Meta’s bidding needs conversion volume to learn from. Spread a small budget across several ad sets and none of them ever exit the learning phase, so you pay exploration prices forever. Below this floor, consolidate rather than hire anyone.
Meta’s ranking systems now do most of the audience finding themselves. What you control is the quality of the information you give them and the quality of what you put in front of people.
Browser events are lost to tracking prevention, ad blockers and refused consent. Server-side events recover much of that — but only if event IDs match so the two streams deduplicate, and only if enough customer information is attached for Meta to match the person. An installed CAPI with poor match quality is doing far less than the green tick suggests, and checking it is the first thing we do.
Every ad set is a separate learner, and each one needs its own steady stream of conversions before it stabilises. Ten ad sets on a modest budget means ten permanent learners and inflated costs across all of them. Consolidating is uncomfortable because you lose granular control, and it is usually the single biggest improvement available in a small or mid-sized account.
Meta’s automated campaign types work well when there is genuine conversion volume and clean signal, and they work badly as a way to avoid thinking. We use them where the data supports it, hold structured control campaigns alongside, and keep an honest read on which is actually winning rather than assuming the newer product must be better.
Budget changes, creative swaps and audience edits all restart the learning phase. Accounts managed by nervous hands never leave it. Changes get batched to a schedule, with a documented reason, and then left alone long enough to be judged.
When the algorithm chooses who sees the ad, the ad decides who responds. That makes creative throughput the main constraint on most accounts.
A concept is a reason to buy: the problem it solves, a demonstration, a comparison, a founder explaining, someone handling the objection that stops people. Fifteen colour variants of one image is one concept. Genuine testing moves between angles, because no amount of editing rescues an angle nobody cares about.
Frequency climbing while click-through falls and cost per acquisition rises is fatigue, and each account has a fairly predictable interval. We track it per concept so replacements are planned, not panicked, which is what stops the monthly cycle of "it suddenly stopped working".
Most creative tests never reach a result — too many variants, too little budget, called after four days on a difference that is noise. We size tests so they can finish, and we call them with a significance test rather than a hunch. Our own A/B calculator is free to use if you want to check ours.
A restricted ad account stops the business, not just the campaign. Recovery takes days when it works at all, and it is almost entirely preventable.
Proper business verification, two-factor on every user, partner access instead of shared logins, no personal profiles carrying the whole account, and a second admin who is not one person’s login. Most bans we are called in to fix trace back to one of these.
Health, finance, employment, housing and anything with a before-and-after have specific rules, and repeat disapprovals count against the account itself. It is cheaper to check a concept at brief stage than to appeal it after it has cost you standing.
Housing, employment, credit and social issues restrict your targeting and require declaration. Declaring late forces a rebuild and resets everything the account has learned. Getting this right on day one is free; getting it wrong in month three is not.
A real audit of your live campaigns — structure, tracking, creative and where the budget is leaking. Free, no obligation, and yours to keep even if you never work with us.
No, and the answer changes by account. It performs well with real conversion volume and clean signal. On thinner accounts, or where you need control over what is shown to whom, a structured campaign still wins. We run both and let the numbers decide rather than adopting whatever is newest.
In our experience it is almost always one of four things: a consent banner update cutting your event volume, a tracking change nobody flagged, creative fatigue arriving on schedule, or someone making frequent small edits and holding the account permanently in learning. The audit will tell you which.
More than you expect. The specific number depends on spend and audience size, but the honest answer is that creative supply is the ceiling on most Meta accounts. We will size it in the audit, and if your pipeline cannot support it we will say so before you sign rather than blame it in month three.
Sometimes. Appeals work more often when they are specific, calm and written quickly, and much less often after several rushed attempts. What matters more is the prevention work, which is why we do it on day one.
Same platform, same account, and usually the same campaigns with placement-appropriate creative. Where Instagram deserves its own budget, it is because the creative format genuinely differs, not because the tab does.
You do. Ad account, pixel, datasets, custom audiences and creative all stay in your name and we are a partner on them, removable in two clicks.
Structure, signal quality, creative and account safety — a written read within two working days.