Marketing: KPIs
Measures for a knowledge-led marketing function, why published volume and impressions are excluded, and the content decay review nobody runs.
Markdown. No sign-up, no email.
Once producing content is nearly free, every volume measure becomes noise. These measure whether anything reached the right person and whether it kept working.
The six that matter#
| Measure | Definition | Target | How it gets gamed |
|---|---|---|---|
| Qualified organic sessions | Visits from search matching the customer profile | Growing | Counting all traffic, which rewards attracting the wrong readers |
| Content earning traffic at 90 days | Share of published pieces still bringing visitors | Above 50% | Publishing more, which lowers the ratio and hides it in the total |
| Conversion to a named contact | Readers who identify themselves | Above 1% | Gating content, which trades compounding search value for a short-term list |
| Ranking pages | Pages placing in the top 10 for a target term | Growing | Targeting terms nobody searches, which rank easily and deliver nothing |
| Cost per qualified contact | Total marketing cost divided by qualified contacts | Falling | Excluding people-time from cost |
| Source attribution | Contacts whose origin is known | Above 80% | Recording "direct", which is what unknown looks like when nobody chases it |
Content earning traffic at 90 days is the one that exposes the content-farm failure. A function publishing four strong pieces a month where three still earn traffic in a quarter is beating one publishing forty where six do, even though the second looks ten times busier on every activity report.
Two counter-metrics#
| Counter-metric | Catches |
|---|---|
| Average time on page for new content | Falling while volume rises is the signature of filler |
| Share of site that is thin | Pages under the quality threshold drag the whole domain. Growth in this number is a slow, invisible penalty |
Content decay review, quarterly#
The review nobody runs, and the one with the best return.
| Question | Action |
|---|---|
| What ranked last year and does not now? | Update it. Refreshing a decayed page beats writing a new one, almost always |
| What has never earned a visitor? | Improve it or remove it. Keeping it costs domain quality |
| What is out of date? | Anything dated, technical or regulatory. A wrong page is worse than no page |
| What ranks on page two? | The cheapest wins available. Small improvements move these; nothing moves page five |
What is deliberately not measured#
- Pieces published. Free to produce, so it measures nothing but effort spent.
- Impressions and reach. Rented attention that ends when spending does.
- Followers. Weak correlation with anything commercial.
- Keyword rankings in aggregate. An average across hundreds of terms hides both the wins and the collapses.
Cadence#
| Weekly | New qualified sessions, what published, what converted |
| Monthly | Ranking pages, cost per qualified contact, attribution completeness |
| Quarterly | Content decay review, thin-content share, positioning check |
The honest position today#
This function's measures are currently dominated by a single fact: traffic is the binding constraint, not conversion. Optimising conversion on a site nobody visits is arithmetic performed on zero. Until qualified sessions are meaningfully non-zero, the useful numbers here are the ones above the funnel, and saying so is more useful than reporting a conversion rate calculated from a handful of visits.