The governance gap is bigger than most teams realize.

The median score: 38 out of 100

Across 18 Condition Score assessments conducted between 2024 and 2025, the median content system maturity score was 38 out of 100.

That number is worth sitting with. A score of 38 means the typical content system we assessed has no named owner, no documented terminology, no structured metadata, and no defined escalation when something goes wrong. Individual contributors are doing good work. The systems around them are not holding. And the AI deployments now reading that content are inheriting every weakness in it.

Why 60 is the governance threshold

We have identified 60 as the governance threshold. It is not arbitrary. Below 60, governance exists as intention: teams have ideas about quality, preferences about tone, informal agreements about how things should work. Individual pieces of content may be excellent. But the system that produces them is fragile. It depends on the right people being in the room, on institutional memory, on someone remembering how we did it last time. AI does not have institutional memory. It has whatever the retrieval system returned.

Above 60, governance becomes infrastructure. Standards are documented and enforced. Patterns exist with usage rules. Quality is measured on a cadence, and those measurements feed back into system improvements. New team members can ramp without months of osmosis. RAG pipelines retrieve from content that is structured, dated, and unambiguous. The system produces consistent output even when the team or the model changes. The full mechanism is documented in our operating model: Signal, Score, System.

Only 3 of 18 organizations we assessed crossed that threshold.

The plateau at 38

The distribution reveals a clustering pattern that's worth examining. Twelve of the 18 organizations scored between 28 and 47, a narrow 19-point band. This is not a normal distribution with a long tail. It is a plateau. Most teams reach a similar ceiling and stop, not because they lack ambition but because crossing the governance threshold requires a qualitatively different kind of investment than what got them to 38. The pattern looks the same when you break the score apart by dimension, as we cover in our dimension breakdown.

Getting from 0 to 40 is largely a craft investment. Hire good writers. Develop a tone guide. Build some templates. Care about quality. These are investments in output, and they yield visible results. Getting from 40 to 60 requires something different: investments in infrastructure. Documenting terminology that everyone assumed was obvious. Assigning ownership over systems that currently run on goodwill. Adding metadata that machines can read. Measuring things that people would rather leave unmeasured. It feels bureaucratic. It is not. It is the difference between an AI system that can be trusted to retrieve from your content and one that cannot.

The cost of absent governance

The gap between the plateau and the threshold has a cost. We estimated rework rates across the organizations we assessed by asking how often content required revision after initial stakeholder review. The pattern was stark. Run a free check on a single artifact with the Condition Signal tool to see where your own content lands.

Organizations below 30 averaged 3.4 revision cycles per piece of content. Those above 60 averaged 1.1. The difference is not marginal. Each revision cycle involves a writer, at least one reviewer, often a stakeholder, and frequently a legal or compliance check. At 3.4 cycles per piece across a team producing 40 pieces per month, the compounding cost of absent governance is measured in thousands of hours annually. The follow-on effect on velocity is examined in our governance and velocity analysis. And that is only the visible cost. The invisible cost is what happens when the same ungoverned content is fed to a chatbot that confidently cites whichever version appeared first.

The governance threshold is not a score to aspire to. It is a structural boundary. Below it, every piece of content is a negotiation, and every AI retrieval is a coin flip. Above it, the system does most of the negotiating for you, and the AI returns answers you can defend.