Meta Ads looks simple from inside Ads Manager. That is precisely what makes it expensive. Here is what a consultant actually does, and what to judge one on.
A Meta Ads consultant is not there to press buttons in Ads Manager, which anyone can do. They are there to decide what to measure, what to test, and above all when to stop. Since Andromeda made manual targeting largely redundant, most of the work has shifted to measurement and creative. A good consultant is recognised by what they refuse, not by what they promise.
A pixel alone loses part of your conversions. The Conversions API restores the signal server side and gives the algorithm something real to learn from.
On Meta, creative drives most of the performance. I build ads that speak to a precise situation rather than generic visuals resized ten ways.
I tell you when a campaign does not deserve more budget. Money running on an audience that does not convert costs more than switching it off.
Most people who contact me about Meta Ads have already tried it themselves. They created an account, boosted a post, spent a few hundred euros, and looked at the result without knowing what to make of it.
That is normal. Meta Ads Manager is designed to make it feel easy. Three clicks and your ad is live. That is true, and it is exactly where the problem starts.
This page covers what a Meta Ads consultant actually does, why the platform is trickier than it looks, and how to recognise someone serious when you are looking for one. Including if that someone is not me.
The difficulty is not technical. Creating a campaign takes ten minutes.
The difficulty is that you cannot see what is happening. You spend real money inside a black box that decides on its own who sees your ad, when, and at what price. You cannot debug it. You can only observe it.
Three things make that observation misleading.
The signal is incomplete. The Meta pixel is browser side tracking: blockers, tracking restrictions and cookie refusals eat part of it. So you never see all your conversions, and neither does the algorithm. It learns on data full of holes.
Learning needs volume. Meta documentation places the exit from the learning phase at around fifty optimization events per week per ad set. Below that, results stay unstable, and you make decisions on noise while believing you are reading a trend.
The platform shifts under your feet. The global Andromeda rollout, completed in April 2025, largely made manual interest targeting redundant, the very thing every tutorial used to teach. Matching now works from the signals in the creative itself. An account structured according to 2023 best practice is today structured backwards.
Not clicking in the interface. Anyone can click in the interface.
They are there first to decide what to measure. Before anything else: which event counts as success, at what moment, and what it is worth to your business. Without that decision you optimize a number that means nothing. A campaign producing form fills at three euros each is a disaster if none of those forms becomes a customer.
They are there to repair the signal. The Conversions API sends the event from your server rather than from the browser, restoring part of what the pixel loses. It is the most profitable and most systematically neglected building block, because it is invisible and it pleases nobody in a meeting.
They are there to produce creative, a lot of it. Since Andromeda reads the creative to find the audience, the ad itself has become the main lever. It is no longer a matter of one format resized ten ways, it is a matter of different angles for different situations.
And above all they are there to tell you to stop. That is the part nobody sells, and the one that pays best. A budget still running on an audience that does not convert costs more than switching it off. Someone who has never advised you to cut a campaign is not looking at your results, they are looking at their invoice.
Here is what I most often find when opening an existing account.
Half installed tracking. The pixel is there, the Conversions API is not, or both are there without deduplication and every purchase is counted twice. The account then shows beautiful results matching no real sale. It is the most frequent and most expensive failure, because it makes every subsequent decision wrong. A pixel that fires before the visitor has consented belongs to the same disorder, and my free Cookie Audit extension makes it visible in a minute.
A fragmented structure. Fifteen ad sets splitting the same budget and competing for the same people. Each stays below the learning threshold, so none of them learns. Consolidating into one or two sets is often the move that unlocks everything, and it costs nothing.
A single creative, resized. One idea in ten formats is not ten ads, it is one ad in ten sizes. The platform needs genuinely different propositions to work out who to talk to.
The market is full of people selling screenshots of return on ad spend. Here is what I would look at if I had to pick one.
They ask to see your numbers before talking about results. Without knowing your margin, average order value and conversion rate, no advertising target means anything. Someone announcing a return multiple before seeing your account is telling you a story, not a method.
They start with measurement, not creative. Less seductive, and the correct order. Optimizing campaigns whose results you measure badly is speeding without a speedometer.
They tell you what they will not do. A written scope with a list of exclusions beats ten promises. It is also what prevents disputes in month three.
They accept saying they do not know. Nobody knows the detail of how Andromeda works, Meta does not publish it. Anyone claiming otherwise is selling certainty, the most expensive and least useful product in marketing.
And they sometimes advise against advertising. If your offer is unclear or your landing page does not convert, putting money in front simply amplifies the problem. Advertising repairs nothing, it accelerates what already exists.
Five questions, and the answers will teach you more than a portfolio.
How do you measure a conversion, and how do you check the number is right? The answer must mention the Conversions API and deduplication. If it only mentions the pixel, walk away.
What minimum budget do you recommend, and why that amount? The right answer starts from your cost per conversion, not from a generic package.
What happens if it does not work after two months? An honest answer exists, and it contains the word stop.
Who produces the creative, and how much per month? Since Andromeda, this is the most decisive and most frequently dodged question.
What is not included in your price? If there is no list, it will be improvised later, at your expense.
I do not manage Google Ads. I worked on Google campaigns in previous roles, but I would rather name two areas of mastery than present myself as an expert in everything.
I promise no return on ad spend multiple before seeing your account, your offer and your margins. Anyone who does is selling a story.
And I do not do monthly retainer ad management. That is a deliberate choice: the result depends on a budget I do not control, and billing a recurring fee on that basis puts me in a position I find dishonest.
Measurement comes before ad budget, whatever the platform. If your conversions do not report correctly in Analytics, start with the GA4 and GTM tracking rebuild, which begins with a quick audit. It is the prerequisite for any serious optimization, on Meta and everywhere else.
If your question is really about the visibility you do not buy, the understanding SEO page explains how the other half of the subject works, and why it asks for patience where advertising asks for budget.
Paid advertising is not among my packaged services today, and that is deliberate: I would rather frame a scope I can hold to than sell a monthly retainer whose results depend on a budget I do not control. Meta Ads work happens case by case, after a conversation.
A review of structure, tracking and creative, with fixes ranked by impact.
The most profitable brick and the most often neglected. Without it, optimization stays guesswork.
Build, launch, iterate. Steering happens on business indicators, not platform indicators.
You need enough volume for the algorithm to exit the learning phase, otherwise results stay unstable. The threshold depends on your cost per conversion: the higher it is, the higher the ramp up budget must be. We calculate it together from your real figures rather than a generic amount.
Because the pixel is browser side tracking, so it is affected by ad blockers, tracking restrictions and cookie refusals. The Conversions API sends the event from your server, which restores part of the lost signal and improves how the algorithm learns.
Yes, but not with the same logic as B2C. In B2B the sales cycle is long and conversion volume is low, so we often optimize on an intermediate event and invest in the content that feeds the decision. It is lead generation mechanics, not immediate sales.
They ask to see your margins and conversion rate before announcing any target, they start with measurement rather than creative, and they tell you in writing what is not included. The most reliable sign is this one: someone who has never recommended you stop a campaign is not looking at your results.
Far less since the global Andromeda rollout completed in April 2025. Matching now works from the signals in the ad itself, image, text and context, rather than from hand stacked audiences. In practice the effort has moved from targeting to producing genuinely different creatives.
My specialty is organic search and Meta media buying. I have worked on Google campaigns in previous roles, but I prefer to state my two areas of real mastery rather than present myself as an expert in everything.