Analytics & data
Data you can trust to decide where to invest
We implement and audit GA4, Google Tag Manager, Consent Mode v2, Search Console and Looker Studio dashboards. We start from a measurement plan that turns your business goals into events, conversions and KPIs, and we comply with European privacy rules without giving up more visibility than necessary.
The starting point
If you don't trust your data, you can't optimise anything
In almost every account we review we find something similar: conversions counted twice, events with different names for the same thing, internal traffic mixed in with customers, forms that don't register on mobile, or a cookie banner that blocks more than it should… or less than the law requires.
With data like that, every SEO, Google Ads or redesign decision is made blind. So our analytics work always starts with a business question, not a tool: what do you need to know to make better decisions? From there we design what to measure, how to name it, how to respect consent and how to present it so that people actually use it.
Analytics is the foundation the other services share: it feeds Google Ads bidding, CRO tests and the monthly SEO report.
What we do
Four layers of reliable measurement
Measurement plan
A living document linking business goals, KPIs, events, parameters and conversions, with a naming convention the whole team follows.
Implementation
GA4 and advertising tags managed through Google Tag Manager, with a data layer specified together with your development team or e-commerce platform.
Privacy and consent
A consent management platform, Consent Mode v2 and, where it adds value, server-side tagging to control what data leaves your site and where it goes.
Reporting and decisions
Looker Studio dashboards combining GA4, Search Console and Google Ads, designed to answer specific questions rather than pile up charts.
Consent Mode v2
Basic versus advanced implementation
Consent Mode v2 passes four consent signals to Google: analytics_storage, ad_storage, ad_user_data and ad_personalization. The last two were added in version 2. There are two ways to implement it, and the difference matters.
| Basic | Advanced | |
|---|---|---|
| Tag loading | Google tags do not load until the user accepts. | Tags always load, with consent set to denied by default until the user decides. |
| If the user declines | No data is sent to Google. | Cookieless pings without identifiers are sent (consent state, event type, technical information). |
| Modelling | Google can only apply general conversion modelling. | Google can use the pings to model conversions and, in GA4, behaviour, if the property meets the volume requirements. |
| Privacy | The more conservative approach: nothing leaves before consent. | Technical data is transmitted before consent; this is worth assessing with your privacy adviser. |
| When we recommend it | Sites with a strict privacy policy or in sensitive sectors. | Accounts with significant ad spend that need modelling, once it has been validated from a legal standpoint. |
AI search
Measuring traffic from AI assistants
When ChatGPT, Perplexity, Gemini or Copilot cite your site with a link, some of those visits arrive with an identifiable source (for example chatgpt.com or perplexity.ai), and some assistants append parameters such as utm_source=chatgpt.com. In GA4 we create a custom channel group that separates these visits from generic referral traffic.
Treat it as a qualitative signal rather than a complete measure: many AI answers generate no click, some apps do not pass the source, and traffic from Google's AI features is counted within Google organic search, not broken out separately in Search Console. What is useful is the trend, the landing pages that receive these visits and how those visitors behave. We combine it with the citation tracking we do as part of generative engine optimisation.
Process
How we run an analytics project
Data quality audit
We review the GA4 property, the Tag Manager container, the consent banner and the Google Ads conversions. We check whether events fire when they should (and only then), whether there are duplicates, whether internal and spam traffic is excluded, whether data retention is set and whether consent signals arrive correctly.
We also cross-check sources: Search Console clicks against organic sessions, Google Ads conversions against real sales in your CRM or shop.
Activities
Outputs
- Data quality report with prioritised issues
Measurement plan
We agree with you which decisions you want to make with the data and translate them into KPIs, events and conversions. Each event is documented with its name, parameters, trigger and purpose, so that anyone on the team, or a future agency, can understand it.
Activities
Outputs
- Documented measurement plan
- Technical specification for developers
Implementation and validation
We configure the tags in Tag Manager, Consent Mode v2 with your consent platform and, where appropriate, a server-side container. We validate in preview mode and with DebugView, across browsers and devices, both accepting and rejecting cookies.
Activities
Outputs
- Published, versioned container
- Test log
Dashboards and monitoring
We build Looker Studio dashboards for each audience (leadership, marketing, sales) and connect Search Console, which keeps 16 months of performance data. For longer-term or more granular analysis we enable the GA4 export to BigQuery. We check data quality every month as part of the report.
Activities
Outputs
- Dashboards by audience
- Monthly data review
FAQ
About analytics and measurement
Do I need server-side tagging?
Not always. Server-side tagging gives you more control over what data is sent to each vendor and can improve site performance, but it adds infrastructure and maintenance costs. We mainly recommend it for e-commerce and accounts with significant ad spend. It is not a way around consent: the same rules still apply.
Why don't GA4, Search Console and Google Ads match?
Because they measure different things: Search Console counts clicks on Google results, GA4 counts sessions from users who have consented (or modelled data, depending on the implementation) and Google Ads attributes conversions to its ads using its own model. Differences are normal; what you should watch for is a sudden change in the relationship between them.
What data retention should I set in GA4?
In a standard property you can choose between 2 and 14 months for the event-level data used in explorations. Standard reports are not affected. If you need to analyse granular data over a longer period, the answer is the BigQuery export.
Can you measure how often ChatGPT cites me?
Not with GA4: GA4 only sees visits that reach your site. To measure citations and mentions in AI answers we use a set of representative queries that we review regularly. It is complementary work that we explain on our generative engine optimisation page.
Does my site need a cookie banner?
If you use cookies or similar technologies that are not strictly necessary (analytics, advertising), yes, and it must let people reject as easily as they accept. If you don't, you don't need one. We help with the technical side; the legal wording is best validated with your adviser.
Who owns the accounts?
You do. We create the GA4 property, the Tag Manager container and the dashboards under your company's account, and work with user access you can revoke at any time.
Related
Related
GA4 and Consent Mode v2
GA4 and Consent Mode v2 in the EU: why it exists, the four consent signals, basic vs advanced mode, implementation with a CMP and Tag Manager, and testing.
Google Ads & PPC
Google Ads and Microsoft Ads management: search, Performance Max, Shopping and remarketing campaigns with reliable tracking and accounts in your name.
Generative Engine Optimization
Generative Engine Optimization: we help ChatGPT, Gemini, Perplexity and Google's AI Overviews find, understand and cite your website as a source.