DCM Becomes Agent Orchestrator: The Next Generation of Campaign Management

About twelve months ago, inside the opua brand family, we asked ourselves a strategic question that subsequently shifted the roadmap picture of the Digital Campaign Manager. If marketing operations are currently shifting from a tool-operator world to a steering world, then there must be a place where this steering actually happens. Today, in most marketing teams, this place is a mix of Excel reports, Slack threads, media-plan reviews, and an inbox full of escalated anomalies. This makeshift infrastructure still works in the current state, where campaigns are set up weekly. It no longer works in the target state, where agents make decisions in real-time feeds. We have therefore decided to extend the Digital Campaign Manager — DCM for short — from a campaign tool into an agent orchestrator. This extension is not a marketing label on an existing product. It is an architectural shift that adds four new functions and redefines the existing DCM identity as a cockpit for steering.

Before describing the four new functions, it is worth briefly recalling what DCM has been so far. DCM is a multi-channel campaign manager with marketing-mix-modeling integration, consolidating performance data from Meta Ads, Google Ads, GA4, and other sources. The product emerged with a clear focus on the Swiss mid-market and mid-sized performance departments — teams too small for their own MarTech department but too large for pure self-service tools. DCM offers KPI consolidation, anomaly detection, cross-channel reporting, and a direct connection to the strategic MMM backbone for allocation recommendations. This identity remains. What changes is the second function DCM now additionally takes on: cockpit for the steering of agentic marketing operations. Both functions live in the same product because the same data — channel performance, MMM recommendations, anomalies — forms the foundation for both.

The first new function is the machine-readable KPI hierarchy. A KPI hierarchy is the structured arrangement of goals, trade-offs, and tie-breakers with which a marketing lead describes their steering logic. So far, this arrangement has lived in most teams in the head of the marketing lead, at best in a strategy deck. For human tool operators that was sufficient. For agents it is not sufficient, because an agent needs a data structure, not a slide. DCM therefore allows explicit modeling of the KPI hierarchy as a data object. The primary goal might be cost-per-qualified-lead, the secondary goal brand lift, the trade-off rule 'if CPQL rises more than twenty percent, cut brand spend in favor of performance spend', the tie-breaker 'at parity, the highest incrementality decides'. This hierarchy is configured once in the DCM cockpit and automatically forwarded to all connected agents. What used to be negotiated in the marketing meeting now becomes code.

The second new function is the guardrails cockpit. Guardrails are the explicit boundaries inside which an agent may decide autonomously. DCM bundles them in three classes. Decision authority — which kinds of decisions may the agent take independently, which must be escalated. Spend authority — which budget corridors may they shift without human approval, from which threshold does approval become required. Brand-safety corridors — which inventory categories are admitted, which excluded, which conditionally usable. These three classes are maintained in the DCM cockpit, continuously checked by the system against agent behavior, and forwarded to escalation workflows in case of breach. A marketing lead who maintains the cockpit for one hour per week has a better overview of the risk profile than a team working manually through anomaly emails every day.

The third new function is the translation of measurement signals into real-time feeds. So far, the weekly MMM update landed as a PDF in Sharepoint and had to be translated by the performance team into DSP configurations. In the new DCM architecture, MMM outputs flow as AMDP payload — agent-marketing-data-protocol — directly into the cockpit and from there onward to the connected agents. A changed channel mix from MMM-Wizard reaches the buy-side agent within minutes, not within days. A changed brand-lift insight from AiCMO modifies the pacing strategy automatically, the moment it is validated. What used to be a weekly report becomes the steering logic of the operational stack. The marketer remains responsible for reviewing outputs and checking bias risks — but translating them into operational action is no longer their job.

The fourth new function is the audit-trail viewer. Every agent decision that DCM coordinates leaves an audit trail — who decided what, when, why, against which KPI hierarchy, on which data basis, inside which guardrails. These audit trails become visible in the DCM cockpit as a time-sorted stream and can be filtered by agent, channel, brand, or decision type. Behind the scenes, the individual entries are signed with SHA-256 audit hashes in canonical form, forming a complete reproducibility chain. For a marketing office operating in a regulated industry, this is the structural precondition for being allowed to use agentic operations at all. Compliance officers can reconstruct concrete decisions bit-identically three months later. Internal audits can rebuild the decision logic from the stored KPI hierarchies. External regulators receive a depth of documentation no US-based SaaS vendor delivers in comparable form.

The four functions together form the cockpit for the steering role. The question that follows is which agents DCM actually coordinates today. In the current roadmap state, four productive agentic integrations are foreseen, all inside the opua brand family. Nexbid is the sell-side agent stack with verified auctions and the Verified Agent Badge as a trust surface for publishers. AiCMO is the AI-citation-tracking agent, monitoring where and how a brand appears in the outputs of the major LLM systems. Pruefstand is the verify agent, running automated output reviews and bias audits and feeding them back into DCM. MMM-Wizard is the strategic attribution backbone, not itself agentic, but essential as a data source for all three agents. These four components are not assembled from four different vendors. They are developed in the same brand family, with the same mathematical DNA in Lean 4, the same audit-hash standard, and the same governance logic.

Strategically this produces an architecture we internally call the four-tier stack. DCM forms tier one, the tactical steering with last-click attribution and multi-touch tracking in real time. MMM-Wizard is tier two, the strategic backbone with Bayesian inference, adstock, and saturation. Nexbid stands for tier three, the agentic activation through buy-side and sell-side protocols. AiCMO delivers tier four, the AI-citation attribution for the new visibility in LLM outputs. The four tiers are clearly separated by question, timescale, and data level. But they share the same audit-hash standard, the same Lean 4 DNA, and the same AMDP protocol as a binding fabric. Anyone buying a marketing stack today assembled from four different US-based SaaS vendors will discover in eighteen months that the audit trails do not align, the governance logics contradict each other, and the bias definitions are not translatable. With us, the mathematical core is the same across all four layers.

What pilot customers can expect in Q3 2026 is a limited beta with selected brands from insurance, banking, and pharma — industries with the highest compliance and audit requirements, and therefore the strongest fit for the architecture's USP. The KPI hierarchy function and the guardrails cockpit are productively ready. The real-time feed translation runs in its first iteration for the AMDP connection to Nexbid and MMM-Wizard, with AiCMO and Pruefstand following in subsequent sprints. The audit-trail viewer is feature-complete in its base functionality and will be extended in the pilot by industry-specific audit templates. What is not included in the pilot and is scheduled for 2027 is direct connection to third-party agents outside the opua brand family — such as Adform buyer agents or Pacvue commerce agents. These integrations are technically in the roadmap, but deliberately out of scope for the first pilot, because control over the own stack is the precondition for clean testing.

Anyone wanting to test the DCM Agent Orchestrator as a pilot customer can sign up at demo@digital-opua.ch for the pre-beta phase. The pilot capacity is deliberately small because we accompany the first implementations closely — typically with two workshop days for setup, a two-week configuration phase, and a thirty-day monitoring phase with weekly review sessions. Anyone wanting first to check whether their own stack is ready for DCM at all can request a two-hour setup audit at audit@digital-opua.ch, which analyzes KPI hierarchy, measurement signals, and bias risks in the current setup. And anyone wanting to understand the technical foundation of DCM more deeply can find the open-source components of the agentic integration — that is, the Lean 4 theorems of Nexbid and MMM-Wizard verification — at github.com/nexbid-dev/protocol-commerce, licensed under MIT. The shift from campaign manager to agent orchestrator is not the last word on the future of the marketing tool. But it is the next necessary stage.