Content audit

Take stock of what a company already publishes, channel by channel, so the next plan builds on real numbers instead of guesses. The output is a diagnosis: what exists, how often, on what topics, at which funnel stage, what works and what's missing.

When to use it

Use it before writing a content strategy, when taking over a content programme, or when someone asks "what are we actually publishing and is it working?".

Don't use it to write posts or articles, to build the strategy itself (that's /content-strategy, which reads this audit), or for a one-channel quick look you can answer in a paragraph.

Inputs

Read these first if they exist. Agency users: swap context/ for clients/<client>/context/.

• context/icp/: who the content is for and the problems they have. Needed to judge relevance.
• context/positioning/ and context/messaging/: what the company wants to be known for.
• context/goals/: what content is meant to deliver (pipeline, awareness, hiring).
• context/competitors/: optional, for a light benchmark.

Ask the user for:

• Company name and website URL (required).
• At least one content channel URL: blog or newsletter, LinkedIn company page or founder profile. Optional: YouTube, events page.
• Time range. Default to the last 6 months.

If a context file is missing, run anyway and say in the output which judgements are weaker because of it (for example, "relevance scored without an ICP file").

Tools

If you have Firecrawl or Exa, use them to list and read pages. Otherwise use web search with site: queries, or ask the user to paste URLs or exports. List pages first, keep only the ones in scope, then read those. Don't crawl a whole domain to count blog posts.

Steps

1. Find the channels. Check the site footer and navigation, then search site:domain.com blog OR newsletter OR resources and the company name on LinkedIn, YouTube and event platforms. Produces: a channel list with URL, active or inactive, and last-published date.
2. Audit the blog or newsletter. Sample the 10-15 most recent pieces. Record cadence (posts per month), topic, format, funnel stage, and quality signals: headline specificity, depth, a clear call to action, any lead magnet. Produces: a per-piece table plus a cadence figure.
3. Audit LinkedIn. Pull the last 20-30 posts. Record cadence, format (text, carousel, video, image, poll), topic, and visible engagement. Read 5-10 for voice and hook patterns. Produces: the same table for LinkedIn plus voice notes.
4. Audit YouTube and events, if they exist. Cadence, content types, recent view counts, production quality; for events, type, virtual or in-person, and community signals. Produces: one short section each.
5. Map topics to the ICP. Group everything into topic clusters and check each against the problems in context/icp/. Produces: a cluster table showing which ICP problems are covered, thin or missing.
6. Check the funnel mix. Tag each piece top (problem awareness), middle (how to solve it) or bottom (why this product). Produces: the mix as percentages per channel and overall.
7. Score. Rate 1-5 on volume, consistency, quality, ICP relevance, funnel balance and cross-channel alignment, then an overall health score. Each score gets one line of evidence.
8. Find what works. Name the 3-5 pieces or formats that clearly outperform the rest and say what they have in common.
9. Write the recommendations. Split into quick wins (this week), strategic fixes (next 30 days) and longer bets (next quarter). Each one names the gap it closes.

What good looks like

• Every channel listed with its URL, status and last-active date.
• Numbers, not adjectives: "3.2 posts a month, mostly Tuesdays", not "posts regularly".
• Every key finding links to the specific post or page it came from.
• No invented engagement figures or guessed intent. Where data isn't visible, write "not available" and say how to get it.
• Gaps are stated in ICP terms: "nothing on [problem the ICP names]", not "needs more educational content".
• Recommendations are specific enough that someone could start one tomorrow.

Mistakes to avoid

• Judging quality by taste. Tie every score to evidence or to the ICP's problems.
• Counting only the blog. LinkedIn and the founder's profile are often where the real output is.
• Reading likes as results. Engagement shows reach, not pipeline; say which you're looking at.
• Padding the recommendations. Five that matter beat fifteen that don't.

Output

Write to outputs/content/MMYY-content-audit.md with this shape:

1. Header: company, website, audit period, channels audited, date.
2. Summary: three to five sentences on overall health and the biggest gap.
3. Scorecard: the six dimensions plus overall, 1-5, one line of evidence each.
4. Channel sections: inventory table, cadence, topics, formats, notes.
5. Topic and funnel analysis: cluster table against ICP problems, funnel mix.
6. What works: the outperformers and the pattern behind them.
7. Recommendations: quick wins, strategic, longer bets.
8. Input for the strategy: current baseline, gaps to close, voice patterns worth keeping.
9. Data gaps: what you couldn't see and how to get it.

Next step: run /content-strategy, which reads this file as its baseline.