Meta Ads

First we check the ad account, the page, the pixel, the catalogue and the events. We set up or repair the tracking of the key actions: product view, add to cart, checkout started, purchase or request. Where it's possible we connect server-side conversion tracking to cut down on lost data.

Then we build the campaign structure, launch the ads and keep an eye on budget, audiences, frequency and cost per result. We switch off combinations that spend money without producing the target action, and move the budget onto the ones backed up by requests or orders.

Once a week you get a report you can actually read: how much was spent, how many target actions came in, what they cost, what we changed and what we're checking next.

The structure depends on the job and on how much data there is. Usually we keep three things apart: reaching a new audience, bringing site visitors back, and dynamic product ads from the catalogue. That way you can see where a result came from, and people seeing the brand for the first time don't get mixed in with people who were already choosing a product.

At the start we check the events and pick the action Meta should actually be learning on: a purchase, a completed request or another confirmed action. Then we run a limited number of clear tests, so the budget isn't scattered across dozens of ad sets.

We make changes based on accumulated data. Frequent random edits restart the learning phase and make it impossible to tell what worked.

The client or their content team hands over finished photos, videos, banners, product information and current offers. We match the material to a specific audience and funnel stage, pair it with ad copy, a headline and a call to action, and launch several versions we can measure.

We judge by the data, not by personal taste: does the ad hold attention, do people click through to the site, and do they complete the target action. Weak versions get stopped, strong ones stay and get scaled as far as the budget allows.

When material burns out, or a test needs an angle, format or offer that doesn't exist yet, we give a specific brief for the next piece. Producing content stays with the client or their SMM team.

Management costs from $400 a month for one agreed channel; the exact fee depends on the number of markets, ad accounts, products, languages, funnels and the amount of analytics work.

The ad budget is not part of the management fee. The client attaches their own card and pays Meta directly for the impressions. We don't take the ad budget into our own account and we don't add a hidden commission to it.

Before launch we agree a monthly budget and an acceptable spending pace. Any increase is discussed separately, especially while a campaign is still in its learning phase.

The account, the pixel and the page stay yours — even if we part ways.

You handle the shoot: the client or their content team films the photos and video. We work with the finished raw material — editing, subtitles, voiceover where agreed, and preparing formats for upload to Meta.

If you need creative material prepared on an ongoing basis, there's a separate creative package — $350 a month. It includes:

• a script hook, an offer and ad copy;
• three short videos a month;
• two statics or carousels a month;
• editing, subtitles and voiceover where agreed;
• formats prepared for launch on Meta;
• one agreed round of revisions.

Anything beyond the package is billed per piece: a video is $100, a carousel is $60, a static is $40.

Management combined with the creative package is from $750 a month. The ad budget is still not included in it.

We'll need access to the Facebook page, the ad account, Business Manager, the pixel and the site. If none of that infrastructure exists yet, we create it on the client's side and hand every right over to the owner. Paying for ads needs a card attached.

We'll also need current prices, stock levels, where you sell, your margin or the order cost you can live with, delivery terms and finished ad materials. Telling us quickly when stock or offers change matters: advertising something you don't have burns budget and ruins the statistics.

On the client's side we need someone who signs off offers and tells us promptly about lead quality and actual sales.

Meta's counter can show events without tying them to real orders. The causes are usually a badly installed pixel, duplicate events, no Purchase event after payment, ad blockers, browser restrictions, cookies declined, a purchase from a different device, or an order placed by phone or in a messenger.

We check separately whether a purchase is being reported before the order is actually completed. Otherwise Meta learns on people who only got as far as the form and never finished. We match the account's events against the site's or the CRM's orders and find what's going missing or being counted twice.

An exact match between Meta and an accounting system usually isn't possible, because they attribute differently. Our job is to make the gap explainable and small enough to make decisions on.

The first few days are for checking that the data is coming through and for the initial learning. The first two weeks show a direction, but a small sample can still swing on a handful of orders. A more solid assessment usually comes after three or four weeks, or earlier if the campaign has already gathered enough target actions.

We don't look only at clicks and messages. We take into account real orders, what they cost, the revenue, cancellations and lead quality. If there isn't much data, we say plainly that the conclusion is still provisional.

An obviously dead combination can be stopped sooner. We're not going to wait out a month for the sake of a round number while ads spend budget with no sign of demand.

We don't name a guaranteed ROAS, a number of sales or an exact cost per lead before launch. Those depend on the product, the price, the season, the competition, the site, stock availability, how fast enquiries are handled and how much data the account holds.

What we can do is give market benchmarks, set control figures and agree in advance at what numbers we change or stop a campaign. We back decisions with data, not with a promise to “set it up once and let it run”.

Meta can reject ads, restrict categories and change how its algorithms work. We follow the rules, fix the causes we can reach and resubmit material for review, but the final decision is the platform's.