I build and run the system described in README.md: measurement → decision → fix →
publish → proof → report, running daily on a schedule, with dated artifacts for everything.
How I work: asynchronous, in writing. You get artifacts — JSON, Markdown, dated proof documents,
a per-domain evidence page — not status meetings. If something needs a call, that is a signal that the
reporting is not good enough yet, and I will fix the reporting.
---
| AUDIT | FULL | FULL+ | |
|---|---|---|---|
| Price | €49 one-off | €149 / month | €349 / month |
| See the actual format | PRIMJER-AUDIT-aicommandcenter.pro.md — real audit generated from measured data; every number cites its source file (and says "not measured" where nothing was measured) | same format, daily | same format, daily |
| Delivery | File by email | Daily run + evidence page + 2 Telegram messages/day | FULL + programmatic pages + live mini-tool |
| Domains | 1 | 1 | 1 (each extra domain €99/mo) |
| Technical SEO audit (canonical, meta, H1, JSON-LD, internal links, crawl) | ✅ | ✅ | ✅ |
| GEO readiness score + structured-data validation | ✅ | ✅ | ✅ |
| Search Console + GA4 + PageSpeed + Cloudflare measurement | baseline only | daily, with trend alarms | daily, with trend alarms |
| Fixes prepared with preflight + rollback, published per change | — | ✅ | ✅ |
| Dated proof artifact for every change — before publication (archive + SHA256) and after it (published snapshot + SHA256) | sample | ✅ | ✅ |
| Weekly + monthly report | — | ✅ | ✅ |
| Public evidence page per domain (numbers pulled from the machine snapshot) | — | ✅ | ✅ |
| Programmatic pages from real demand | — | — | built from measured demand; 8–12 pages per domain per run, reviewed before publishing — measured 2026-10-08: 22 Serbian pages live on the two Serbian sites and 35 English pages live on the three English sites, all in the daily measurement rotation |
| Mini-tool built and deployed on your domain | — | — | ✅ |
| No meetings required | ✅ | ✅ | ✅ |
Keys: for AUDIT I need no access at all — you send me the domain, I send back a file.
What is measured live in that file (opening the site, nothing else needed): availability, title,
meta description, H1, canonical (and whether it is inside <head>, where Google actually reads it),
JSON-LD structured data, word count, robots.txt, sitemap.
What is not in an AUDIT — because it cannot be read without access or is not mine to see: Search
Console impressions/clicks, GA4 sessions, rankings. Those say "not measured" in the file rather than
being guessed. If you want them, the daily run does that once access is set up.
For FULL I need read access to Search Console + GA4 (service-account invite) and a deploy path.
You keep ownership of everything.
---
1. Two Telegram messages — one positive (what improved), one negative (what broke or went down).
Hard commitment: silence is a bug, and it is treated as one.
2. A dated proof artifact — every publish leaves HTML + SHA256 + archive, plus a "is it really
live?" check. If a change did not land, the report says so.
3. Trend alarms — the machine watches for drops and tells you before you notice.
4. Your evidence page — one page per domain, no login needed. The numbers are read from the machine's own snapshot, so they match the JSON exactly. It refreshes when the daily run republishes it — it is a dated snapshot, not a live dashboard, and it says the date on it.
---
1. Send: domain + what you want measured (impressions, clicks, conversions, AI visibility).
2. I reply with the AUDIT file within 48 h (€49) — factual, with the raw numbers and where each
one came from.
3. If it is useful, we move to FULL: you invite the service account to Search Console + GA4, and the
daily run starts the next morning at 05:10.
Two separate things, both in writing (details on the Guarantee page):
1. 7-day money-back guarantee on the first FULL/FULL+ payment — unconditional. Ask once in
writing within 7 days and you get that payment refunded. No form, no call, no explanation.
2. 30-day exit — after the first month you are never locked in. Cancel any time; cancellation
stops future billing. No cancellation fee, no minimum term.
The 7-day guarantee is the unconditional one. The 30-day exit is not a results guarantee — I do
not promise that metrics will move (see *What I will not claim*), so a refund is never conditioned
on a number reaching a target.
---
AI citation measurement did not exist yet, so it was not sold. **That sentence stayed on the page after
the module was built — exactly the kind of drift this offer is supposed to prevent.** What exists now:
3 providers (Perplexity, OpenAI, Google Gemini), **8 purchase-intent questions that never contain
your brand name**, every answer stored with a SHA256 hash, and the detection logic itself tested on
constructed answers (7/7, including negative controls that must not produce a false citation).
What is reported is the method, the raw answers, and the frequency across days — **never a
promise of citations. Brand sentiment (NLP) remains not implemented** and is listed as such.
REST API and Webflow, but I have not yet shipped to a live WP/Webflow site. I will say so in
writing rather than learn it on your domain.
dated artifact. If a number does not move, the report says that too.
---
audit refund rule, VAT / reverse-charge / invoice date, and the measurement method in full.
retention, incident response, DPA on request. It also states plainly that there is **no SOC 2 and no
ISO 27001** — so a security questionnaire never gets a surprise.
/.well-known/security.txt — the route for reporting a vulnerability.Async only — no meetings, no calls. Two ways that work:
1. Open an issue on the public repository (fastest, nothing else needed):
https://github.com/dejanswat-sketch/full-vidljivost-dokaz/issues/new
Write the domain and what you want measured; you get a written reply with artifacts.
2. Email or Telegram — you get the same thing in writing, with artifacts attached.
What happens next: within 48 h you receive the AUDIT file (€49) — factual, with raw numbers and
the source of each one. If it is useful, the daily run starts the next morning at 05:10.
Start async — no calls, no meetings. Write the domain and what you want measured; you get a written answer with the artifacts.