A framework for reading an enterprise marketing stack the way an economist would. Three layers, one direction of flow, and the leaks that explain why so much current spend evaporates.
Every enterprise marketing function in Europe is now running AI pilots. Most CMOs can list a dozen tools they have bought in the last eighteen months. Few can point to a durable improvement in unit economics, customer lifetime value, or cost of acquisition that those tools have produced.
Adobe has a new CEO. Salesforce is rewriting its pitch around Agentforce. Every vendor at MWC and CES is shipping an agentic story. The category is moving at speed. And yet, if you sit with a CFO and ask what has actually changed on the P&L, the answer is usually the same shrug you got in 2022.
The standard explanation is execution. Data quality, change management, organisational resistance. Those factors are real, but they describe symptoms, not the root cause. The root cause is structural. Enterprises are investing heavily in AI capability without a clear model for where durable value from that investment actually accumulates, and where it evaporates.
Value in a marketing stack does not accumulate evenly. It pools in specific layers. The tools most enterprises are buying do not sit in those layers, which is why the boardroom feels the spend curve climbing while the value curve stays flat.
Since GPT-4 dropped, vendor-led narrative has trained enterprise buyers to look for value in the most visible part of the stack, which is also the part that commoditises fastest.
The layer that looks like the product, the chatbots, the content generators, the experience orchestrators, is the layer where every competitor eventually has the same thing. Whatever advantage exists there has a half-life measured in months. The cost of accessing a capable generative model, a predictive engine, or an autonomous agent is falling toward zero. Differentiation at that level is an event, not a position.
The layer that actually holds durable economic value sits underneath it. The data architecture, the identity graph, the consent and governance logic, the audit-grade workflow scaffolding. These are harder to buy off a shelf and harder to change once installed, which is exactly why they matter. Value that reaches this layer is sticky. Switching costs are highest here, not because the technology is unique but because the organisational context, the data models, the workflow integrations, and the compliance dependencies embedded around it are extremely costly to replicate elsewhere.
Adobe, Salesforce, and Microsoft compete intensely to own this layer precisely because of its gravitational pull. Their most strategically significant recent investments are not foundation models. They are Real-Time CDP, Data Cloud, and Microsoft Fabric. The platform contest moved.
Any point in a revenue architecture where AI-generated value fails to flow downward and become embedded in the governed, high-switching-cost foundation, causing it to evaporate, commoditise, or remain trapped in fragile, low-defensibility pilots. Gravity Leaks are the single most expensive failure mode in an enterprise marketing stack, and they are almost always invisible until a platform migration or compliance audit reveals how little of the AI investment actually compounded.
Value Gravity™ is the lens we use to explain that asymmetry. It is not a maturity model. It is a diagnostic. You point it at an existing stack and it tells you where the economics of your next decision actually live.
Economic value accretes downward through the stack. The deeper a layer sits, the more durable the value it holds and the harder it is to replace.
Value Gravity is not a taxonomy of tools. It is a claim about economic forces operating across the layers simultaneously.
AI capability at the top commoditises fast. Whatever is a competitive advantage in Q1 is a line item on every vendor roadmap by Q3. The cost of accessing capable models is falling toward zero. Differentiation at this layer is measured in months, not years.
Switching costs, consent regimes, and governance obligations all sit at the base. The deeper a capability lives in the stack, the harder it is to replace. This is not a technology lock-in argument. It is an organisational context argument. The value of a well-governed CRM estate is not in the software, it is in the decade of customer data and workflow logic built around it.
Identity graphs and consented data compound in value over time. Every properly governed AI activation, scored, attributed, audited, makes the foundation more defensible. The top layer consumes that weight. The base produces and holds it, which is why good foundations pay back over years, not quarters.
These are not predictions about emerging trends. They are the structural consequences of the gravity model that show up in enterprise commercial stacks today.
The race to build the best foundation model is already narrowing. Capability differences between major models are converging faster than most enterprise roadmaps anticipated. The strategic contest that matters for commercial stacks is different. It is about who controls the governed context plane where customer identity, buying-group history, compliance status, and orchestrated data flows live.
Adobe, Salesforce, and Microsoft understand this. Their most strategically significant recent investments are not in foundation models. They are in the base layer: Adobe Real-Time CDP, Salesforce Data Cloud, Microsoft Fabric. These are the investments that build switching costs, not just capability. Enterprise leaders who evaluate AI spend primarily by what new capabilities it unlocks are optimising for the layer where value is shortest-lived.
The investments that generate the most boardroom excitement, the pilots, the copilots, the generative content tools, have the shortest value half-life. The investments with the lowest short-term visibility, the identity resolution architecture, the governed data flows, the audit trail infrastructure, have the longest compounding returns.
Organisations that measure AI returns quarterly will systematically underinvest in the base layer and accumulate a growing inventory of high-capability, low-defensibility experiments. The correction, when it comes, is usually painful. It is triggered by a governance incident, a compliance audit, or a platform migration that suddenly reveals how little of the AI investment actually compounded into the commercial foundation.
Enterprise organisations have a growing inventory of impressive AI proofs of concept that never became production capability. The standard explanation is change management, data quality, or organisational resistance. The more precise explanation is that most pilots are designed to demonstrate the capability of the tool, not to test whether value can be embedded at the base layer.
A pilot that generates high-quality predictive scores that never land as persistent, governed records is not a change management failure. It is a value gravity failure. The intervention that converts experiments into durable returns is not better adoption planning. It is redesigning the pilot success criteria around a single question: does this compound in our governed commercial foundation?
Value Gravity™ is the methodology. IDADAY International Services is the practice that uses it.
IDADAY works with a small number of enterprise marketing and data leaders in the Benelux and wider Europe on the decisions that sit at the foundation layer. Data architecture, identity and consent design, measurement models, and the governance scaffolding that keeps all three coherent. The Gravity Scan is one of the instruments the practice uses, not a standalone product.
If any of the argument on this page has reframed how you are thinking about your own stack, the most useful next step is usually a conversation.