Most of the money lost on Meta is lost before a single ad runs, in the twenty minutes where nobody asks the uncomfortable questions. These are the ten I would ask if I were hiring me. Each one comes with what a straight answer sounds like and what should end the call.
Why this exists. I lose deals to agencies that promise a guaranteed number, and I am not going to start promising one. I would rather you take this list to them than take my word for anything. If they answer these ten well, hire them.
Creative volume is the whole game, and it is the first thing an agency quietly cuts because it is the expensive part. In my own e-commerce account six ads carry almost all of a 69,705 dollar result and the rest produced close to nothing. Nobody knows in advance which ad that will be, so the only way to reach it is to keep putting new ones in front of the algorithm.
A number, a person attached to it, and an explanation of what happens to the ads that lose. Five to fifteen new creatives a week is a working range for an account with a live offer.
“We refresh creative when performance drops.” That means two or three ads a month, and it is the most common reason a coach tells me their last agency tried Meta and it did not work.
Anyone can show you a winner. What you are actually buying is the judgment that gets used when something is not working, which is most of the time in the first six weeks.
A specific account, a specific reason and what was changed. The offer was wrong, tracking was broken, the audience was too small, the budget was below what that objective needed.
A portfolio with no losses in it. Either the losses are being hidden or there have not been enough accounts to have any.
This is the question that costs people the most money and almost nobody asks it. If the agency runs your ads inside their own Business Manager, your pixel history and your custom audiences can leave with them, and that history is worth more than the campaigns.
Your Business Manager, your ad account, your pixel or dataset, your Facebook and Instagram assets. The agency gets partner access and you can revoke it in a minute.
“We run everything through our BM, it is simpler.” It is simpler for them.
The account is not where campaigns fail. The offer, the funnel and the tracking decide the outcome, and all three sit outside Ads Manager.
An audit of the offer and the funnel, a rebuild of Meta Pixel and Conversions API tracking, and a written statement of what is being optimised for. Days rather than weeks, but never zero.
Ads live on day one. That is a promise about speed, not about outcome.
Every reporting number you will ever see is an estimate under an attribution window. Pulling one of my own campaigns through the Meta API returned 204 leads for an ad that Ads Manager reports as 196. That is a four percent gap on the same ad on the same day.
A plain explanation of the attribution setting in use, and a warning up front that the platform number will not match your CRM or your bank statement exactly.
Reporting that matches your internal numbers perfectly. Someone is copying one into the other.
Meta will sell you the cheapest version of whatever you ask for. Conversations started, profile visits and form fills are advertising results, not sales. In one of my accounts a profile visit costs six cents and a messaging conversation costs 1.36 dollars, and adding those two numbers together would be meaningless.
A named objective, the reason it fits your business, and an honest account of what has to happen after the ad does its job.
A dashboard full of cheap results with no line connecting them to revenue. Cheap leads are easy. Cheap leads that buy are the job.
A screenshot without a date range is not evidence. Neither is a monthly summary that quietly moves the window whenever a different window flatters.
A fixed reporting cadence, an account review more often than the report, and figures that always carry the period they cover.
Reports that arrive only when you chase them, or numbers with no date attached to them.
An agency that says yes to everything is either lying or spread so thin that nothing is deep. I run one channel and I say so on the first call.
A short, specific list. Channels they do not run, industries they will not take, situations where they hand the money back.
“We do everything.” Then ask who personally will be inside your account each day, and watch how long the answer takes.
Paid traffic is an amplifier. If an offer does not convert warm traffic, colder traffic will only lose money faster, and it will look like the ads failed.
A direct answer about your specific offer, including the possibility that it is not ready yet. If the funnel or the offer are not there, the honest move is to say so before taking anyone's money.
No conditions at all. Everyone can be helped, apparently, which is a sales position rather than an assessment.
It will. Every winner decays, and the account either has a queue of tested replacements or it has a panic.
A description of the testing structure that runs alongside the winners, and what share of the budget it holds.
“We will make a new one when that happens.” By then you have spent a month at a worse cost per result.
A result with no window attached can be any window at all. Every table on my results page prints the exact period its screenshot covers, and the numbers in the text are the numbers visible in the picture.
When an account runs two objectives side by side, Meta prints a dash instead of a total, because a total would be adding conversations to profile visits. An agency that shows you a clean total in that situation typed it themselves.
Ads sorted by result show winners at the top and near zeroes underneath. That shape is normal and it is the proof that testing happened. A screenshot cropped to three good rows tells you nothing about the account.
Accounts do not fail politely inside business hours. The answer to this one tells you the real size of the team.
If you want to see how I answer my own ten questions, the account screenshots are on the results page and the rest is a conversation. Write to info@aleksaadvision.com.