Set decision rights before the next disagreement.

Content Governance defines who may approve each type of content, which risks require another signature, and how exceptions are handled. Decisions no longer depend on whoever happens to be available.

You walk away with written decision rights, risk tiers that have been tested against real cases, and a named exception path people can actually use.

Design a governance model

Some of the 500+ brands we've worked with

See all references

We begin with how decisions happen today, then test and introduce a model the team can follow.

Map how decisions actually get made now
We look at real approvals rather than the org chart. Who actually said yes last time, how long it took, and where things stalled or got skipped entirely.
A picture of current decision rights, gaps, and recurring friction points.
AI assist
Agents can read a year of approval history and show where work stalled, how long each stage held, and which exceptions kept recurring. That record describes the current state, it does not propose who should hold authority.
Human gate
The people who actually approved things last time confirm the map is accurate.
Name the risk tiers
Not every piece of content carries the same risk. We sort content by what could go wrong if it's wrong and match review weight to that.
A risk-tiering model with the review level each tier requires.
AI assist
Agents can group past incidents by the kind of harm involved, which is a useful starting shape for the tiers. Where the line between tiers falls is a risk judgment your leadership makes.
Human gate
Whoever holds real authority signs off on which risk tier a given content type belongs to.
Write the rules down
Who can approve what, who needs a second signature, how exceptions get requested, and where AI-assisted work needs disclosure or an extra check, in language people will read.
A governance charter and decision matrix mapped to the risk tiers.
AI assist
Agents can sort a year of approval records quickly. The strategist looks for the few patterns worth turning into rules.
Human gate
The charter includes only historical patterns that a strategist judges material enough to become a rule.
Test it against real cases
We run a handful of past decisions through the new model to see if it would have produced a sane answer, and fix anything that would have jammed.
A validated model with edge cases and exception paths worked out.
AI assist
Agents can replay real past cases against the drafted rules and show which ones would have routed differently. Whether the new route is better is argued through with the people who lived those cases.
Human gate
The named approvers confirm the model would have produced a sane call on each past case before it goes live.
Roll it out and watch adoption
We hand the model to the people who'll use it, train them on the exception path, then watch whether real decisions start following it.
A rollout plan, adoption signals to watch, and a review date.
AI assist
Agents can track which approvals are following the documented route after rollout. A drop in adoption is investigated by talking to the team, not by tightening the rule.
Human gate
Whoever owns the charter reviews the adoption signals and decides if the model needs a revisit.

Who gets to say yes is not a question a model can answer

A year of historical approvals shows where decisions stalled and which exceptions kept recurring. An agent can sort that record, summarize the stalls, and compare current rules with how similar teams structure theirs. The record still cannot assign authority. Someone with actual sign-off power has to decide which risks need a second signature, when a bottleneck is worth keeping, and who owns the exception path. We keep those choices with the people who will use the rules, and each person remains named against the tiers they approve.

The charter records the rules, while the decision map shows why those rules are credible.

Decision-rights map
Who currently approves what, where that broke down, and what a cleaner version would look like.
Governance charter
The written rules for who signs off on what, by risk tier, including where AI-assisted content needs an extra check.
Rollout plan
How the new model gets introduced, who trains on it, and what we'll watch to see if it's being followed.

We call it done when: the decision rights are written down, tested against real cases, and the people who'll use them have seen the model and know where the exception path is.

This creates a practical rulebook for approvals, escalations, and exceptions.

A good fit when

The same approval dispute keeps returning, because nobody can point to a written rule showing who gets the final say.
Content now touches legal, brand, and multiple teams, and nobody's written down who has final say.
AI-assisted drafting has raised review and disclosure questions your current policy doesn't answer.

Better handled as other work when

You want day-to-day workflow steps rather than a decision-rights framework. That's Editorial Workflow Design.
Every content decision is expected to pass through one person, so the proposed rules would preserve the bottleneck they are meant to resolve.
The charter needs a signature, but nobody with actual approval authority is willing to own the resulting risk tiers and exception path.
Work is simply moving too slowly, while the team already agrees who can approve each content type. Editorial Workflow Design addresses that operating problem.

Some of these consultants write the pages themselves. Others plan distribution or check how a piece performs once it's live, so the team below covers more than one stage of the work.

Acrolinx
the codified rule set that turns a written policy into an automatic check
Notion
the decision-rights table itself, linked from every workflow it governs
Airtable
a filterable risk-tier database for testing rules against real cases