~$~/nine/index.tsx
LIVEv0.9 · mainEST. 2003

The Nines/CONTENT/Content auditing: how we figure out what's working before we touch anything.2024_10_07

Content auditing: how we figure out what's working before we touch anything.

author

Brian Aldrich

tag

content

filed

2024.10.07

read_time

10 min

---

section summary

tone direct

---

Most content audits are just inventories with traffic numbers next to them. Ours are decision documents — keep, kill, rewrite, merge — backed by AI-assisted analysis on the parts that used to take humans weeks.

Before we publish a single new piece for a client, we audit what's already there. Not to feel busy — to figure out which assets are earning their keep, which ones are dragging the rest of the site down, and which ones are one rewrite away from being the best-performing thing on the domain.

What a real audit is — and isn't.

A spreadsheet of every URL with sessions next to it isn't an audit. It's an inventory. Inventories tell you what exists. Audits tell you what to do.

Our audit ends with a recommendation on every URL: keep, rewrite, merge, redirect, or kill. Nothing else. If a piece doesn't earn one of those five labels, we haven't finished thinking about it.

How the audit actually runs.

1. Pull everything into one place.

Every published URL on the domain — blog, resources, glossary, case studies, landing pages. We attach search performance, organic landing data, conversion paths, internal link counts, and last-modified dates. The goal is one row per URL, with every signal we'll need to make a call.

2. Cluster by topic, not by category.

Most sites are organized by the marketing team's mental model, not by the way search engines or buyers actually group topics. We re-cluster every URL into topic groups based on real query overlap. This is where we find cannibalization — three pages competing for the same query, none of them ranking.

3. AI-assisted content analysis.

This is the part that used to eat weeks of analyst time. We run every page through an LLM-powered scoring layer that evaluates depth, alignment to the topic cluster, recency of claims, and overlap with sibling pages. The model doesn't make the final call — humans do — but it gets us 80% of the way to a recommendation in a fraction of the time.

It also catches things humans miss when they're tired: contradicting claims across pages, stat citations from sources that no longer exist, links to dead products.

4. Decide and document.

Every URL gets one of the five labels and a one-line reason. The output is a decision doc the client's team can actually act on — not a 200-page PDF that goes in a drawer.

What we look for that other audits miss.

  • Cannibalization at the cluster level. Two pages ranking on page two for the same query are usually one merged page that would rank on page one.
  • Decay curves. A piece that was great in 2021 and is now half-wrong is a liability, not an asset.
  • Internal link orphans. Pages with zero internal links pointing in are pages search engines have effectively forgotten about.
  • Brand-vocabulary drift. Pages using language the company stopped using two years ago.
  • Conversion dead ends. Pages with traffic but no path forward — no related links, no CTA, no next step.

Where this fits in the larger system.

An audit is the first step of a four-channel content engine — auditing tells you what to keep, production fills the gaps, omnichannel strategy decides where each piece needs to live, and distribution pushes it to the audiences that haven't seen it yet. Skipping the audit means the other three steps are guessing.

That's the part most agencies skip. They lead with new content because new content is what clients think they're paying for. We've found over and over again that 60-80% of the immediate gains a client can capture are already on the site — they just need the right edits, the right merges, and the right kills.

What the deliverable looks like.

  • A URL-level decision doc with five-label recommendations.
  • A topic-cluster map of the existing site, showing overlap and gaps.
  • A prioritized rewrite queue — biggest opportunities first.
  • A redirect plan for everything we're killing.
  • A short list of new topics the audit revealed are missing entirely.

From there, the client either runs the recommendations themselves or hands them to us to execute. Either way, no part of it is theoretical — every line item has an owner, an effort estimate, and an expected outcome.

When you should audit.

Annually, at minimum. Anytime you've changed positioning. Before any major content investment. After any platform migration. And anytime you can't honestly answer the question: which of our published pages are pulling their weight right now?

If you're not sure, you're due.

Ready to put us to work?

next_step

~$nine init --audit

Start with an Insight Genesis audit. Six weeks. Fixed scope. A written diagnosis of where your marketing actually stands — plus a working agent prototype tailored to your business.