Ideas don’t create growth.
Execution does.
Most organic growth initiatives stall not because of weak strategy — but because of fragmented implementation, unclear priorities, and no accountability loop. This is the operating model that makes the system move.
The backlog keeps growing
Items are added faster than they are shipped. Nothing ever feels done.
Priorities change every week
No stable sequence. The team reacts instead of builds.
Improvements exist in documents
Audits and reports sit unimplemented. The gap between plan and live is wide.
No feedback loop after shipping
Changes go live, but impact is never measured or connected back to decisions.
Months of activity.
Nothing compounds.
The same pattern appears across companies of different sizes: clear diagnosis, a reasonable plan — and then months of scattered activity with no compounding effect. The cause is usually structural — and I’ve been the third partner in a row often enough to recognize it.
Nobody owns the number
SEO waits for development, development waits for content, content waits for analytics. Everyone owns their part. When I ask who owns the outcome, the room looks at the door.
Velocity collapses after kickoff
The first weeks move fast. Then reviews multiply, approvals stack up, scope creeps. By week six the cadence is gone. I’ve watched this happen to good teams with good plans — which is why the model below has one fixed date, and everything else is pulled.
Strategy resets instead of iterates
When results are slow, the instinct is to start over: new approach, new tool, new partner. Each reset throws away the one thing that was compounding — the measurement. The plan was different each time I arrived. The missing loop was the same.
Nothing waits
for the next sprint.
Work is pulled, not batched. There is one ranked queue, one or two things in flight at any moment, and a review at the end of every month that re-scores what’s left. The only thing on a fixed schedule is the review — the work itself moves when it’s ready.
Baseline & diagnosis
Access set up, data reviewed, current state mapped. What’s technically broken, which pages are losing ground, where the funnel drops.
A ranked queue, not a wishlist
Every candidate is scored by impact, confidence and ease before it enters the queue. New findings are scored the same way and slot in where they belong — not at the back because they arrived late, and not at the front because they arrived loudly.
One or two things, all the way out
The top of the queue gets built and shipped. Nothing new starts until it is live. QA runs before anything goes out.
Compared to its own baseline
Each change is measured against the state before it. Did the metric move? Why, or why not? The answer is written down before the next item is pulled.
The one fixed date
Once a month we look at everything that shipped, what it did, and what the data now says. The queue is re-scored against the new picture. This is the only thing in the model that runs on a calendar.
What you will not get: a 60-page PDF with 200 recommendations. A long roadmap meeting with no follow-up. A month of “strategy” with nothing shipped. The first thing that goes live is the first thing that gets measured — and that’s what informs the next decision.
Two ways to work.
Same operating discipline.
The model adapts to your environment. The cadence, the prioritization logic, and the measurement layer are the same. What differs is who holds the keyboard.
Direct Implementation
I do the work. Strategy and build in the same pair of hands, so a decision and its implementation are not separated by a handoff or a developer queue.
Structured Oversight
I hold the queue, the priorities and the quality bar. Your team builds. I am not managing your people — I am removing the ambiguity that makes them wait.
In both modes the queue, the scoring and the measurement are identical. The difference is who writes the code — not how decisions get made.
When I’m not the right choice
If you need a content production team, a simultaneous multi-market rollout, or a dedicated account manager for internal stakeholder reporting, a larger agency structure fits better.
Continuous work also needs continuous decisions. If your approvals run on a three-week cycle, the loop spends most of its time waiting.
And I don’t work on Drupal.
Not everything deserves action.
The framework that decides.
Every candidate task is scored before it enters the queue. The scoring model prevents gut-feel prioritization and ensures that constrained time goes toward the highest-leverage work first.
How far the metric in scope moves if this works. 1–10.
How much evidence backs the hypothesis. Data beats pattern beats opinion. 1–10.
How little time and how few dependencies it takes to ship. Not how simple it sounds. 1–10.
I × C × E. Maximum 1000.
Anything under 200 does not enter the queue. That is not a rejection — it is an instruction: either break it into a piece small enough to ship and measure on its own, or run the cheap test that raises the confidence first. The content cluster above came back as a single decision-stage page for the highest-value term, scored 7 × 7 × 7 = 343, and shipped.
AI supports the scoring process by accelerating pattern detection across data sources — identifying which technical issues correlate with ranking drops, which pages have the largest conversion gap, and which intent clusters have the weakest content coverage. The scores are structured judgment, not guesswork.
Illustrative — the scoring model applied to a representative backlog
Strategy without execution
is theory.
Execution is not the last layer in the stack — it is the operating layer that runs through every other layer. Measurement insights feed the queue. SEO and GEO findings become queue items. UX hypotheses are shipped and measured in cycles.
Measurement
Data flows in. Attribution is clarified. The queue is written from insight, not assumption.
Layer 02SEO
Technical fixes, page architecture, intent mapping — all become queue items with defined scope and success metrics.
Layer 03GEO
Entity structuring, schema implementation, authority architecture — shipped in short cycles, monitored across quarters.
Layer 04UX & CRO
Friction is identified in one cycle, removed in the next, measured in the one after. The loop makes improvement compounding.
Execution
The operating model that connects all four layers into a running system — and ensures nothing stays in a document.
Pricing model
What it costs.
I charge €50 per hour. Every figure below is hours × €50, so you can see what each budget buys.
Hourly consulting€50 / hour
Two example projects
B2B website
4 templates
Template build · choose one
Static, AI-built
Four templates built in code. Content changes go through your developer or me.
€1,600
32 h · one-off
With an editing interface
The same four templates with Sanity, so your team can edit content in the browser.
€2,800
56 h · one-off
WordPress
Four templates as a custom WordPress theme, edited in the WordPress admin.
€2,800
56 h · one-off
Alongside the build
Project management
Priorities, coordination and follow-up until the site is live
€50 / hour
billed by the hour
Webshop
Shopify or WooCommerce, 8 templates
Template build
Shop templates
Eight templates, from the homepage to the checkout, built on Shopify or WooCommerce.
€4,000
80 h · one-off
Alongside the build
Project management
Priorities, coordination and follow-up until the site is live
€50 / hour
billed by the hour
These are indicative budgets for building the templates. Research, information architecture, content writing and moving existing content are quoted separately.
No VAT is added to these figures — I’m VAT-exempt under Hungarian law.
Why is project management billed by the hour?
A build has a clear end, so it can be estimated in advance. Project management depends on how many people are involved, how fast decisions are made and how much changes along the way.
Billing it by the hour means you pay for what the project actually needs, not for a monthly estimate.
What a template build covers
A template is the layout one kind of page uses: the homepage, a service page, a blog article. I build each template once, test it on phones and desktops, check its speed, and make sure search engines can read it. Every page of that kind then uses it.
The build includes the technical basics: clean markup, structured data where it fits, redirects from the old site’s addresses, and a sitemap.
Static, editing interface or WordPress
A static site built with AI-assisted development is the fastest and the cheapest to run: no database and no plugins to update. Content changes go through the code, which suits teams with a developer, or with me on hand.
With an editing interface, your team edits in the browser. Sanity suits sites with several editors and structured content. Decap CMS is a cheaper alternative, about 44 hours (€2,200) for the same four templates, and works well when the content is simple.
WordPress is the familiar choice when your team already knows it. The templates are built as a custom theme, not bought ready-made.
Webshop: Shopify or WooCommerce
Shopify runs the shop for you: hosting, payments and updates are covered by its monthly fee. WooCommerce runs on WordPress on your own hosting, which gives more control and more to maintain.
Eight templates cover what a shop needs: the homepage, category, product, search, cart, checkout, account and content pages. Product data, payment and shipping setup, and moving an existing catalog are quoted separately.
Project management
Project management covers the priorities, coordination with your team and other suppliers, and following the work until it is live. It is billed by the hour, and the hours are reported with the work.
Priced separately
Research and information architecture, which decide what the site needs before it is built.
Content writing and moving existing content, including an external writer if needed.
Design beyond the templates, such as a new visual identity.
Plugins, apps, themes and hosting that carry their own fees.
Before you start
You receive a proposal with the templates, the estimated hours and the fee. It also lists what your team needs to provide, such as content, access and decisions.
There is no lock-in. Any change to the scope is agreed before the extra work starts.
What people ask before
starting an engagement.
How involved do I need to be?
Minimal, but regular. One alignment conversation at the start — 30–45 minutes. After that, brief async updates when something ships, and one review a month. That review is the only meeting in the model.
What if I already have an internal development team?
That’s the Structured Oversight mode. I own the queue, the priorities, the QA logic and the measurement design; your team builds. It removes the ambiguity that usually makes internal teams wait, without replacing their capacity.
How do we know whether something worked?
Success criteria are written before an item is built, not after. Every item in the queue carries its own definition of success. When it ships, it’s compared to its own baseline — and if it didn’t move, that’s documented and the queue is re-scored.
What is the minimum commitment?
There isn’t one. Work is billed by the hour, project management included, with no fixed term.
Does AI replace human decision-making here?
No. AI accelerates the diagnostic and scoring work — pattern detection in data, anomaly identification, content gap analysis. But every priority call, every success criterion, and every decision about what ships next is business judgment, not generated output. AI provides speed and coverage; the strategy stays human.
The same split applies to building. Producing a page is fast now, and that is not where the work is: what goes on it, in what order, and against which measurement was decided in the research. The speed changes what a small budget can reach, not what a decision is worth.
If growth lives in documents
instead of in production —
Let’s identify the highest-leverage starting point and start with the diagnosis. No lock-in, no fixed term. The first thing shipped is the first thing measured.