias off the hook easily.

Jon sees advertisers fixate on things that don’t matter: placements, detailed targeting, age and gender, micromanaging which ad gets budget. Many decisions rest on “people under 30 don’t buy” without data. That’s a decision based on vibes rather than evidence.

Humility is an overlooked but essential trait: when results tank, the answer usually starts with you.

He also cautions against acting on meaningless data. Small sample sizes produce random results that advertisers wrongly attribute to targeting or creative. Correlation isn’t causation, and there is no magical campaign construction or guru hack that fixes fundamentally weak ads.

Above all, Jon rejects universal rules. His honest answer is often “it depends,” because context (industry, price, budget, goal) changes everything. He wants advertisers to understand how things work rather than blindly follow his advice, so they can spot their own exceptions and let results be their guide.

Attribution, Breakdowns, and Conversion Reporting Literacy

The number in the Results column is convoluted: its meaning depends on attribution settings, conversion count, and data quality. What separates good advertisers from bad is what they do with that number. Taking it at face value leads to either overconfidence or misplaced frustration with Meta.

Different conversions carry different weight:

• Click-through (especially 1-day) reflects the strongest intent.
• Engage-through shows interest but no website click.
• View-through is the flimsiest: no click of any kind verifies influence. A high share of view-through conversions makes your results mean much less.

Jon rejects the common complaint that Meta “steals credit” or reports “vanity metrics.” Those impressions and clicks genuinely happened before the conversion. The same arguments apply to Google and other channels.

When results look inflated, a reasonable explanation is far more likely than Meta making up numbers: an event setup problem, a deduplication issue, or one person converting multiple times.

Two reporting tools are essential. Breakdown by Attribution Settings segments reported results into clean rows (1-day click, 2–7 day click, 1-day engagement, 1-day view) with no math, but only under the Standard model. Compare Attribution Settings goes further, uncovering conversions that happened under settings you didn’t select, which is a regular part of Jon’s routine.

He also distinguishes First Conversion (isolating unique customers) from the default All Conversions, which can inflate the picture.

Attribution shapes delivery, not just reporting. The setting tells Meta what kind of conversion to optimize for and who to find. Jon recommends the default 7-day click-through for purchases, giving Meta room to find people on a slightly longer consideration cycle.

For free, low-deliberation offers, or when he wants clean data without view-through inflation, he prefers 1-day click. He remains skeptical of optimizing for incremental attribution, citing worse costs and no clear quality gain in his tests.

Breakdowns (by age, gender, placement, platform, location, audience segments, attribution, and creative) help you understand results. But the outdated approach of trimming “underperforming” ages or placements to force efficiency just restricts the algorithm and usually hurts results.

Instead, use breakdowns to detect concentrations of cheap, low-quality results tied to your performance goal, then act with a value rule. For long buying cycles, Jon keeps click windows and creates varied ads for different awareness levels rather than switching objectives.

Creative Diversification Is the Primary Lever

Since targeting moved into the ads, your creative now does the work audience targeting once did. Ad creation today is fundamentally about creative diversification: providing meaningfully different visuals, formats, angles, and text so Meta’s delivery system can match the right message to the right person.

This is enabled by Andromeda, Meta’s ad retrieval engine, the first step of delivery that narrows tens of millions of ads to a few thousand candidates. Jon stresses that Andromeda is just retrieval:

• It’s not an algorithm, and it doesn’t rank winners (GEM does).
• It doesn’t change your targeting strategy, and it didn’t kill your results.
• It “doesn’t want” anything. Focus on what people want rather than on pleasing the system.

It was built to handle today’s explosion of creative variations, which is precisely why diversification matters.

Jon tried pumping out 20, 30, or 50 ads at once and found it wasteful: Meta concentrates budget on a small handful regardless. The goal is genuine diversity across formats, concepts, personas, and pain points rather than a pile of near-duplicates.

He also warns against obsessing over which single image, video, or text “wins,” since successful ads have many winning combinations working together in aggregate.

Jon builds diversity in phases, which he calls a “creative diversity stack.” Each phase shares a theme; combined across phases, they produce genuine diversity. He starts small, publishes, and lets results guide the next phase.

His tactics include:

• Customizing creative by placement (9×16 video for Reels and Stories, static or 4×5 elsewhere)
• Using all five primary text and headline options at varying lengths
• Staying open to Advantage+ Creative enhancements

Meta’s creative testing tool is central to this workflow. It runs a test of two to five ads (possibly up to 10) within your active ad set, spending budget evenly with no auction overlap; when the test ends, Meta resumes normal delivery.

This is far superior to the old approach of building separate test campaigns, since an ad that wins in isolation isn’t guaranteed to perform once moved. Jon sometimes starts new ad sets with a test, manually assigning 100% of budget, though he does this less often now that Push Delivery can answer the same question for existing ads.

He warns against underpowered tests: five ads splitting $50/day won’t generate meaningful purchase data.

When interpreting results, don’t hunt for a single champion. Identify general themes that worked and apply them to your next batch, keeping text variations similar in style, persona, or angle so results aren’t watered down. You don’t need separate ad sets per persona; place persona-focused ads in the same ad set and let Meta match them.

Meta’s new creative workflow (currently in test) allows up to 10 assets per ad with per-creative performance breakdowns, a level of transparency advertisers have long lacked, and appears set to replace both Flexible Format and Dynamic Creative.

Remarketing Happens Automatically, and Audience Segments Prove It

Remarketing (reaching website visitors, email subscribers, customers, and engaged users) defined Jon’s approach for a decade, making up 90–95% of his strategy during his peak years.

Times changed. Reach costs multiplied, audiences shrank after iOS 14, and Meta invested heavily in algorithmic targeting. Most importantly, Meta now prioritizes remarketing automatically, defaulting to past conversions, pixel activity, and prior ad engagement when Advantage+ Audience is on.

You can prove this with Audience Segments, a reporting-only feature (Sales objective required) that breaks results into Engaged Audience, Existing Customers, and New Audience. Breaking down a Sales campaign this way typically reveals that Meta spends roughly 20–25% of budget on remarketing without any inputs (sometimes 10–40%, higher for new campaigns and bottom-of-funnel goals).

Jon credits this tool with transforming his understanding of algorithmic targeting.

Define your segments properly:

• Engaged Audience: define it broadly. All website visitors (180 days) plus your entire email list, capturing everyone who has engaged.
• Existing Customers: define it precisely. A website custom audience of all purchase events (now up to 730 days) plus an email audience of all paying customers.

Don’t worry about overlap; Meta counts shared users as Existing Customers. Failing to define these segments, or defining them incompletely, is one of the most common mistakes and a red flag in account audits.

The implication is decisive: if Meta already reaches these people, a separate remarketing campaign just reaches them twice, creating auction overlap. Jon almost never restricts by custom audiences anymore.

Remarketing’s famously low costs and high ROAS are often propped up by view-through conversions, many of which would have happened anyway, and recent attribution changes may hit remarketing results hardest.

Before excluding existing customers, confirm a proven problem exists (reaching them too often, lower average order value, or they’d have bought anyway). Then use custom audience exclusions, the Customer Lifecycle Strategy option, or a value rule to bid less, rather than removing them outright.

Note that Customer Lifecycle Strategy is simply a second place to manually select audiences to exclude; it isn’t automated magic and only works because you’ve defined your segments.

Jon isn’t 100% anti-remarketing. The rare valid exception is a tiered model: advertise a low-ticket offer, then support the upgrade to a high-ticket product with a dedicated campaign that uses:

• The Awareness objective with Maximize Reach
• Only your custom audiences, with Advantage+ Audience off
• A frequency cap
• Messaging that mimics your sales approach

Don’t Blame Meta. Take Responsibility and Evaluate in Aggregate

When results are poor, Jon insists the first place to look is your ads and setup before you consider targeting, construction, or a “broken” algorithm. You have nearly full control over your results, and that starts with humility and taking responsibility.

Blaming Meta, chasing guru strategies, and floating conspiracy theories are all unproductive.

The breakdown effect explains much of the confusion. Early results reflect small samples; as data comes in, Meta moves budget toward the asset it predicts will maximize total results. Because costs rise with spend, the asset that received more budget may show a higher average CPA, which makes it look like Meta backed the loser.

Averages hide the trends and inflection points behind these decisions.

Consequently, one ad taking 50% of budget is normal and usually not a problem. It may simply reflect that Meta knows the other ads are less likely to convert. Creating separate ad sets to force delivery to a specific asset triggers auction overlap and hurts performance.

Jon advises worrying less about which ads get shown and more about how overall performance is doing.

Evaluate results in aggregate. The combined performance of all your ads, placements, and audiences matters more than any single top performer. Those elements work together, and one asset may perform better for a specific group even if it looks weaker on average. Manual intervention (turning off the ad getting the most spend, forcing budget elsewhere) usually does more harm than good.

Jon troubleshoots with three sequential questions:

• Why aren’t people clicking?
• If they’re clicking, why aren’t they converting?
• If they’re converting, why isn’t it attributed?

He warns against common misdiagnoses. Creative fatigue is rarely the true problem today unless you’ve restricted your audience or limited variations. Frequency is context-dependent rather than a magic threshold. And short-term dips are often just randomness, auction overlap from recent changes, competitive pressure, website issues, or event delays.

Evaluate seven-day windows at least two weeks in rather than reacting to daily swings.

When results struggle, simplify first to isolate the problem, then critique your ads honestly through the customer’s eyes: Do they lead with pain points rather than features? Is the creative scroll-stopping? Is the offer irresistible?

The most reliable improvement lever is always creating new, better, uniquely different ads. For those with access, the Push Delivery feature can test whether Meta’s distribution is correct (pushing an underused ad from 10% to 30% for a week), though the most likely outcome is confirmation that Meta had it right.

Build a Reliable Conversion Tracking Foundation

Conversion tracking tells Meta when key events happen. Without it, Meta can’t report results, and its algorithm can’t learn to find more converters. Tracking powers both reporting and optimization.

The website pixel is no longer dependable on its own due to privacy laws and browser settings. You should still use it, but pair it with the Conversions API (CAPI) to send first-party data you own; Meta reports an average 17.8% lower cost per result when CAPI is implemented.

There are two primary data types: website data (which piggybacks off the pixel) and CRM/customer data, valuable if you collect leads and close sales offline.

For setup: Shopify offers an easy automatic process; otherwise, the Conversions API Gateway generates the same events as your pixel, deduplicates them, and sends them via the API regardless of how your site is built. Meta’s new no-cost, one-click CAPI option for web is now active and worth watching, though Jon advises confirming its limitations before relying on it.

Most “inflated” or suspicious results have logical explanations:

• Event setup errors (firing on the wrong pages)
• Attribution quirks (team members loading confirmation pages)
• Deduplication issues from sending both CRM and web events without proper matching

None of these involve Meta fabricating numbers. Before assuming Meta broke something, check your event tracking, CAPI, landing page, and website performance, since these are within your control.

Where volume allows, optimize for a deeper, quality-reflecting event passed via CAPI rather than the standard event. More on that in the next section.

Prioritize Lead Quality and Value Over Cheap Volume

Lead generation (collecting contact information in exchange for something of value) is one of the best uses of ad dollars for many brands, since cost per lead is a fraction of a purchase and steady list building compounds over time.

But unlike e-commerce, where failure is obvious, lead generation problems are easy to miss, making it a science that rewards discipline.

A lead might cost under a dollar or hundreds of dollars. Determine your value per lead by tracking what leads do after submitting, then set a realistic cost goal. A $5 lead means nothing without knowing conversion rate and average revenue, and some funnels take months or years to produce a sale.

Lead quality is largely controlled outside your ads. Meta’s literal algorithm will exploit weaknesses (age, country, placement) to deliver cheap leads it doesn’t care about converting, and those are easy to monitor and address with value rules (bidding down on 65+, for example).

But the biggest factors are your follow-up process, sales script, speed of outreach, and email deliverability. Verify your CRM, tell people to check spam and promotions folders, and keep emails clean of excessive images and links.

Jon also experiments with optimizing for a quality event rather than the standard Lead event: tag subscribers who take meaningful actions (opening emails, clicking links), create a custom event via CAPI, and optimize for that. This requires the event to fire within the 7-day click window and enough volume (ideally 50+/week) to exit the learning phase.

Use disqualifying questions and conditional logic on instant forms to screen leads early so Meta learns from the right ones. Note that Meta now defaults Leads campaigns to combined “Website and Instant Forms,” so expand the collapsed “Single” option to send people to just one location.

For high-ticket products, Jon recommends going straight for the purchase before building a lead funnel. If you build one, attract the right leads with a tightly targeted lead magnet instead of a small discount.

The Bottom Line

• Trust Meta’s literal algorithm. Optimize for a true conversion (usually a purchase or lead), leave most targeting inputs alone as the suggestions they now are, and stop trying to control targeting like it’s 2018.
• When you have data Meta doesn’t, use value rules to bid down on problem groups (age, gender, placement, audiences) rather than restricting and disabling Advantage+.
• Start with one campaign and one ad set optimized for a conversion; consolidate budget and add complexity only when a proven exception applies and you can fund each piece.
• Before any change, ask “What problem does this solve?” and reject decisions based on feelings, vibes, or confirmation bias.
• Don’t take the Results column at face value. Use Breakdown by Attribution Settings and Compare Attribution Settings to separate view-through from click-driven results, and use breakdowns to diagnose problems, never to trim ages or placements.
• Creative diversification is your primary lever: genuinely diverse formats, angles, and personas, built in phases with the creative testing tool, instead of near-duplicates or dozens of ads at once.
• Meta already spends ~20–25% on warm audiences automatically. Define your Audience Segments to prove it, and don’t run separate remarketing campaigns that just create overlap.
• When results dip, look at your ads first, evaluate performance in aggregate over meaningful windows, and take responsibility instead of blaming Meta.
• Build a reliable tracking foundation by pairing the pixel with the Conversions API, and look for logical explanations before assuming Meta invented conversions.
• Ignore cost per lead in isolation. Determine value per lead, optimize for quality events where volume allows, and remember that lead quality is mostly won through follow-up and offers.