Because until this is done, every other decision is a guess with a decimal point. And it’s the cheapest work on this site to verify: pick one week, count your orders, and compare it to what analytics says. If the two don’t match, you already know the answer.
Web analytics consulting
Marketing without measurement is structured guessing.
Most sites are optimized against numbers nobody has checked. Measurement runs twice in every cycle: it opens the work as the baseline and closes it as the proof.
One consultant for the whole measurement layer: what gets tracked, whether it’s true, and which few numbers are worth deciding from. Google Analytics 4 (GA4), Google Tag Manager, and the reporting on top of them — for ecommerce and B2B sites.
Problem
Most companies collect data. Very few build a decision system.
GA4 has been installed. Search Console is set up. A dashboard was created. But when you ask “what channel generates actual revenue?” — the room falls silent.
That’s not a data issue. It’s a structural one. Data without architecture is noise; dashboards without decision rules are expensive decorations. And increasingly the gap isn’t the tracking at all — it’s that nobody on the team can turn the numbers into a decision, so the analyst is whoever has time on Friday.
A measurement architecture defines how data moves, connects, and informs decisions — before anyone looks at a chart. And it includes the part most setups skip: teaching the team to read what it produces.
Symptom 01
GA4 events fire but mean nothing
There is tracking. There is no taxonomy. Each event has been named differently. The funnel stages are invisible.
Symptom 02
Attribution is permanently contested
The paid channel is credited with conversions. SEO gets credit for assists. None are demonstrably correct. Budget disputes occur repeatedly.
Symptom 03
Server-side data is missing
Browser-side tracking has lost — by industry estimates — 20 to 40% of events to blockers and privacy settings. Revenue is under-counted.
Symptom 04
Nobody reads the data
The tracking works. The reports arrive on time. And nobody can say what to do differently on Monday, because interpreting them is a skill the team was never given.
Symptom 05
Reporting creates more questions
Each monthly report is long. Each decision is short. The link between each metric and action is missing.
Symptom 06
No single source of truth
GA4 states something. The CRM states something else. The ad platform states something different. Trust in the data falls apart.
What I work in
The stack — and what each piece is for.
These are the tools I set up, reconcile and read from every day. Most of this is Google Analytics consulting; the rest is what makes GA4 true. If yours are on this list, you’re in the right place.
- Google Analytics 4
- The record of what happened: sessions, events, conversions. Every number on this page starts here — and most of the errors do too.
- Google Tag Manager
- How tracking gets onto the site without a developer for every change. Plumbing, not insight. But wrong plumbing means wrong numbers.
- Server-side tagging
- The same tracking, sent from a server you control instead of the browser — harder to block, and first-party by design. Only where the loss is big enough to justify it.
- BigQuery
- GA4’s raw data, joined with your orders and your CRM. For the questions the interface can’t answer. Most sites don’t need it; the ones that do, need it badly.
- Looker Studio
- The reporting layer. Its only job is to show the numbers you defined, with the definitions attached.
- Microsoft Clarity
- What people actually do on the page: recordings, heatmaps, rage clicks. Free, and usually not installed.
- Google Search Console
- What Google shows you for, what gets clicked, and what doesn’t. The bridge between search and measurement.
- Bing Webmaster Tools
- The same for Bing — which is one of the search providers ChatGPT draws on when it searches. Almost nobody opens it. That’s the point.
If your measurement lives in a different stack — Adobe, Matomo, a custom warehouse — someone who works in that one every day will serve you better than I will.
The measurement stack
Every layer has a role. Every layer connects.
GA4 Event Architecture
Event taxonomy clean-up, funnel stage tagging, lead quality signals, tracking form interactions, scroll & engagement model development.
GA4GTMData stream
The foundation of all downstream analysis. Without clean events, every report is unreliable.
Server-Side Tracking
Critical conversion events are routed through a server-side container you control instead of straight from the browser — which is where ad blockers, iOS limits and third-party cookie erosion all take effect.
GTM server-sideStape.ioConversions API
Keeps conversion data that browser-side tracking loses after the event fires. Critical for accurate paid channel optimization and SEO attribution.
Attribution Modeling
Understanding first touch impact, last touch conversion and assisting channel in each part of the customer journey (organic, paid and direct).
GA4 attributionBigQueryCustom models
Makes the case for SEO investment visible. Stops budget debates rooted in last-click bias.
Data Warehouse & Consolidation
When necessary: GA4 raw data, Search Console, CRM signals and behavioral data combined into one queryable source — so patterns show up that no single tool can see.
BigQueryMicrosoft ClarityCloudflare logsServer logs
Not every company needs this. High-growth environments and ecommerce teams often do.
Reporting & Decision Layer
Creating an executive level summary; creating operational monitoring; and establishing iteration based on feedback loops. Not pretty dashboards — clear decision making.
Looker StudioCustom reportsAI analysis
Fewer internal arguments. Faster, more confident prioritization. And a team that can run the next question itself — with AI doing the querying, and the interpretation staying human.
Deciding what to change from those numbers — and testing whether the change worked — is conversion work, not measurement work. UX & conversion →
Server-side tracking
The data you’re losing
is the data that matters most.
Where server-side tracking is worth it
Ad blockers
Ad blockers stop browser-side tracking scripts before they load; the visits they block are never recorded anywhere. Server-side collection moves the endpoint to infrastructure you control, so it’s harder to block and the data stays first-party. It doesn’t recover what a blocker stops in the browser before it’s sent.
Apple’s privacy limits
Intelligent Tracking Prevention caps cookies set by scripts at seven days — one day when the visitor arrived from a classified tracker on a link carrying a query string or fragment. A returning buyer shows up as a new user, and multi-session attribution breaks on the client side.
Meta & Google Conversions API
Both platforms now recommend server-side event matching for accurate campaign optimization. Browser-only signals degrade ad performance.
How it works
How the work runs
01
Inventory — what’s tracked today.
Every event, goal and tag that exists, and what each one actually fires on. Usually the first surprise: half of it was set up for a question nobody asks anymore.
Campaign tagging is part of the same inventory. UTM parameters written by three people in three formats become three channels in the report — and a paid campaign that arrives looking like direct traffic. The fix is a naming convention and someone who enforces it.
Tools: GA4 · Google Tag Manager · Google Search Console · campaign URL conventions
02
Definitions — what counts as a conversion.
The business question first, the implementation after. This step is a conversation, not a configuration screen.
Tools: your order or CRM data
03
Implementation — make it fire correctly.
Events, parameters, and server-side tracking where the loss is big enough to justify it. Consent handling is part of the build, not a bolt-on: what fires before consent, what fires after, and whether the data you keep is data you are allowed to keep under GDPR. On WordPress and WooCommerce I implement this myself, and on a site built on Astro and Cloudflare the tracking is part of the build.
Tools: GA4 · Google Tag Manager · consent mode · server-side tagging
04
Validation — does it match reality.
Analytics against the back office: orders, inquiries, revenue. A number that doesn’t reconcile isn’t a number, it’s a rumor.
Tools: GA4 · BigQuery · your order data
05
Reporting — the few numbers you’ll use.
Not a dashboard with forty tiles. The handful you’d actually check on a Monday morning, with the definitions written next to them.
Tools: Looker Studio · BigQuery
The audit
The audit
A GA4 audit — what’s tracked, what’s true, and what’s missing.
The audit is the beginning of the work, not a document before it. Every event and goal inventoried, the numbers reconciled against your order or inquiry data, and a written list of what to fix — in the order that matters.
If nothing is tracked yet, skip this — the setup below is where you start.
Why this first
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
Around 500 pages
Audit and setup
- 01
Measurement audit
Every event, goal and report checked, and the numbers reconciled with your inquiry data.
€600
12 h · one-off
- 02
Setup
What the audit found, built and documented.
€800
16 h · one-off
- Events and conversions10 h
- Consent and tags6 h
or
Standalone audit
Audit & handover
A written fix list your team or developer can work from
€800
16 h · one-off
Ordered separately
Server-side collection
Data collected on your own domain: first-party, more stable, still consent-based
€600
12 h · one-off
Looker Studio dashboards
Search, AI visibility, technical health and visitor paths, built on your own data
€1,200
24 h · one-off
Webshop
Around 5,000 products
Audit and setup
- 01
Measurement audit
Every event and report checked, and revenue reconciled with your order data.
€800
16 h · one-off
- 02
Setup
Ecommerce tracking rebuilt where needed, and documented.
€1,000
20 h · one-off
- Ecommerce events14 h
- Consent and tags6 h
or
Standalone audit
Audit & handover
A written fix list your team or developer can work from
€1,200
24 h · one-off
Ordered separately
Server-side collection
Cart and purchase data collected on your own domain: first-party, more stable, still consent-based
€800
16 h · one-off
Looker Studio dashboards
Revenue by channel, categories, search and technical health, built on your own data
€1,600
32 h · one-off
These are indicative budgets, not fixed prices. The platform, the number of tools, what is already in place and how many dashboards you need all change the final quote.
No VAT is added to these figures — I’m VAT-exempt under Hungarian law.
Why is the audit lower when I also do the setup?
An audit on its own has to work without me. Every finding is written up with what to change and how to check it, so your team or developer can carry it out.
When I do the setup myself, the audit only has to point me to the problems. The detailed write-up happens in the setup itself, which is why the audit costs less upfront.
What the audit covers
I go through what is set up today: Google Analytics, Tag Manager, Search Console and any other tool that collects data on your site. Every event and goal is inventoried, and the numbers are reconciled against your order or inquiry data, so you know which figures you can trust.
You get a written list of what to fix, in the order that matters, and what each fix changes in your reports.
B2B website: inquiries you can count
On a B2B website, the numbers that matter most are the inquiries: form submissions, calls, email clicks and booked meetings. The setup makes each one a conversion you can trace back to its source.
Consent is set up so that analytics runs only after visitors agree.
Looker Studio dashboards are a separate project, because building them well takes longer: agreeing what to track, connecting the data sources, and testing the numbers against each other. They can be ordered with the setup or on their own.
Webshop: revenue that matches your orders
On a webshop, the numbers have to match your orders. The setup covers the full ecommerce path, from product views through cart and checkout to purchase, with revenue that reconciles with your shop’s own data.
Server-side collection, ordered separately, moves data collection to your own domain, where it is more stable and stays first-party. It does not recover data that a blocker stops before it is sent, and it still runs only with consent.
The dashboards, ordered separately, follow revenue by channel, category performance, search and technical health.
After the setup
The setup ends with a handover: documentation of every event, tag and dashboard, so your team knows what is measured and where.
If something changes later, such as a new form, a site update or a new tool, I work by the hour. There is no monthly retainer for measurement work.
Priced separately
Developer work on your site beyond placing tags and events, including external developer fees.
Paid tools and hosting, such as a server-side tagging server or dashboard data connectors.
Ongoing reporting and analysis. Reading the numbers month by month is part of the SEO, GEO or CRO work.
Before you start
You receive a proposal with the scope, the estimated hours and the fee. It also lists the access your team needs to provide, such as Google Analytics, Tag Manager and the site itself.
There is no lock-in. Any change to the scope is agreed before the extra work starts.
Related areas
This connects to the rest of the loop
Search / SEO consulting
Rankings are diagnostics; this is where you find out whether they turned into revenue.
Search / SEO consultingAI visibility (GEO)
Harder to measure, and measured differently. The baseline matters more, not less.
AI visibility (GEO)Technical SEO
Whether the pages you’re measuring can be crawled and rendered at all. Half of “no data” turns out to be “no index”.
Technical SEO consultingExecution
The fix list is written in order. This is where it gets built — events, server-side, reporting — or handed over with the method written down.
ExecutionQuestions
Questions people ask about measurement
Do I need BigQuery?
Most sites don’t. It earns its place when you need to join analytics with order or CRM data, or when GA4’s sampling and retention limits start hiding what you need. Below that, it’s a cost with no answer attached.
Is GA4 enough, or do we need server-side tracking?
GA4 is enough for most sites. Server-side is worth it when the measurable loss — blocked events, poor mobile capture — is bigger than the setup and maintenance cost. That’s an arithmetic question, and the audit answers it with your numbers rather than a general claim.
Can you fix what we already have instead of rebuilding it?
Usually yes, and it’s cheaper. Rebuilding is only the right answer when the definitions themselves were never agreed — in which case what exists isn’t a setup, it’s an accident with history.
We have a dashboard nobody looks at. What then?
That’s a definition problem wearing a reporting costume. If nobody uses it, it’s answering questions nobody asked. The fix starts by deciding which three or four numbers would actually change what you do.
Does consent handling break the numbers?
It changes them, and pretending otherwise is how a report drifts from reality. With consent mode the modeled part is visible and separable — you know which portion was measured and which was estimated. A setup that still reports a clean 100% after a consent banner went live is not accurate. It is broken.
Next step
Want to know if your numbers are true?
Pick one week. Count the orders or inquiries you actually got, and tell me what analytics says. If the two don’t match, that’s where we start.