Software companies are trying to become services businesses. Services businesses are trying to become software companies. And both are running into the same wall: neither model, as traditionally built, can deliver measurable outcomes at scale with attractive economics. This is not a story about one side replacing the other. It's a structural convergence driven by two forces arriving at the same time.
First, the cost of intelligence has collapsed. Capabilities that required dedicated teams and enterprise budgets just a few years ago can now be embedded directly into scalable infrastructure. What was once gated by resources is now gated by architecture and willingness.
Second, the locus of value has shifted from execution to decision intelligence. Across professional services, the differentiator is no longer how well you do the work. It's whether you know which work to do, and whether you can prove it mattered. Clients aren't paying premiums for effort anymore. They're paying for the quality of the decisions embedded in the delivery.
These two forces meeting simultaneously changes the economics of who creates value and who captures it. And nowhere is that collision more visible than in marketing, in the space between agencies and the software companies that serve them.
The agency model has real strengths that are easy to undervalue from the outside: strategic judgment, creative instinct, client intimacy, and domain expertise. In a world increasingly flooded with automated output, those human capabilities matter more, not less. But the operating model those strengths sit inside was built for a different era.
Agencies sell time. They scale through headcount. Costs move largely in line with revenue. Even well-run firms hit margin ceilings that are structurally constrained by labor intensity. Retention is fragile because switching costs are low, and every new CMO or budget cycle resets the relationship.
The deeper issue isn't talent. It's that the model makes strategic depth non-scalable. The work that actually retains clients is the decision intelligence layer: knowing what's working, proving it matters, reallocating accordingly. That's exactly the work that gets crowded out when teams are drowning in operational execution.
Many agencies have tried to close the gap by adding analytics teams, partnering with measurement vendors, or layering on new tools. But these remain adjacent to the core delivery model rather than native to it. The underlying economics don't change.
Software companies face the mirror image of this problem, and it's just as structural.
They've built the infrastructure that makes sophisticated marketing intelligence possible: CDPs, measurement platforms, AI orchestration layers. But they remain on the tool side of the value chain, selling capabilities while someone else captures the strategic value those tools enable.
The best software platform in the world creates no impact if insights aren't translated into context-specific decisions. And most aren't. Staying in the tool layer means someone else captures the value your infrastructure enables, or no one acts on the insights at all and the tool itself gets blamed for poor results.
Software companies have historically underinvested in the contextual, consultative layers that make their own technology effective. Many have treated implementation and strategy as the customer's problem. That's left them exposed to churn and commoditization in exactly the same way agencies are exposed by their lack of scalable infrastructure.
Both models are incomplete. Agencies have the relationship and judgment but lack scalable infrastructure at the core of delivery. Software companies have the infrastructure but lack the embedded context and decision-making layer. And clients, who increasingly demand decision intelligence, are unwilling to accept either half.
This is why a hybrid model is emerging, often described as "services as software." Services are becoming automated and systematized. Software is becoming more humanized and outcome-oriented. AI is what makes both possible: service delivery with software-like economics that still retains the strategic depth clients actually value.
At an operational level, that convergence resolves into four layers. Each one addresses a structural weakness that exists on both sides of the traditional divide.
A High-Fidelity Data Foundation
Everything begins with signal integrity.
In ecommerce, that means a purpose-built first-party data layer, often a CDP, that captures behavioral and transactional data, resolves identity across touchpoints, and streams clean, real-time signals back into ad platforms.
As privacy constraints increase and platform reporting becomes more opaque, performance depends on the quality of the inputs. Without precise, real-time first-party data, optimization degrades and volatility rises. AI layered on fragmented signals only accelerates bad decisions.
This layer restores signal quality. It's the prerequisite for everything that follows.
An Embedded Measurement System
The second layer resolves fragmentation of truth.
Most CMOs operate in a world of competing narratives. Platform dashboards say one thing, analytics tools say another, and agencies reconcile them manually. Decision-making slows because no one can agree on what's actually working.
An integrated measurement stack combines multi-touch attribution, marketing mix modeling, and incrementality testing into a unified performance framework. Not quarterly reports, but always-on evaluation that both agencies and their software partners can operate against together.
The shift is from interpreting reports to operating against a shared, continuously updated system of record. That's what moves both sides from selling effort or selling tools to selling validated decisions.
AI-Orchestrated Insight and Action
Measurement alone doesn't change economics. Action does.
The third layer embeds AI directly into delivery workflows: campaign-level diagnostics, anomaly detection, creative fatigue analysis, incrementality monitoring, budget reallocation triggers. Instead of generating abstract insights that sit in dashboards, AI continuously surfaces signals that are ready to act on.
This absorbs the analytical volume that traditionally consumes senior time on the agency side while solving the "last mile" problem on the software side. Human expertise shifts upward, toward prioritization, experimentation design, and strategic trade-offs that neither tool nor automation can replace.
AI, in this model, doesn't commoditize either side. It increases the leverage of human judgment while giving infrastructure a pathway to outcomes.
A Cross-Merchant Intelligence Layer
At scale, a fourth layer emerges: aggregated, anonymized intelligence across merchants.
This enables benchmarking beyond platform metrics, early detection of category-level shifts, and predictive performance signals that no single brand can see in isolation.
It's not required to operate. But it's what allows the model to compound. Over time, advantage shifts from individual account optimization to contextual market intelligence.
I've spent over a decade across both sides of this. Agency environments first, then building SaaS as the AI wave accelerated, and now at Blotout, where we're building the data infrastructure and measurement layer for ecommerce. From this vantage point, the convergence doesn't feel theoretical. It shows up in how clients evaluate partners, how agencies are trying to evolve their delivery, and how software platforms are moving closer to outcomes.
The agencies that treat data infrastructure and AI orchestration as core to how they deliver will change both their economics and their client relationships. The software companies that find ways to get closer to outcomes, without assuming they can replace human strategic expertise, will unlock value the tool layer alone cannot capture. Both sides have to move.
The line between "agency" and "software company" is already blurring. What emerges on the other side will be shaped by those willing to combine infrastructure and expertise into something neither side has built alone. The competition isn't between agencies and software companies. It's between those who see the convergence now and those who'll spend the next few years wondering what changed.
