Meta Ads is the platform where activity is most often mistaken for results. A low CPA in the dashboard can hide inflated attribution, saturated audiences and creative that only converts remarketing traffic. As a Meta Ads agency, our work starts by separating what the platform says from what the business confirms.
To start from the data: the dead spend auditor reads your Ads Manager export and flags the budget that hasn't produced any result, without the file ever leaving your browser.
Measurement first: CAPI and real attribution
Since iOS 14.5, Meta models more than it measures. That is why the first thing we set up is the server-side Conversions API (CAPI), deduplicated against the pixel and connected to your source of truth (Shopify, CRM, backend). Without CAPI implemented properly, the algorithm optimises on partial data and scaling decisions are made blind.
We also use 7-day click attribution as the main reference and always cross-check it against business data (actual sales, margin, P&L), so the ROAS we report is the one that appears in your income statement, not the one Meta wants to show you.
Campaign structure with a rationale
We don't stack campaigns. We work with a clear structure that separates:
- Prospecting: broad audiences and lookalikes built on high-value data (buyers, SQLs), excluding existing customers.
- Remarketing: tight windows and distinct creative so it doesn't cannibalise prospecting.
- Advantage+ Shopping (where it applies): with control over the existing/new customer ratio and a budget cap to stop the algorithm settling into remarketing.
The structure depends on the business, not on a template. Ecommerce, lead gen and B2B have different architectures because the funnel is different.
Creative driven by data, not intuition
On Meta, the creative IS the targeting. That is why creative production can't be disconnected from the account. Our Nash Generative Lab analyses the performance of every asset (hook rate, hold rate, CTR, cost per result) and produces iterations based on what works, not on what people like internally.
Volume matters: Meta needs variants to find who each angle speaks to. But volume without analysis is noise. We produce variants systematically: new angles, iterations on winners and retiring fatigued creative based on data, not opinion.
Scaling without breaking things
Scaling on Meta isn't raising the budget on a Monday and hoping. Every significant budget increase resets the ad set's learning phase if it isn't handled properly. We work with controlled increases, duplicate ad sets to preserve learning and monitor the breakdown effect daily (when CPA rises as you scale, we know why and what to do about it).
And when Meta stops being the most efficient channel for your next euro, we tell you. We also manage Google Ads, LinkedIn Ads for B2B and TikTok Ads, and the mix is decided with data, not preferences.
Frequently asked questions
Do you only manage Facebook and Instagram, or WhatsApp and Messenger too?
We run campaigns on Facebook, Instagram, Audience Network and Messenger. We implement WhatsApp Ads (click-to-WhatsApp) when the sales cycle justifies it, mainly in lead gen and services.
What about attribution since iOS 14?
Server-side Conversions API and 7-day click attribution are the foundation. On top of that we cross-check against real business data to confirm that what Meta reports matches actual sales. Without that check, we don't scale.
How much budget do I need to start with Meta Ads?
It depends on the objective and the ticket size. In ecommerce, €3,000–5,000 a month already gives enough signal to optimise. In high-ticket B2B, it can work with less spend if the ICP is well defined. We size it for you in the audit.
Do you produce the creative, or do I need to supply it?
Either. If you have a creative team, we work with them and give them data-based briefs. If not, our Nash Generative Lab produces the assets (static, video, UGC-style) as part of the service.
Is your paid media a profitable system or just a pile of campaigns?
Tell us about your case and we return an actionable diagnostic: what to scale, what to keep and what to pause. Free, no strings attached.