Structural Content

Serve the business, not engagement

6 April 2026

The biggest problem with content so far has been murky ROI

In most areas of a business, you generally know how the money you are spending is impacting specific business outcomes. When you add a sales headcount, you measure the return in extra revenue. When you buy a machine for the production line, you know exactly which role that machine plays in delivering the product. When it comes to content, the line from spend to business impact is usually not so clear. Outside of direct-response marketing content such as paid ads or email campaigns, where you can track immediate conversion rates, ROI gets murky. Which is a problem, as the majority of content companies produce today actually falls outside of digital marketing.¹ Across brand-building content, community content, PR content, thought leadership pieces, employer branding, customer education, internal comms, success is not typically measured in terms of specific business outcomes. It’s measured by proxy, in the form of “engagement”. With content usually being one of the bigger cost centers in companies, not knowing exactly what you are getting for your spend is all the more uncomfortable.²

The root cause is structural, not strategic

We have not had murky ROI in content for lack of trying. The root cause has been structural: Human limitations have prevented the content teams from accessing the more granular business needs their content could serve, forcing them to fall back to serving more broad, generic needs such as “build the brand”, or “drive engagement”.

The “business teams” are no content experts. They can’t know the full depth of what content can do for them, outside of generic use cases like “market my product”. As a result, they miss asking the content team to assist with more specific business needs where content could in fact have helped. Content requests like “help user research region west close user intelligence gaps on feature X through interactive content”, “reduce human error incidents during engineering process Y through targeted educational content” or “reduce the number of deals lost to specific objection Z by arming reps with localized objection-handling stories” rarely land on the content team’s desks. The content team, on the other hand, knows exactly what concrete business use cases content could in theory help with. However, they can not be integrated into the business deep enough to stay on top of all of the current specific needs they could potentially serve.³ Even with a live database of every business job-to-be-done (JTBD), human bandwidth is too limited to scan hundreds of simultaneous jobs across departments and filter for content opportunities.

However, the content teams have been hired to create content. So, they look elsewhere to find what kind of content to produce. Outside of direct-response marketing content, they are left to serve high-level business cases like “drive awareness”, “build the brand” and “build community”. With targets this big, it becomes difficult to track content’s actual business impact. How did campaign X contribute to driving brand awareness? And how did this brand awareness help the business? Hard to tell. Instead, content success is measured by proxy, in the form of engagement metrics. The more people viewed, liked and commented, the more successful the content is deemed to be. Companies pour money into social media teams and agencies to create “viral moments”. The buzz looks good. But how exactly the buzz helped the business remains unclear.

Generative AI alone couldn’t fix this

The emergence of generative AI and Large Language Models (LLMs) could help somewhat with this problem. Content strategists could now drag and drop much more business context into AI models to help them identify specific business opportunities their content could serve. In theory, if a company already had all their business JTBDs organized in a central database, strategists could have gone as far as dragging and dropping copies of this database into LLMs to identify concrete opportunities. However, you still needed to manually feed the LLM this context.

Enter agentic AI. With agentic AI, AI can now connect to databases and pull information itself to perform tasks and actions. And with that, there is no more excuse for missing a business JTBD content could help serve. We now have the opportunity to rid content of murky ROI.

Introducing Structural Content

Structural Content is content that serves specific internal business needs, functioning as a load-bearing component of a company’s business operations, rather than an extraneous activity layered on top of them. As such, it is always tied to quantifiable business outcomes. It is produced using a network of agentic and generative AI and supporting infrastructure to scan the full width of a company’s operational jobs-to-be-done (JTBDs) at regular intervals, identify opportunities for content to contribute, and trigger respective content requests for production and deployment.

Structural Content does for content what supply chain management did for manufacturing logistics: it replaces human scanning of an impossibly complex system — a company’s full set of ongoing JTBDs — with AI that monitors the full picture continuously and triggers demand signals.

The mechanism: A company’s JTBDs are maintained in a central live system, typically synced automatically with and/or translated from data in tools already in use (Asana, Salesforce, Jira, product roadmaps). Agentic AI with professional-grade content and marketing skills scans the jobs at set intervals to identify opportunities where content can help, and generates content job tickets in response. The content department makes go/no-go decisions, prioritizes and processes the job tickets through production, deployment and iteration. The content affects the business outcomes it has been triggered to serve, which in turn changes the JTBD backlog. Content becomes part of the narrative of a company’s progress towards its desired business outcomes.

The impact: Broad use cases like “drive brand awareness” are replaced with specific use cases like “Drive awareness among [specific segment] of [specific brand image] that is a prerequisite for entering the consideration phase for [specific product].” Brand building becomes the cumulative result of many targeted exercises, each serving a specific need, each measurable. In addition, the AI JTBD-scanning is likely to unlock new granular business use cases for content outside of marketing, in areas like user research, product development, production and human resources.

Across all use cases, every single piece of Structural Content is tied to a specific business outcome it is serving. Your content ROI is clear, per piece, across content categories, and across your entire content function. Content no longer serves engagement, it serves the business, by way of engagement. There is no more room for murky ROI, content has become pragmatic. And, with that, Structural Content opens the door to a host of additional AI content use cases.

Setting the stage for near-autonomous content operations

Knowing exactly which role content is to play allows for precise production, deployment and iteration instructions, which, in turn, unlock the ability to safely use AI a fair share more in the content lifecycle.

Have AI produce more content without risking “slop”: When what AI produces does not serve our needs, taste and preferences, or only does so at a shallow level, we perceive it as “slop”. Without precise instructions, AI can’t help but produce generic, “loveless” outputs. With Structural Content, each content job is bespoke and rich in context, as a defined part of the greater value chain. The inputs can now be precise and unique, and so the AI-generated outputs can be precise and unique. The risk of AI producing “slop” is greatly reduced, setting the stage for handing more and more content production over to AI with confidence.

Safely use AI to deploy content and run campaigns: Similarly, the rich context provides a dependable set of guardrails that allow for using AI to publish the content and run the marketing campaigns (more) autonomously. This can be a huge value unlock, as AI can tweak ad campaigns in real time 24/7, for levels of cost-efficiency humans could previously not achieve.

Use AI to self-optimize content: Structural Content also provides the basis for using AI to iterate on the content assets themselves. Knowing in concrete terms what content is to achieve, AI can compare expectation with reality, and run iteration loops tweaking content variables until the content asset performance vs. the intended ROI is maxed out.

Effective, reliable AI lives from clear instruction. Structural Content produces the foundation for clear instruction: being able to define exactly what the intended outcome is. And, with that, we’re closer to a future where AI can reliably run content operations near-autonomously.

What the content team becomes

The more autonomously AI runs content operations, the leaner content operations can become over time. The first jump step in efficiency gains comes when AI handles content production to the degree that a company no longer needs to rely on external vendors, such as agencies — cutting cost, churn and administrative bloat. The second big step up comes from no longer needing to spend staff bandwidth on production. The role of the in-house content team shifts from writing, recording, editing and deploying content themselves to using their domain expertise to engineer and tune the AI-driven machinery that produces these outputs. As time goes on, the parameters of this machinery become more and more dialed, and less operational human intervention is needed, allowing for leaner teams or the redistribution of content bandwidth to more high-impact work like bigger picture R&D and vision.

Caveats

Structural Content comes with some caveats.

Structural Content requires central digital organization of a company’s JTBDs. In reality, JTBDs often live partially offline, in people’s heads, or are strewn across various workflows, tools and databases. Resolving this might require larger organizational transformation. But, odds are, this is already in the works, as companies today recognize that visibility of JTBDs equals opportunity for AI to share the work.

Content production and deployment needs to be faster. The more granular the content need served, the higher the chances that it needs to be served quickly. “Brand building” is ongoing, whereas content equipping a sales rep to counter a current frequent objection can become outdated in weeks, days or hours. If not yet using AI to near-instantaneously generate content, early adopters may need to filter out jobs that are too granular to serve in time, until their production speed has caught up.

You can only measure content ROI as precisely as you are measuring the ROI of the business JTBDs it serves. Structural Content directly serves the desired outcome of the JTBDs it is responding to. When this outcome is not clearly defined, content’s contributions to the company also remain unclear. In this case, the company has larger success measurement issues to deal with. However, Structural Content’s main benefit remains untouched: it is now serving the company as precisely as other business functions. It no longer lags behind.

Structural Content might not be flashy. Pragmatic content impresses through results, not show. It is not for those who relish big content moments for the sake of the attention. It is for those who look at content as another opportunity to serve more value to their customers.

Feasibility and outlook

The tools and knowledge needed to implement Structural Content are already available today. Agentic AI frameworks trained on content expertise can periodically connect to project management software and other databases to extract business JTBDs, identify where content can help and produce — and prioritize — the Structural Content job queue. With the level of job specificity that comes with Structural Content, the capabilities of today’s generative AI suffice to have AI also produce the content assets. And companies have already switched to AI-deployment of the assets in campaigns. From there, the jump to AI also analyzing performance vs. business outcome and self-optimizing the content assets and the campaigns is a logical conclusion.

We’ve been making content not knowing exactly how it contributed to producing value for our customer. We had an idea, but accepted the murky ROI as unavoidable in content. Now, with Structural Content, both the technology and the approach to using this technology exist to make this a problem of the past, of “content 1.0”. Companies that adopt the Structural Content approach are subscribing to making content a load-bearing part of business operations. They are saying “I want content to be pragmatic. I want to understand exactly why we are doing content. And I want to be able to direct my content investments by business ROI markers.”

Notes

1. Based on The Long and the Short of It by Les Binet and Peter Field with the Institute of Practitioners in Advertising (IPA), the marketing industry adopted the so-called 60/40 rule: in studying nearly 1,000 effectiveness case studies over 30 years, it was found that a 60/40 budget split between brand building and direct response content is the most promising balance for long-term marketing success. Note that the recommendation is for the majority of content to not be direct-response.
2. According to Gartner’s 2025 CMO Spend Survey, companies spend an average of 7.7% of total revenue on marketing — the bulk of which goes to producing, managing, and distributing content, and the infrastructure around it. And 59% of CMOs report that budget is still insufficient.
3. The introduction of the Product Marketing Manager (PMM) helped integrate marketing further into the business, but would require unfeasibly large teams of PMMs to cover all business areas.