The machinery
Find the lead, or find the leak.
These do it.
Four systems, all of them running on real work rather than waiting to be built. Nothing on this page is a roadmap item. If it is listed here, it has run.
- System 01
- Audit engine · 2 paid audits through it
- System 02
- WhatsApp follow-up · installed per client
- System 03
- Content engine · feeds the retainer pages
- System 04
- Measurement · running on a property I own
System 01 · finding the leak
The audit workflow.
A crawler that refuses to give retired advice, and a six-dimension review on top of it. Two paid audits have run through it.
- Server-side crawl. Up to twelve pages. Every hostname and every redirect hop is re-checked against private address ranges before a connection is made, so the scanner cannot be pointed somewhere it should not reach.
- Six categories, out of 100. Anything it could not measure prints as insufficient data, never as a zero.
- A published ban list. The engine refuses to recommend FAQPage markup, HowTo markup or an llms.txt file, and prints the reason. Every report carries a section titled "what I did not tell you to fix".
- A verdict per AI crawler. GPTBot, ClaudeBot, PerplexityBot, Google-Extended and the rest, each with the rule it came from, never assumed.
System 02 · stopping the leak
The private WhatsApp follow-up system.
Built on your own number and your own approved customer data. Not a broadcast tool — a follow-up system for people who already asked to hear from you.
- Several sending numbers, each with its own daily limit and queue.
- Write the follow-up once. It sends the next step after the days you choose, and only if they have not replied.
- One list showing where every enquiry stands.
- Sequences stop the moment someone answers.
Attached to a build at Format 03, or installed on its own once the visibility work has earned real enquiries.
System 03 · earning the lead
The content engine.
The three to four pages a month inside the Rank Engine come out of this, and so does the proof published under DAINER.AI. One piece of real work becomes the lesson, the long version, and the cuts.
- 01 Real work, or a recording of it
- 02 The lesson and the proof, pulled out
- 03 The long version
- 04 Distribution copy and two to four proof cuts
- 05 Human review, then approved publishing
A person reviews every step before anything goes out. Approval to produce is never approval to publish.
System 04 · proving it moved
First-party measurement.
A local collector console built for one of my own properties. It runs the same discipline the Rank Engine reports under: a source that returned nothing is shown as unavailable, and never averaged into a trend as a zero.
- Indexation is the first queue. URL inspection across the live sitemap, with URLs unknown to Google treated as not indexed for planning.
- Work sorted by distance from the result. Positions 4–10, 11–20, high impressions with low click-through, and rows with no impressions at all — each a separate bucket.
- Sources are never blended. Raw traffic and consent-gated analytics sit side by side, and are never averaged together.
- Blank means no rows. An empty chart says the source returned nothing. It does not say activity was zero.
Shown as an operating example on a property I own. Client reporting is set up per client, on their own accounts, which they keep.
Tell me how enquiries reach you now and I will say which of these would help — or that none of them would, and the leak is somewhere else.
Your website score is not your Academy fit.
Take the 2-minute Fit Check12 questions · Full result before any contact · It can say “not yet.”