Handbook · by Dr. Louis Korczowski

One ad eats half your Meta budget? Keep it. The creative strategist handbook to scale ads without cost inflation.

Splitting a test campaign from a scale campaign costs a small account money and rarely fixes rising costs. Feed one purchase campaign from three engines instead: organic content that finds hooks for free, a squad of creators paid on results whose posts run as partnership ads, and offers and statics that close the sale. Eleven recipes, US rules, a prompt for your AI on every step.

Ask your AISummarize this handbook and apply it to your situation: ChatGPT ClaudeMarkdown AI skills on GitHub

TL;DR

Meta ads get more expensive when one campaign is fed too little, too similar content. Splitting a "test" campaign from a "scale" campaign costs money at small budgets and rarely fixes it. Run three engines instead. An organic content engine that finds hooks for free. A small squad of creators paid on results, whose posts run as partnership ads. And a creative strategy that turns the winners of both into ads with strong offers and statics. All of it goes into one campaign optimized for purchases, with a test lane inside it, a creative count sized to the budget, and a verdict read on blended return over weeks and months. Below about ten orders a day, daily ROAS is noise. The AI reads the data. You keep the offer, the brief and the decision.

The method

Three engines feed one campaign.

Content strategy finds the hooks for free. Performance creators multiply them from handles people trust. Creative strategy turns the winners of both into ads that close. Eleven recipes, each with one thing to do today and a prompt for your AI assistant.

Fig. 1 · Three engines, one campaign● the campaign stays simple, the inputs get rich

When	What	Time
Every day	Organic posts, replies to comments, creator chat	1 hour
Every week	Size live ads, run the test lane, partnership codes, offers and statics, 7-day read	3 to 4 hours
Every month	Contribution after ad spend, creator leaderboard, organic winners to ads	2 to 3 hours
Each sale	Retargeting lane with a lifetime budget	30 min

01Name the problem before you touch the structureDiagnose · 30 min · once, then monthly
02Keep one campaign and optimize for purchasesStructure · 20 min · once
03Size the number of live ads to your budgetStructure · 10 min · weekly
04Let one ad take the budget, and test new ones in a laneStructure · 15 min · weekly
05Speedrun the account with your organic winnersCreative strategy · 1 hour · once, then monthly
06Build an organic engine that feeds the adsContent strategy · 1 hour a day · daily
07Run a squad of 5, then 10 creators paid on resultsPerformance UGC · 2 hours a week · weekly
08Run creator posts as partnership adsPerformance UGC · 30 min · per creator
09Close the sale with offers and staticsCreative strategy · 1 hour · weekly
10Run the campaign weekly, judge it monthlyMeasure · 30 min weekly · 1 hour monthly
11Run a limited-time sale without breaking the main campaignMeasure · 30 min · per sale

Why costs rise

The auction got more expensive in 2026. Your inputs decide how much of it you pay.

Meta's average price per ad rose 14% year over year worldwide and 23% in the US and Canada in the third quarter of 2026 (Meta Q3 2026 results via PPC Land). Common Thread Collective measured portfolio CPMs at $18.01 in March 2026, a four-year high, and calls it structural repricing from advertiser demand (CTC, May 2026). Three things in your account decide what you pay: how many distinct concepts compete, whose handle they run from, and how strong the offer is when the click lands. Advertisers already move this way. Meta CFO Susan Li said partnership ads grew over 97% year over year and passed a $15 billion annual revenue run rate (Meta Q3 2026 earnings call via PPC Land).

The recipes

Eleven recipes for scaling a Meta account.

Diagnose

Name the problem before you touch the structure

30 min · once, then monthly

An unprofitable campaign has two suspects: the creative or the offer. Check both before you rebuild anything.

You need

Ad-level results for the last 14 to 28 days, with 3-second views and ThruPlays or 15-second views
Your store analytics: sessions, add to cart, checkout, orders
The product page each ad sends to

Steps

Creative suspect: read hook rate (3-second views divided by impressions), hold rate, click-through rate and cost per click for each ad. People who stop and leave point to the creative. People who click on the wrong promise point to the targeting the creative creates.
Offer suspect: people click, land, see the price and leave. Read add-to-cart rate and checkout rate on the landing page. Strong clicks with weak add-to-cart point to the offer or the page.
Open the page on a phone, as a stranger, right after watching the ad. If you cannot find the product, the price and the promise of the ad within about five seconds, the page is part of the problem.
Write one sentence per suspect: guilty, innocent or unknown. Work on both when both are guilty.

Creative is where the buyer meets the brand, and Meta's own line to advertisers in 2026 is outcome-based: "diversify your creative, test, and let the system optimize" (Meta spokesperson to Marketing Brew, September 2026). The five-second page test is a working rule from my own accounts: a visitor who cannot match the page to the ad leaves.

Avoid

Rebuilding the campaign because of one bad day
Blaming the algorithm before reading the page
Fixing the creative when the price is the problem

TodayWatch your best ad on your phone, tap through to the page and time how long it takes to find the price.

For your AI

Copy this prompt into your assistant:

Here are my ad results for the last [14 or 28] days: [paste ad name, spend, impressions, 3-second views, ThruPlays or 15-second views, CTR, CPC, purchases] and my store funnel: [sessions, add to cart, checkout, orders]. Diagnose whether the creative, the offer or the page is the bottleneck, ad by ad, and tell me what data is missing.

Skill: statur-meta-ads-scaling · install: npx skills add lkorczowski/content-strategist-AI-cookbook --skill statur-meta-ads-scaling

Structure

Keep one campaign and optimize for purchases

20 min · once

One campaign, broad targeting, purchase as the only optimization event. A separate test campaign spends money that a small account cannot afford.

Steps

Keep one sales campaign with campaign budget (Advantage+ campaign budget) and broad targeting. Exclude only audiences you can justify.
Optimize for purchase. Keep it there even when the ad set says "learning limited".
Do not create a "Winners" campaign and a "Test" campaign. In my accounts, an ad that spends in a test campaign often stops spending once moved. Each split also divides the data the system learns from.
Keep your existing ad account and pixel when you change store or domain. Add the domain and connect the Conversions API instead of starting from zero.
Add a second campaign only when the first one gets well over 50 purchases a week. Split by job: one campaign per bid strategy, or one for existing customers and subscribers. Every campaign stays optimized for purchases.

Meta's learning phase needs roughly 50 optimization events a week, and "every extra campaign and ad set waters down your budget" (Jon Loomer, July 2026). The 50 are counted per ad set (Jetfuel, August 2026). Common Thread Collective tells accounts that split into several campaigns without 20 conversions a week each to "consolidate your best-performing ads into a single, broad-targeted campaign" (CTC, May 2026). My rule: optimizing for add to cart to exit learning faster finds exactly what you asked for, people who add to cart and stop there. In my accounts with an average order around $50, purchase optimization gets enough signal. Some agencies accept a cart objective when purchases are very rare (Jetfuel, August 2026). I keep purchase on every account that already sells. The larger accounts I work with spend four figures a day. They run a few purchase campaigns split by bid strategy and by customer type, and each one sells well above 50 times a week.

Avoid

Switching to add-to-cart optimization to leave "learning limited"
Splitting a $60 daily budget across three campaigns
Creating a new ad account to "reset" the algorithm

TodayCheck the optimization event of every active ad set and write it down.

For your AI

Copy this prompt into your assistant:

Here is my Meta account structure: [campaigns, ad sets, budgets, optimization events, audiences, number of ads]. My daily budget is [amount] and I get about [n] purchases a week. Tell me how to consolidate it into one purchase campaign, what to pause, and what I lose or keep by doing it.

Skill: statur-meta-ads-scaling · install: npx skills add lkorczowski/content-strategist-AI-cookbook --skill statur-meta-ads-scaling

Structure

Size the number of live ads to your budget

10 min · weekly

Give each live ad $5 to $10 a day. Divide your daily budget by 7.5 to get the number of ads that can learn.

Steps

Count your live ads. At $60 a day, 6 to 12 ads can learn; 20 cannot.
Count distinct concepts. Two videos of the same person that differ only by on-screen text likely count as one creative for the system.
Pause the ads that never spent and the near copies of your winner. Keep the winner and the ads that brought purchases.
When the account is profitable, add $10 a day of budget for each genuinely new concept. This is my working rule.

The $5 to $10 rule is my own heuristic from running small accounts, and other practitioners make the same point. Jon Loomer: "five ads splitting $50/day won't generate meaningful purchase data" (July 2026). Meta does not publish how it groups similar ads. One agency media director told Marketing Brew that two videos of the same person with different on-screen text "would actually be considered the same creative" (Billion Dollar Boy in Marketing Brew, September 2026).

Avoid

Launching 20 variations of one video
Judging an ad that spent $3
Adding ads without adding budget

TodayDivide your daily budget by 7.5 and compare the result with your number of live ads.

For your AI

Copy this prompt into your assistant:

My daily budget is [amount]. Here are my live ads with spend and purchases over the last 14 days and a one-line description of each: [paste]. Group them into distinct concepts, tell me how many ads my budget can support at $5 to $10 per ad per day, and which ones to pause.

Skill: statur-meta-ads-scaling · install: npx skills add lkorczowski/content-strategist-AI-cookbook --skill statur-meta-ads-scaling

Structure

Let one ad take the budget, and test new ones in a lane

15 min · weekly

A winner that takes a large share of the budget is normal. Never cut an ad that sells. Give new concepts a small, timed test lane inside the same campaign.

Steps

Leave the winner alone while blended results are good.
Launch new concepts in a batch of 2 to 5 through Meta's creative testing feature inside the existing ad set, or in one testing ad set inside the same campaign.
Cap the test at about 20% of the budget and run it about 7 days.
At the end, keep the concepts that sold at an acceptable cost and let the campaign decide how much they get. Do not expect a test winner to take over the budget.
Intervene on concentration only when blended results are weak and you are learning nothing about the other ads.

Spend concentrates by design. In Motion's benchmarks of 6,015 accounts, about 5% of ads became winners and took 55% of Meta spend. The share was 23% in accounts under $10,000 a month and 64% in the largest (Motion, Creative Benchmarks 2026). Motion sells creative analytics. Loomer recommends leaving an ad that takes the budget alone when results are good, and starting new ads with a creative test (Jon Loomer, updated April 2026). Meta's creative testing feature tests 2 to 5 ads inside an ad set. It ran 7 days by default in Loomer's tests, and Meta recommends no more than 20% of the budget (Jon Loomer, updated January 2026). Foxwell: intervene only "if spend is heavily skewed and you're not learning" (Andrew Foxwell, June 2026).

Avoid

Pausing the winner to "force" spend on new ads
A test campaign with its own budget
Reading a test before its planned end

TodayPick the 2 to 5 most different concepts you have and plan this week's test.

For your AI

Copy this prompt into your assistant:

One ad takes [share]% of my spend. Blended results over the last 14 days: [spend, orders, revenue, blended ROAS]. New concepts ready: [describe]. Tell me whether to intervene, and plan a creative test inside my existing campaign: which 2 to 5 ads, budget cap, duration and the metric that decides.

Skill: statur-meta-ads-scaling · install: npx skills add lkorczowski/content-strategist-AI-cookbook --skill statur-meta-ads-scaling

Creative strategy

Speedrun the account with your organic winners

1 hour · once, then monthly

Start paid with your best organic videos, a strong offer and a call to action. In my accounts, this is the quickest way to find a first profitable ad.

Steps

Pick the organic videos that beat their format's usual results on mature data.
Keep the first three seconds. Add a call to action and the offer on screen and in the first line of copy.
Run the original post as the ad where you can, so its likes and comments stay visible.
Put them in the test lane of recipe 04 against your current best ad.

Taylor Sicard on Arrae, a supplement brand: "Arrae's 1,000 ads did not start as 1,000 ideas. They started as a handful of organic winners the team scaled" (July 2026). Organic and paid have different jobs, so keep the hook and rebuild the rest (Ayala Daly, June 2026). The full method is in Organic winners make your ads cheaper.

Avoid

Boosting a video because the team likes it
Changing the hook of a proven video
Running an organic winner with no offer

TodayList your three best organic videos of the last 90 days.

For your AI

Copy this prompt into your assistant:

Here are my organic videos of the last 90 days with views, 3-second retention, saves and shares: [paste]. Shortlist the ones that fit paid, and for each write the call to action and the offer line to add, keeping the first three seconds identical.

Skill: statur-organic-to-paid · install: npx skills add lkorczowski/content-strategist-AI-cookbook --skill statur-organic-to-paid

Content strategy

Build an organic engine that feeds the ads

1 hour a day · daily

Start organic before you push spend, and run it every day. Volume finds the hooks your ads will use.

Steps

Pick one primary channel and post every day, with formats you repeat and judge as a group.
Set a volume target. For a consumer brand that wants organic to move sales, I aim at 100,000 views a day across channels, about 3 million a month. Below that, organic rarely moves the numbers on its own.
Log every video and read it on day 7. Promote the outliers to recipe 05.
Make educational and reaction formats around your category, and answer comments with videos. They open new audiences and give the ads proof.

Jetfuel puts it simply: "Organic is the cheapest ad testing you will ever run" (Jetfuel, August 2026). The 100,000 views a day target is my rule of thumb from the brands I work with, and no public benchmark gives a threshold. Read it as an order of magnitude. The daily routine is in the Creative-Led Growth Handbook, the weekly reading in the Content Testing Loop, and the analytics in Own your social media analytics.

Avoid

Waiting for organic to "work" before running any ad
Posting only product videos
Judging a format on one video

TodayCount your organic views of the last 30 days across every channel and divide by 30.

For your AI

Copy this prompt into your assistant:

My organic views over the last 30 days per channel: [paste]. My product and audience: [describe]. Plan the daily organic routine and the formats that would take me toward 100,000 views a day, and tell me which videos should become ads.

Skill: statur-creative-led-growth · install: npx skills add lkorczowski/content-strategist-AI-cookbook --skill statur-creative-led-growth

Performance UGC

Run a squad of 5, then 10 creators paid on results

2 hours a week · weekly

In my experience, a small group you coach outperforms a marketplace of strangers. Pay on results, and pay a small fee for each approved video while a creator ramps up.

You need

A group chat (WhatsApp or Discord)
Product to send
A one-page brief with your best hooks and formats
A short written agreement per creator

Steps

Recruit 5 creators who already make content in your category. Start with followers and customers who create.
Send product and the brief. Share your organic winners in the group every week.
While a creator ramps up, in the first one to three months, pay a flat fee per approved video. Then move each creator to performance pay: a share of the ad spend or sales their videos drive.
Rank creators on a weekly leaderboard by sales and margin.
Grow to 10 creators once the first 5 deliver one usable ad each per week.

Operators in 2026 move proven creators to performance and keep a small floor for new ones: cutting the base to zero for everyone dries up the pipeline (Roster, October 2026). Trybe, a creator program platform, ranks creators on real-time leaderboards and charges 1.25% of ad spend, 1.5% of attributed sales or 1.5% of flat creator payouts (Trybe pricing page). Measure the creator program on margin. In the worked example of zeroto1, an agency that runs Trybe programs, the same program reads 7.5x on revenue and 1.5x on margin (zeroto1, October 2026). In the US, free product given with the expectation of a post is a material connection the post must disclose clearly. The brand needs a reasonable program to train and monitor its creators. The platform's paid partnership label may not be enough on its own (FTC Endorsement Guides FAQ). Platforms, payment models and pitfalls are in the Performance creator programs dossier.

Avoid

Paying for views, which invites fake views
A group of 30 creators nobody coaches
Product sent with no written terms

TodayList 10 people who already post about your category and follow your brand.

For your AI

Copy this prompt into your assistant:

My product: [describe]. My best hooks and formats: [paste]. Write the one-page brief for 5 creators, the welcome message for the group, the ramp-up fee per approved video and the performance terms after it, and the weekly leaderboard columns.

Skill: statur-creator-program · install: npx skills add lkorczowski/content-strategist-AI-cookbook --skill statur-creator-program

Performance UGC

Run creator posts as partnership ads

30 min · per creator

The best creator videos run from the creator's handle, as partnership ads, in the same campaign as your own content.

Steps

Ask each creator for a partnership ad code or ad permission on their account through Meta's creator tools.
Run their existing post by its ID, so the engagement stays on the original post.
Put partnership ads in the same campaign as your in-house ads. My working target: at least a third of your live ads come from creators.
Replace each creator winner after about five weeks with a new video from the same creator. This is my working rule.
On TikTok, the equivalent is Spark Ads with an authorization code.

Meta ran 20 lift studies with advertisers in Europe, the Middle East and Africa in the first half of 2025 and published them in an October 2026 white paper. Campaigns with at least 30% partnership ad assets showed a 45% lower median cost per incremental purchase. This is Meta's own data, at 92% confidence, below the usual 95% bar (PPC Land). A second dataset comes from Agentio, a vendor of creator ads, and it is more modest. Across $130 million of spend, partnership ads had 19% higher click-through and 5% lower cost per acquisition than licensed creator content run from brand accounts. Their CPM was 19% higher. Winning ads started to fatigue after about 36 days on average (Agentio via Net Influencer, August 2026). In June 2026, Meta announced it would merge Creator Marketplace and the Partnership Ads hub into one Meta Creator Marketing Hub (Net Influencer). Meta's developer guides cover partnership ad permissions and creation (Meta for Developers); TikTok documents Spark Ads (TikTok Ads Help).

Avoid

Downloading the creator's video and running it from your page only
A separate "creator" campaign
A creator post with no clear disclosure

TodayAsk your best creator for a partnership ad code on their best post.

For your AI

Copy this prompt into your assistant:

Here are my live ads, split between in-house and creator content, with spend and purchases: [paste]. Tell me my share of creator ads, which creator posts to run as partnership ads next, and which creator winners are older than five weeks and need a replacement.

Skill: statur-meta-ads-scaling · install: npx skills add lkorczowski/content-strategist-AI-cookbook --skill statur-meta-ads-scaling

Creative strategy

Close the sale with offers and statics

1 hour · weekly

The video earns attention. The offer and the static close the sale. Make the offer hard to refuse and identical on the ad and the page.

Steps

Write one first-order offer that removes every excuse: a strong first-purchase discount, an extra product, a gift, early access. Test the size of the first order too.
Show the offer the same way on the ad and on the page. A strikethrough price is the easiest form to keep consistent.
Make statics from your winning hooks: the product, the price, the offer, one proof. Keep them in the same campaign as the videos.
Treat every product page as a landing page: one promise, the price, the offer, proof, the button.
Change the offer before you change the creative when clicks are good and sales are weak.

Across 243 DTC ad teardowns, 38 of the 46 ads that carried an offer lost it partly or fully on the page. A discount behind an email form failed 31 times out of 32. A dollar saving or a strikethrough price failed least often, in 6 of 11 cases (Taylor Sicard, August 2026). In 17 US accounts and $14.9 million of spend from June to August 2026, statics took 32% of spend and 43% of each account's top 10 ads. In the 12 accounts with enough purchases, their median cost per purchase was within 4% of video. Biddyco sells statics, so treat this as one agency's data (Biddyco, September 2026).

Avoid

An offer hidden behind an email form
A different price on the ad and on the page
Statics that look like every competitor's AI template

TodayWrite three first-order offers and pick the one a skeptical customer could not argue with.

For your AI

Copy this prompt into your assistant:

My product, price and margin: [describe]. My current first-order offer: [describe]. Write five offers that remove the buyer's excuses without killing my margin, the static ad for the best one (headline, visual, proof, offer line), and check that my product page [paste] shows the same offer.

Skill: statur-meta-ads-scaling · install: npx skills add lkorczowski/content-strategist-AI-cookbook --skill statur-meta-ads-scaling

Measure

Run the campaign weekly, judge it monthly

30 min weekly · 1 hour monthly

Below about ten orders a day, daily ROAS is mostly noise. That threshold is my working rule. Operate on a rolling 7-day view and judge blended return and margin over the month.

Steps

Weekly: read the last 7 days against the 7 days before. Spend, orders from your store, revenue, blended return (store revenue divided by total ad spend) and cost per new customer.
Monthly: compute contribution after ad spend from your store and your margins. This is the verdict.
Check that the Conversions API is connected and deduplicated. Ask buyers "how did you hear about us?" after checkout to catch the sales the pixel misses: post-purchase survey apps such as Fairing are free up to 200 orders on Shopify (Fairing on the Shopify App Store).
Read Meta's reported ROAS as one signal among others. Test the incremental attribution setting against your store numbers before you trust it.
Look at the day only during a limited-time sale.

Loomer: "Evaluate seven-day windows at least two weeks in rather than reacting to daily swings" (July 2026). CTC: "Stop optimizing ad-level ROAS" and manage account-level efficiency (May 2026), anchored on contribution margin (Taylor Holiday, February 2026). Meta's reported numbers can move while sales stay flat. Since March 18, 2026, only outbound link clicks count as clicks for attribution. Across 50 large accounts, reported 7-day click ROAS fell 13% overnight with no change in business results (CTC, April 2026). In Haus experiments from July 2025 to June 2026, campaigns optimized to Meta's incremental attribution earned 1.26 times the incremental return of standard attribution, up from 0.80 times a year earlier. Haus, which sells incrementality tests, calls the reversal "real but still developing" (Haus, July 2026).

Avoid

Doubling the budget after one good Saturday
Cutting it after one bad Sunday
Judging the month on Meta's ROAS alone

TodayPut your store revenue and total ad spend for the last four weeks side by side.

For your AI

Copy this prompt into your assistant:

Here are my last [8] weeks: ad spend, store orders, store revenue, new customers, Meta-reported purchases and ROAS: [paste]. My gross margin is [x]%. Build the weekly table with blended return and cost per new customer, the monthly contribution after ad spend, and tell me what changed and what to do next week.

Skill: statur-meta-ads-scaling · install: npx skills add lkorczowski/content-strategist-AI-cookbook --skill statur-meta-ads-scaling

Measure

Run a limited-time sale without breaking the main campaign

30 min · per sale

A sale with an end date, such as Black Friday or Cyber Monday, gets its own small retargeting lane with a lifetime budget. The main campaign keeps running.

Steps

Create one small campaign with ad set budget and one ad set aimed at people who engaged or clicked in the last 30 days.
Put in statics and short videos that say the deadline: "ends tonight", "last hours".
Use a lifetime budget that ends with the sale. Keep it small next to the main campaign.
Update the main campaign's copy with the offer for the same dates. Pause the lane the day the sale ends.

I run this lane with ad set budget so I control who sees the deadline and how much it costs. In my accounts, the broad main campaign keeps reaching past visitors and keeps learning while the sale runs. This is the one case where reading results day by day makes sense.

Avoid

Moving the main budget into the sale campaign
Keeping deadline ads live after the deadline
A sale lane with no end date

TodayPut the dates of your next sale in the calendar with a budget for the lane.

For your AI

Copy this prompt into your assistant:

My next limited-time sale: [dates, offer]. Main campaign budget: [amount]. Plan the retargeting lane: audience window, budget, three statics and two short videos with deadline copy, and the copy changes for the main campaign.

Skill: statur-meta-ads-scaling · install: npx skills add lkorczowski/content-strategist-AI-cookbook --skill statur-meta-ads-scaling

Coming next

The creative strategist toolkit behind this handbook.

Five companion playbooks are in progress. Each one gets its page and its skills in the public cookbook.

Ads data to AI

Pull Meta ad results into one place your AI can query: spend, hook rate, hold rate, purchases per ad and per creator. Meta opened an official ads connector for Claude and ChatGPT in beta in April 2026 (PPC Land).

Organic and partnership in Ads Manager

One naming system for in-house posts, creator posts and statics, so every ad traces back to its concept, creator and offer.

Creator bonuses and leaderboard

Tag every creator ad, compute bonuses from spend and sales, publish a leaderboard for the squad and a private page per creator.

Attribution dashboard

Blended return, contribution after ad spend and post-purchase answers in one internal dashboard.

AI statics that look like you

Generate static ads from your own photos, proofs and brand rules, so they stop looking like every competitor's template.

For your AI

The same method, written for your assistant.

Skills in the open Agent Skills format, maintained on GitHub so every update reaches you. Upload them in Claude, paste them in ChatGPT, or install them with one command.

statur-meta-ads-scaling

Diagnose a Meta account, size live ads to the budget, plan the test lane, partnership ads, offers and statics, and run the weekly and monthly reads.

statur-organic-to-paid

Turn proven organic videos into ads and test them against the current best ad.

statur-creator-program

Design, run and audit a creator squad paid on results.

Rules every skill follows. Drafts only: the AI never sends or publishes for you. No passwords: you sign in yourself. No promised views.

The method behind it

This handbook connects four playbooks: the Creative-Led Growth Handbook for the organic engine, the Content Testing Loop for reading it, Performance creator programs for the squad, and Organic winners make your ads cheaper for the handoff to paid. Every fact links to its source. My own working rules are labeled where they appear: $5 to $10 per ad per day, 100,000 views a day, the five-second page test, purchase over add to cart for any account that sells, splitting large accounts by job, a third of live ads from creators, a new creator video every five weeks, and about ten orders a day before a daily read means anything.

Questions

Scaling Meta ads, answered.

Should I split my Meta ads into a testing campaign and a scaling campaign?

At small budgets, keep one campaign. Every extra campaign splits the data the system learns from, and in my accounts an ad that spends in a test campaign often stops spending once moved. Keep one purchase campaign and test new concepts in a small, timed lane inside it, capped at about 20% of the budget.

Why does one ad take most of my budget?

Because the system bets on the ad it predicts will sell, and near-identical ads likely count as one. It is normal while blended results are good. Reduce the number of near copies, size your live ads to the budget, and add genuinely different concepts in batches.

How many ads should I run for my budget?

About one live ad per $5 to $10 of daily budget. At $60 a day, 6 to 12 ads can learn. Count distinct concepts: two files of the same concept count once.

Should I optimize for add to cart to exit the learning phase?

Keep purchase if you already sell. A cart objective finds people who add to cart and stop there. Keep purchase as the optimization event and accept "learning limited" at low volume.

How do I stop Meta ad costs from rising?

Meta's price per ad rose 14% worldwide and 23% in the US and Canada in the third quarter of 2026. The lever you control is what you feed the auction: more distinct concepts, creator posts run as partnership ads, organic winners with a strong offer, and statics. Judge the result on blended return and margin.

End note · selling in France or the EU

This handbook follows US rules. Three things change for a brand that sells in France. Creator posts must show the mention "Publicité" or "Collaboration commerciale" for as long as the promotion runs. Under the June 9, 2023 influencer law, a breach can bring up to two years in prison and a 300,000 euro fine, and the brand shares liability with the creator (GHR summary of the law; Journal du Net). Since January 1, 2026, a written contract is required once payments and gifted product from one brand to one creator reach 1,000 euros excluding VAT in a year (decree of November 28, 2025, via Eurojuris). Meta's price per ad grew 10% in Europe against 14% worldwide in the third quarter of 2026 (PPC Land). Check your own CPMs before you reuse the per-ad budget in euros. Have a lawyer check your creator agreements before launch.

Version 1.1 · last updated 2026-10-26: Recipe 07 aligned with the creator programs dossier (ramp-up fee for the first one to three months). Ask your AI v2: a source contract tested on ChatGPT and Claude. First edition: eleven recipes across three engines (content strategy, performance UGC, creative strategy) feeding one purchase campaign, a prompt per recipe, US rules with an end note for France and the EU, and one new companion skill in the public cookbook. Changes are logged in the changelog. Author: Dr. Louis Korczowski.

Bring the three ideas you repeat on every sales call.

Work with me

Two-minute application. I reply within 24 hours.