What is included
- Meta ads (Facebook/Instagram): awareness, retargeting, Advantage+ Shopping
- LinkedIn ads for B2B decision-makers: thought leader, document & lead-gen ads
- Creative testing culture: hypotheses, variants, recurring rhythms
- Meta Conversion API (CAPI) & server-side tracking for clean data
- Audience research: lookalikes, customer match, job targeting (LinkedIn)
- Every campaign wired straight into CRM & email journey
Cost savings & workload
Growth in the feed – the effort runs in the system.
Cost savings through automation
Creative fatigue is caught before it costs money
Automated creative rotation and budget steering reduce wasted ad spend: creative fatigue is caught and countered before it burns the budget. Every euro of the campaign works on what is still performing.
Workload reduction
Campaigns run on workflows
Campaign setup, A/B tests and reports run on workflows. Your team makes decisions instead of clicks – and spends its time on creatives, not on the ad manager.
Effect size depends on starting point, team size and tool stack – the concrete figure we name in the audit.
Optional: in 24/7 operation, driven by agents
This service can be operated continuously by an agent stack: budget, data, tests and delivery run on autopilot and your team shrinks to strategy and decisions. An agent stack is an option with us, not a must – we deliver the same service fully manually or in hybrid mode as well. Details below.
Autonomous agent organisation · Paid Social
Paid Social Orchestrator
Creative, audience and budget – in a permanent 24/7 test.
Paid social lives on fresh creatives and precise audience logic – exactly what can be automated. The paid social orchestrator coordinates agents for creative, audience, budget, pacing, anomalies and A/B testing across Meta, LinkedIn & co. The day-to-day operation (launching ads, pausing them, shifting budget, rotating formats) happens without manual input – your team only defines audience, brand and budget frames.
6 specialists on 24/7 duty – each accountable for one KPI
Creative Agent
● runs 24/7
KPI: fatigue score
Audience Agent
● runs 24/7
KPI: CAC per audience
Budget Agent
● runs 24/7
KPI: blended CAC
Pacing Agent
● runs 24/7
KPI: learning-phase status
Anomaly Agent
● runs 24/7
KPI: policy error rate
A/B Test Agent
● runs 24/7
KPI: test coverage
all results flow in
new rules back into the agent stack – the cycle starts again
How to read the diagram: at the top, the orchestrator – it decides which actions run. Below, the agents work 24/7, each accountable for exactly one KPI. All results flow down into the self-learning loop (01 Measure → 02 Analyse → 03 Learn → 04 Apply). The new rules that emerge return into the stack via the left line – so the system gets better every week without anyone fine-tuning by hand.
The agent stack in detail
Every agent works 24/7 in the background: it consumes specific data, produces decisions or content from it and is accountable for exactly one metric.
Creative Agent
runs 24/7Produces new creatives based on proven patterns, rotates out fatigued losers and keeps 3–5 active variants per campaign.
- consumes
- fatigue scores, hook test results, brand guidelines
- produces
- new video and static creatives, 3–5 active variants per campaign
- KPI
- fatigue score
Audience Agent
runs 24/7Segments audiences, builds lookalikes and exclusion logic, tests new audience hypotheses against benchmarks.
- consumes
- conversion and lookalike data, market research
- produces
- new segments, lookalikes and exclusion rules
- KPI
- CAC per audience
Budget Agent
runs 24/7Shifts budget daily to the ad sets with the best CAC/ROAS – within a channel and between Meta/LinkedIn.
- consumes
- CAC/ROAS per ad set, channel benchmarks (Meta, LinkedIn)
- produces
- budget shifts to the ad sets with the best CAC
- KPI
- blended CAC
Pacing Agent
runs 24/7Keeps learning phases stable: no budget jumps that push ad sets out of learning, clean fallback logic when limits hit.
- consumes
- learning status, budget limits, delivery curves
- produces
- stable ad-set launches without breaking the learning phase
- KPI
- learning-phase status
Anomaly Agent
runs 24/7Detects policy violations, delivery drops, CPM spikes and tracking errors in real time – with a concrete fix path.
- consumes
- policy notices, CPM curves, CAPI and tracking events
- produces
- real-time alerts with a concrete fix path
- KPI
- policy error rate
A/B Test Agent
runs 24/7Permanently tests formats (video/static/carousel), hooks, CTAs and audiences – significant winners automatically move into scale ad sets.
- consumes
- test backlog, significance data, format benchmarks
- produces
- continuous tests of formats, hooks and audiences – winners to scale
- KPI
- test coverage
Automated A/B tests
A/B tests run automatically: formats, hooks, CTAs and audience variants are tested continuously. The test agent stops losers once significant, scales winners into dedicated scale ad sets and starts the next test round – social budget works on new variants every single day.
Autonomy levels: who may decide what?
Autonomous – immediate
Bid adjustments within approved frames, budget shifts under threshold, new creatives from the test inventory, negative keywords. The agent executes and documents afterwards.
Orchestrator decision
Channel recommendations, test launches, re-priorisation, rule updates from the learning loop. The orchestrator compares agent recommendations, checks them against frames and decides.
Human approval
Budget increases above threshold, campaign pauses, new audiences, price and promise changes. These actions only run after approval – visible in the dashboard.
The orchestrator’s own duties
The paid social orchestrator steers the creative pipeline: it gives the creative agent patterns and brand boundaries, sets test plans for formats and hooks, distributes budget between Meta, LinkedIn & co. by CAC – and automatically blocks any ad-set launch that violates approved audiences or budget frames.
Orchestrator live dashboard
- CAC per audience
- CPM & frequency trend
- Creative fatigue score
- Learning status per ad set
- Policy warnings
- Scale factor (scale ad sets)
The orchestrator checks these metrics hourly – deviations land in the task queue or become an approval request for the team.
Human-in-the-loop: approval required
- New creatives with brand claims
- Budget > approved daily frame
- Audience change > 30%
- Scaling into a new country/market
The agents work autonomously – up to this point. The orchestrator executes these actions only after approval. That is the difference between automation and risk.
Self-learning: the 4-step cycle behind the org chart
From every creative-fatigue pattern, every hook A/B test round and every CAC trend per audience, the orchestrator learns which formats, hooks and messages work in which market – and sharpens the creative and audience agent briefings from that: today’s creatives are tomorrow’s learning patterns.
The self-learning loop turns permanently – the agent stack gets better week by week without anyone fine-tuning by hand:
Measure
All results flow in: conversions, costs, delivery rates, rankings, engagement. Without clean measurement an agent learns nothing – which is why the tracking setup is the foundation of every stack.
Analyse
The orchestrator compares results against targets and history, separates causality from coincidence (significance) and recognises which changes really had impact.
Learn
From that, new rules emerge: score weights, bid margins, fatigue thresholds, segments. The orchestrator checks every rule against the approved budget and risk frames.
Apply
The new rules enter the agent policies – from the next round, all agents work with the updated knowledge. The cycle starts again: measure, analyse, learn, apply.
Fewer heads, more output: agents take the routine – your team only steers strategy and budget frames.
Agents are an offer, not a must – you choose the automation level
We deliver every one of these services fully manually as well. The agent stack is one option – and you can switch between the modes at any time without losing your setup or your results.
Fully manual
Classic service: our operational team executes every action personally – setup, operation, reporting. The same quality, fully human operation, maximum control.
Hybrid
Agents take routine, monitoring and 24/7 continuous testing; your team and our operational team decide, prioritise and fine-tune. The most chosen mode.
Autonomous with approvals
The agent stack runs fully through – critical actions (budget, audiences, promises) only after your approval in the dashboard. Maximum output, clear accountability.
All three modes use the same systems architecture – you can switch between them at any stage.
A look into the 24/7 operation: our agent dashboard
This is what operation looks like when this service stack runs on autopilot: orchestrator, KPIs, running A/B tests and your approvals – as an interactive preview with sample data.
Who this service is for
Mid-market with a clear growth goal
20–500 employees who want to turn more traffic into more qualified leads and more closes – without overwhelming sales.
Teams juggling several scattered tools
CRM, email, ads and analytics run side by side. We bring them into one system with a single source of truth.
Companies in DACH + international markets
Same systems architecture, local market logic for Italy, Spain, France, the UK – measured in one report.
How the delivery runs
- 01
Audit & target picture
We analyse data quality, existing processes and KPIs – honestly, with numbers. The result: a shared target picture and a prioritised roadmap.
- 02
Build in milestones
Implementation in clearly prioritised stages – each with a measurable goal. You always see what is going live and what it delivers.
- 03
Operation & scaling
We observe, test and optimise continuously. Monthly reporting, clear KPIs, no black box – scaling follows the numbers, not gut feeling.
Typical starting situations we solve
These patterns keep coming up in our audits. If one of them fits, a conversation is almost always worth it.
Leads come in, but nobody knows if they are good – no scoring, no priority.
The CRM is half-set-up and nobody dares to touch it – the data is incomplete.
Emails are sent, but delivery, open and click rates are weak.
Tracking is broken or missing – attribution is guesswork instead of facts.
What we deliver – concretely
Measurable KPI roadmap
Clear goals per milestone – from lead source to revenue. We name the numbers that should change before we start.
GDPR, consent & data hygiene
Compliance is part of every setup: consent, deletion concepts, clean data – no retrofitting after the data-protection audit.
Documentation & handover
Everything documented: setups, workflows, logic. Your team can keep working independently after the collaboration – no lock-in.
Frequently asked questions
Meta or LinkedIn – which channel fits us?
B2C/e-commerce and broad B2B top-of-funnel: Meta (more volume, cheaper CPM). High-price B2B with decision-maker targeting: LinkedIn (higher CPM, more precise audiences). We calculate both before the start – the budget split follows the intent of your audience.
What does a click cost with Meta/LinkedIn ads?
CPC depends heavily on keyword, industry and auction – so we do not think in click prices but in CAC across the entire journey. A cheap click without follow-up is still expensive. We prioritise keywords by CPA and revenue impact, not by the cheapest click.
Do we need our own creatives for paid social?
We deliver structure, hypotheses and test design. Production can be done by your team or a partner we bring in. What matters is a recurring rhythm of new variants – ads fatigue, fresh creatives keep performance up.
What is the difference between paid social and performance marketing?
Performance marketing is the umbrella term (ROAS-driven advertising across all channels). Paid social is the social part of it (Meta, LinkedIn, TikTok). For us both run as one system: shared tracking base, shared landing pages, shared journeys – separate channels, one data flow.
How quickly do we see results from paid social?
First signals usually after 2–4 weeks, a reliable ROAS/CAC decision after about 6–8 weeks. Retargeting streams often deliver earlier, since the audience is already familiar.
What does Paid Social cost with Core Online Marketing?
It depends on scope, tools and data quality. In the first call you get a concrete range and a milestone roadmap – no strings attached. Ongoing operation is available as a monthly package, projects as a fixed price.
How quickly will we see first results from Paid Social?
Usually after 4–8 weeks, depending on the starting point. In the audit we define the KPIs that should change, and we report monthly and transparently – what is better, what is unchanged, what is worse.
Do you work with our existing tool stack?
Usually yes – we are tool-agnostic. If an existing tool makes sense, we keep it; if not, we advise on switching. Partner discounts do not drive recommendations.
Do we need dedicated staff on our side for this?
One contact person for decisions is enough. We handle build and operation and document everything, so your team can continue independently after the collaboration.
Are you set up for international markets (e.g. Italy, Spain, the UK)?
Yes. We are based in Berlin and work with local performance leads for Italy, Spain, France and the UK – the same systems architecture, local market logic, reporting in one currency: your KPIs.
Do you also run the ongoing operation, or just the setup?
Both is possible. Many clients start with the setup and take over the operation themselves afterwards – we document and train. Others let us run the whole operation. In the first call we say honestly what serves you better in your situation.
How do you handle legacy data and historical content?
We start with a data audit: what is usable, what needs to be cleaned, what can be deleted. Historical content is mapped and migrated where it is worth it – not blindly. You start clean, without inheriting debt.
What concrete cost savings and workload reduction does Paid Social bring?
Automated creative rotation and budget steering reduce wasted ad spend: creative fatigue is caught and countered before it burns the budget. Every euro of the campaign works on what is still performing. On the workload side: Campaign setup, A/B tests and reports run on workflows. Your team makes decisions instead of clicks – and spends its time on creatives, not on the ad manager. We show the concrete calculation for your setup in the first call.
Paid Social: Let us calculate what it is worth for you
30 minutes, no strings. We look at your numbers and show where the lever is.
Request a call30 minutes, no strings. No pitch – we look at your data.