Decide What Gets Audited

Lesson outcome

You will define an audit threshold for the agency so teams know when they can move fast and when they need a formal second-model review.

Why this matters in an agency

Without an audit threshold, one of two bad things happens. Either the team audits everything and slows to a crawl, or it audits nothing and lets bad assumptions slip into client work, pricing, implementation, or public claims. Strong audit systems protect speed by concentrating review where the downside is largest.

Inputs, tools, and prerequisites

Use recent agency outputs as examples: a proposal, a code change, a pricing comparison, a workflow rewrite, a report narrative, or a content plan. You also need the capability matrix and orchestration notes that distinguish generation from verification.

Step-by-step walkthrough

Start by sorting your recent work into risk categories. Low-risk work might be internal drafts or disposable brainstorming. Medium-risk work might include internal SOP edits or first-pass analysis. High-risk work includes anything that changes a client promise, commits engineering changes, publishes a factual claim, or informs pricing and strategic choices.

Now write an audit rule for each category. Low-risk outputs can move with light owner review. Medium-risk outputs need a checklist or evidence pass. High-risk outputs require explicit second-model review, source inspection, and a record of the decision. This turns auditing into policy rather than mood.

Then define the cost of failure for each category. A weak internal brainstorm is cheap. A wrong pricing claim or broken code path is not. When the team sees the downside clearly, the audit threshold becomes easier to follow without debate.

Failure modes and verification checks

The biggest failure is judging risk by how polished the output looks rather than by its consequence. Verify the threshold by testing it on three real deliverables from the last month and asking whether the rule would have caught the risky ones.

Implementation checklist

• Sort recent work into low-, medium-, and high-risk categories.
• Define the audit rule for each category.
• Write down the cost of failure for high-risk outputs.
• Share the threshold with the team.

Immediate next action

Take the last five meaningful outputs from the business and classify them under your new audit threshold.