Narrative Audit

Narrative Audit is a Discovery Optimization research methodology that tests what a model knows about your brand from training alone. Every other methodology in this guide measures the live web with search turned on. This one turns search off and asks the model directly about you, so you see what it absorbed into memory from training data, and whether that cached story is accurate, outdated, or simply wrong. It is the only methodology that names your brand on purpose.

When to use it#

Run Narrative Audit when:

• You have enough public presence that models have likely trained on content about you.
• You have rebranded, repositioned, or shipped something that may have changed your story.
• You want to catch cached misconceptions before they harden across model generations.

How the method works#

A worked example#

The other chapters use a hypothetical tool in the project management category. Narrative Audit is different: it only works on a brand the model has actually learned about, so picture running it on an established name in that same category, say Linear.

Identity: "What is Linear?"

Comparison: "How does Linear compare to other project management tools?"

Sentiment: "What do people say about Linear?"

With search off, the answers come entirely from training. You are looking for the gaps: Does the model still describe an old positioning the company has since moved past? Does it miss a major capability shipped after the training cutoff? Does it compare it against the wrong set of rivals, or repeat a criticism that is no longer true? Each of those is a piece of narrative drift baked into the model, and the audit is how you find it before your buyers run into it.

How to read the results#

• The model describes you accurately and current. Your narrative is well-established and consistent. Maintain it.
• The model is roughly right but out of date. Expected, since training lags. The fix is fresh, consistent public content that the next training run can absorb.
• The model is wrong, or slots you against the wrong category or rivals. A real narrative problem. Inconsistent or sparse public signals let the model invent a story. Tighten and repeat your positioning everywhere it reads.

What to do about gaps#

• State your positioning consistently across every public surface, since models average what they read.
• Earn authoritative third-party coverage that reinforces the story you want remembered.
• Re-run periodically. Narrative drift is slow, and catching it early is cheaper than correcting a hardened misconception.

Run it in Rampify#

Narrative Audit runs as a Discovery session. Use the Connect Rampify button at the top of this page to start it. The agent asks the three brand-direct questions with search off, records what the model believes, and turns each correction into a spec you can act on.