GrowthOS
Read-only R&DGrowth decisions should end in business outcomes, not another dashboard.
GrowthOS is an evidence-driven growth intelligence system connecting search visibility, AI discovery, conversion, content and technical performance to first-party business evidence — prioritised, and prepared for a human to act on.
Version 1.34.5. GrowthOS currently runs read-only: it diagnoses and prepares actions, but every write to a production website, listing or campaign requires explicit human review before anything changes.
Open source. If offered as enterprise SaaS, the commercial edition sits within the NxSync portfolio.
Most growth tooling optimises whatever it can measure most easily — traffic, rankings, impressions, click-through rate, AI citation counts, content volume. None of those is the business objective; each is a proxy that can rise while the business itself stays flat, or even falls, if the traffic it buys was never going to become a customer.
GrowthOS is built backwards from the actual objective instead: qualified enquiries becoming opportunities becoming attributable revenue. Every diagnostic score it produces is read as a signal toward that chain, not as a target in its own right.
The north star
Qualified enquiries → opportunities → attributable revenue.
Diagnostic scores are not the objective. They are diagnostics only.
- 01
Discovery
- 02
Relevant visitor or agent
- 03
Trust
- 04
Conversion
- 05
Qualified enquiry
- 06
Opportunity
- 07
Revenue
- 08
First-party evidence
- 09
Stronger discovery
Why this, not a score
GrowthOS starts from "what is stopping more qualified demand from becoming attributable business?" rather than "how do we improve an SEO score?"
A run does not jump from raw data to a recommendation. It separates observed evidence (what a connected source, like Search Console or the live site, actually returned) from diagnosis (what that evidence suggests is the real constraint) from a recommended action (what a human could do about it) — and only the first category is ever asserted as fact. Missing data is reported as missing, not silently estimated.
Why read-only is a decision, not a limitation
`DRAFT_ONLY = true` and `WRITE_OPERATIONAL = false` are not missing features waiting to be switched on — they are the current governance boundary, stated as a fact rather than a roadmap promise. GrowthOS can read connected data sources, diagnose, prioritise and prepare a recommended action; it does not publish content, change a website, alter search infrastructure, or spend budget on its own.
The reasoning is the same one that runs through the rest of this operator’s work: a system that can analyse and recommend, with a human accountable for anything consequential and irreversible, is a different (and currently more trustworthy) thing than a system that can also act unsupervised. GrowthOS is deliberately the first kind, for now.
Seven skills
Growth SEO
Technical SEO, on-page structure, local SEO and backlink research.
Produces: A prioritised technical and on-page audit — not a link-building service.
Does not claim: Does not build or buy backlinks — backlink work is research and planning only.
Growth GEO
AI citation readiness — whether AI search systems can find and cite the site accurately.
Produces: A readiness read against AI answer engines, brand-entity presence and agent protocols.
Does not claim: Does not claim to control ranking or citation inside a third-party AI system.
Growth CRO
Conversion, funnel and trust-architecture analysis.
Produces: Evidence-based, hypothesis-tested recommendations — never auto-applied to a live site.
Does not claim: Does not run experiments or change a live page without human review.
Growth Content
Content inventory, gap analysis and briefs.
Produces: A disposition-based audit of what exists, what is missing, and what to write next.
Does not claim: Does not mass-generate content or publish without review.
Growth Humanizer
Writing-quality and brand-voice methodology applied to growth content.
Produces: Voice-matched, fact-preserving rewrites, with anti-slop patterns removed.
Does not claim: Does not invent facts or evidence in the course of a rewrite.
Growth Intelligence
Monitors upstream methodology and search-ecosystem changes.
Produces: Change advisories and capability updates the other skills draw on.
Does not claim: Does not change methodology on the strength of one article or trend.
Growth Agent Experience
Whether real browser agents and MCP-equipped AI clients can actually use the site.
Produces: An audit of agent protocols, semantic structure and machine-readability.
Does not claim: Does not claim a given agent will visit — only whether the site is legible to one.
- 01
Understand the business
Read the configured market, products and goals.
- 02
Read the website
Fetch the public site and its technical endpoints.
- 03
Read Search Console
Query real search performance where connected.
- 04
Read Analytics
Query landing-page and engagement data where connected.
- 05
Read Business Profile
Inspect local listing signals where connected.
- 06
Check technical health
Evaluate technical performance signals.
- 07
Check the market
Evaluate SERP, trend and competitor signals.
- 08
Research
Run bounded, evidence-gathering queries.
- 09
Diagnose
Analyse traffic quality and the real constraint.
- 10
Prioritise
Build a growth roadmap in order of business impact.
- 11
Prepare actions
Draft implementation briefs for human review.
- 12
Report
Render the output as HTML, PDF, spreadsheet and JSON.
What it took
- A 12-stage pipeline with strict single ownership per stage, so "who produced this number" is always answerable.
- Seven canonical, runtime-neutral skills (SEO, GEO, CRO, Content, Humanizer, Intelligence, Agent Experience) that declare logical capabilities rather than hard-coding one vendor’s tool names, so the same methodology survives a change of AI runtime or model provider.
- A capability registry that tracks each integration’s real, current connection state rather than assuming a connector works because it was once configured.
- An intelligence loop for methodology itself — monitoring, verifying and proposing changes to how GrowthOS evaluates evidence, gated on human approval before anything is adopted.
- A write-safety boundary enforced in the architecture, not just in the product copy — the kind of thing that is easy to promise and comparatively hard to actually hold to across every code path.
What GrowthOS does not do
- Does not modify a production website, robots file, redirect map or DNS record automatically.
- Does not fabricate a score, statistic or connection status — missing data is reported as missing, not estimated.
- Does not auto-publish content, generate fake reviews or testimonials, or send outreach without approval.
- Does not optimise for vanity metrics (rankings alone, domain-authority scores) at the expense of qualified enquiries.