Comment Tester & Scaler Sur Meta Ads En 2026 (Stratégie Complète)

Summary

Matteo Pellegrini presents a Meta Ads testing and scaling system for e-commerce, built around continuous creative research, structured testing, and a steady supply of diverse ads. He says his team uses the approach across multiple brands and markets, with Meta accounting for about 70% of their advertising budgets. His recommendations reflect his team’s experience and are operating guidelines, not fixed rules.

Core process

1. Research before production. Study customer desires and problems, competitor ads, market angles, and audience avatars before scripting. Pellegrini argues that weak research leads to off-target creative.
2. Set a weekly creative strategy. Review what worked and what failed, which audiences and angles produced winners, and what to test next. Adapt the plan to results rather than treating it as fixed.
3. Script and produce to match the strategy. Use customer language and emotional triggers in scripts. Choose formats—such as video, UGC, static, VSL, or interviews—that suit the message, and make sure the visuals reinforce it.
4. Test, scale, and learn. Identify recurring patterns in winning ads and use those findings to guide the next week’s research and creative plan.

Creative testing after Andromeda

Pellegrini says Meta’s Andromeda system makes it less effective to launch many ads with nearly identical creative identities. He recommends prioritizing new concepts, angles, scripts, faces, and footage, rather than relying only on minor edits.

His working mix is approximately:

• 50% net-new concepts
• 20% iterations
• 30% broader variations, or “declinations”

This is a guide, not a permanent formula; the split should change as the account generates learnings.

• Iteration improves an existing ad, often by testing a new hook or a more specific message while retaining some of the original structure.
• Declination carries a proven pattern into a meaningfully different execution—for example, retaining a back-pain angle while changing the format, footage, script, or point of view.

For example, if an ad for a men’s product performs better when women appear in the video, Pellegrini recommends treating that as a pattern to investigate and applying it to future scripts. In a health-related example, a daughter speaking emotionally about her father’s back pain could be tested against direct-to-customer messaging.

Use competitor ads as research inputs, then adapt successful scripts to the brand rather than inventing every idea from scratch.

Campaign structure and testing budgets

ABO testing campaign

Create separate ad sets for different concepts, generally with 4–10 ads per ad set, often testing hook variations. Pellegrini recommends broad targeting—roughly ages 18–65—and, based on his experience, specifying the language for the target market.

He suggests allocating roughly 20–30% of the total budget to testing. A stated starting reference is around 1–3× AOV, but the right amount depends on the account’s performance and capacity to scale. Avoid judging an ad from a single day; running tests over several days can reduce the risk of overreacting to an unusually good or bad day.

CBO campaigns

• CBO Master: Move promising ads into a scaling campaign, grouping ad sets logically by angles, awareness stages, and concepts. Pellegrini recommends no more than 15 ads per ad set and setting a minimum ad-set spend so Meta has a chance to evaluate each one.
• CBO Zombie: Group ads that spend but have weaker results into a separate CBO and let Meta try to find profitable combinations. Continue monitoring and remove ads that remain clearly poor performers.

Advanced advertisers should avoid adding very large batches of ads directly to the Master CBO; Pellegrini says adding 300–500 ads at once disrupted his team’s campaigns.

Beginners and account expansion

For beginners without a separate scaling budget, Pellegrini recommends starting with one CBO for both testing and scaling, with a suggested budget of roughly €300–€500, and adding ad sets gradually. He advises focusing on one strong product or offer before expanding to many campaigns.

More advanced brands can create separate CBOs by product, offer, mechanism, or materially different audience or landing page.

Scaling and performance management

The scaling flywheel is:

Produce more quality creative → find more winners → expand spend → encounter creative fatigue → create new concepts.

Scale only when performance is supported by enough data. Pellegrini recommends evaluating profitability over seven days, rather than making decisions based only on the previous day.

As general guidance, he suggests:

• Increasing budgets by about 10–20%, once or twice daily depending on the account.
• If the account is unprofitable over the recent three days and the previous day, reducing budgets by roughly 15–20%.
• At break-even, gathering more data rather than cutting immediately.

For smaller budgets, he suggests stepping up gradually—for example: €75 → €100 → €120 → €150 → €200 → €350 → €500. Stronger ROAS can justify more aggressive increases, while rapid scaling can expose less-profitable audience segments.

He recommends scheduling automated budget adjustments between midnight and 1 a.m., based on recent profitability. These are his team’s operating rules, not universal platform requirements.

Creative volume and KPIs

Pellegrini’s approximate weekly creative-output benchmarks rise with monthly spend:

Monthly spend Suggested weekly output €0–€50K At least 5 new concepts with about 4 variations each; roughly 20–30 hours of work Around €100K Approximately 10 concepts Around €250K Approximately 25 concepts Around €1M Approximately 100 concepts, with an estimated 400–500 hours of production work

These figures are rough minimums, and some spend-tier details are not fully clear in the subtitles.

Track the creative hit rate: the number of winning ads divided by the number launched. Pellegrini gives 10–30% as a healthy range, while noting that results vary with creative quality, industry, and seasonality. If the hit rate is low, diagnose performance by creative type and approach—such as UGC, AI-generated content, or repeated iterations—and increase what is working while reducing what is not.

A winner should attract spend and maintain stable CPA or ROAS over time. A brief early spike followed by a collapse after three to five days is not, in his view, a durable winner. Stronger “banger” ads can sustain substantially more spend.

Creative diversity and audience coverage

Pellegrini’s planning model combines angle × avatar × awareness stage × creative concept. Execution quality—including visuals, pacing, and emotional impact—also affects how well the message lands. He calls this execution factor a “consistency index,” a term he says he coined.

Diversify by changing the format, point of view, angle, audience, or awareness level around validated messages. Avoid changing every variable at once, since that makes results harder to interpret.

Use customer comments, Reddit discussions, and ad feedback to build specific avatar profiles and identify the language people use. Pellegrini encourages precise ads even if an individual ad reaches a smaller audience. His rationale is that more relevant creative entry points can improve audience coverage, reduce fatigue, and give Meta more signals for matching ads with prospects.

Tracking check

On Shopify, check Customer Events and confirm that Meta tracking—including the Conversions API, if used—is connected and set to “Always on,” rather than “Optimized.” Pellegrini says this helps ensure event data is sent consistently.

Presenter and source

Matteo Pellegrini — YouTube channel: Matteo Pellegrini.