SEO & GEO

Organic

One person can't publish enough. A pipeline can.

Six agents, four data connections, one human approval. Built once, runs weekly.

Connected over MCP

The agent reads the same tools I would.

Google Search Console logo
Search Console

Rankings, clicks, impressions, decay

Google Analytics 4 logo
GA4

What converts once they land

Ahrefs logo
Ahrefs

Keyword research, competitor gaps

WordPress logo
WordPress

Where the draft lands

One connected context

Google publishes no official Search Console MCP. Community servers cover GSC, GA4 and Bing in one connection. Ahrefs ships an official one.

The workflow

Six agents, one human gate.

OrchestratorClaude Code

Orchestrates. Runs sub-agents.

01

Watch

Decay, drops, striking distance

02

Find

Keywords and competitor gaps

03

Read SERPs

Top ten, intent, format

04

Write

Outline, draft, charts

05

Edit

Brand voice, facts, claims

06

Stage

Draft into WordPress

Human gateHuman approves before publish
Watch

Continuous. Pages losing clicks, queries slipping from position 8 to 14, anything sitting on page two. Output is a rewrite queue — refreshing decayed pages usually beats writing new ones, and almost nobody automates it.

Published pages re-enter at step one as monitoring data.

OrchestratorClaude Code

Orchestrates. Runs sub-agents.

01

Watch

Decay, drops, striking distance

Continuous. Pages losing clicks, queries slipping from position 8 to 14, anything sitting on page two. Output is a rewrite queue — refreshing decayed pages usually beats writing new ones, and almost nobody automates it.Watch workflow evidence
02

Find

Keywords and competitor gaps

Monthly. What competitors rank for that enmacc doesn't. Trending queries in the markets we actually win. Output is a topic queue with keywords attached.Find workflow evidence
03

Read SERPs

Top ten, intent, format

Per topic. Reads the pages already ranking to work out what Google is rewarding — the intent behind the query and the format that satisfies it. This is the step most content automation skips, and it's why most content automation fails.Read SERPs workflow evidence
04

Write

Outline, draft, charts

Builds the outline first, then drafts against it. Generates charts and diagrams from data rather than stock imagery.Write workflow evidence
05

Edit

Brand voice, facts, claims

A separate agent with a different brief and no shared context. Brand voice, what enmacc can and cannot claim, factual checks, and permission to send it back. One agent asked to write and then review its own work will approve itself every time.Edit workflow evidence
06

Stage

Draft into WordPress

Lands as a draft with meta, schema and internal links already in place. Never published, only staged.
Human gateHuman approves before publish

In a market where a wrong claim about clearing or margin is a real problem, nothing auto-publishes.

Published pages re-enter at step one as monitoring data.

Output

Three clocks, not one schedule.

Continuous

Rewrite queue

Pages losing ground, flagged before anyone notices in a monthly report.

Monthly

Topic queue

Competitor gaps and trending queries, with keywords attached.

Per article

A staged draft

Outline through to WordPress, with charts, meta and internal links.

An AI that writes blog posts is unremarkable. Three loops on different clocks, feeding each other, is a system.

AI visibility audit · ChatGPT, Gemini, Perplexity · 30 keywords

Visible for what it is. Invisible for what it does.

Ask about a mechanism.

Share of voice for mechanism keywords
43% → 0%

Rank #1 for bilateral and OTC energy trading. Zero share of voice across twelve product and German-language terms.

Ask about a product, or ask in German.

Share of voice for product and German keywords

Every environmental term is zero

Carbon allowances, renewable certificates, EU ETS, Herkunftsnachweise. The markets with no screen and no incumbent — the ones enmacc should own.

Every German term is zero

Gashandel, Energiebeschaffung, Herkunftsnachweise. Stromhandel sits at #98, Energiehandel at #132. A Munich company, invisible in German.

'Gas and power brokers' is zero

When someone asks about the alternative to a broker, enmacc isn't offered as one.

How to read this

Read the prompts, not the dashboard.

Share of voice is a single number, and it hides what the query is actually asking. Every keyword here was checked against the prompts behind it before drawing any conclusion.

"Stadtwerke Energiehandel" returns:

  • →Welche Stadtwerke bieten Energiehandel an?
  • →Was kostet Stadtwerke Energiehandel?
  • →Anbieter für Stadtwerke Energiehandel gesucht
  • →Vergleich von Stadtwerke Energiehandel Dienstleistungen

16 to 31 brands per answer.

Here the models read the query as 'which Stadtwerke do energy trading', not 'which platform do Stadtwerke use'. Same zero, different meaning — and that difference decides whether it's worth spending anything on. Platform-intent terms like Energiehandelsplattform and Herkunftsnachweise kaufen are the ones I'd measure against.

Three moves

A zero is easier to move than a fourteen.

01

Defend the two peaks

43% and 41% at rank #1 is a real asset. Whatever produces that keeps running.

02

Attack five zeros, not twelve

Guarantees of origin · Power purchase agreements · Carbon and EU ETS · Renewable certificates · Herkunftsnachweise kaufen

→ Markets with no screen, no incumbent, and a learning curve. Nobody owns these yet, which is exactly why they're winnable.

03

German as its own workstream

Not translation. Native content in the words the market actually uses. Our own job-ad scan found Direktvermarktung in 31% of German energy postings and 'Guarantee of Origin' in none.

Onsitean explainer per product, both languages
OffsiteMontel, ICIS, EFET, Europex, co-authored work. Models cite sources.
Two enginesBing feeds ChatGPT, Google feeds Gemini
The market printa primary source nobody else can publish

The models know what enmacc is. They don't yet know what it's for.