Warehouse-native KPIs
One metrics layer in BigQuery that feeds GA4 and BI without drift.
How we engage
We start from the decisions you need to make every week: funnel health, feature adoption, campaign efficiency - then work backward into events, dimensions and governance so new releases don’t silently break reporting.
Often paired with: SEO & growth · AI automation · Web development.
- Measurement sprint - initial review, cleanup of event names and rollout of the v1 plan.
- Dashboards - views for leadership and teams with agreed definitions.
- Experiments - test design, power analysis and readouts for product decisions.
From definitions to trustworthy readouts
Naming and QA before yet another custom dimension nobody trusts.
Plan KPIs & definitions
Events Taxonomy + GTM
Report Dashboards
Test A/B design
Privacy Consent & personal data
Measurement plans
Goals, KPIs, segments and definitions of “active” that finance, product and marketing can cite without a 30-minute debate.
We document the primary and guardrail metrics, map them to events and dimensions, and flag where measurement isn’t in place yet.
Plans include expected data freshness and known blind spots (e.g. ad blockers, offline access), so leadership doesn’t draw conclusions from noisy charts.
Event taxonomy and tracking
GA4, GTM (or the tag manager of your choice) and server-side options when client reliability matters - tested with Tag Assistant and real sessions.
We use a consistent naming convention, version breaking changes and run regression tests on key flows after every deploy.
When BigQuery or a data warehouse export is in play, we document the schema and partitioning so analysts don’t get stuck on UI limits.
Dashboards and reporting
Executive roll-ups and squad views with filters that follow how you actually segment customers - not 40 tabs nobody opens.
We prototype in Looker Studio, Metabase or your BI tool, then tie calculated fields to the measurement dictionary.
Annotations for campaigns, releases and incidents are built in, so sudden spikes still have context six months later.
A/B testing and feature flags
Hypothesis, power analysis, randomization checks and readouts tied to the ship decision - even when results are flat.
Before sending traffic, we align on the primary metric, guardrail metrics and the minimum detectable effect. Feature flags are tied to the same event taxonomy.
Post-test summaries include segments to explore and a recommendation: iterate or roll back - no endless multivariate sprawl.
Privacy-aware setup
Consent modes, personal data boundaries, retention and regional specifics - all documented so legal sign-off doesn’t block every tag change.
We map which tags fire before and after consent, redact fields at collection where we can, and separate marketing IDs from product analytics where needed.
DSAR and deletion implications are written down for warehouse tables - not an afterthought when a GDPR email arrives.
Client voice
“We finally stopped arguing about event names in every planning meeting - the dictionary is the source of truth.”
Roadmap
In pilot
Exploring
Coming soon
Exploring
On the horizon
We invest where client roadmaps are heading - carefully piloted, not as a demo for slides.
Experiment premortems
A structured risk review before expensive traffic splits.
Privacy Sandbox readiness
A guide to tracking Chrome Topics and the attribution APIs.
Automated cohort monitoring
Early warning on activation metrics via Slack hooks.
What a typical two weeks looks like
A typical analytics hardening cadence alongside your release train.
Week 1
- Access: GA4, GTM, product database (within permitted scope)
- Initial review and a list of broken events
- Workshop on the decisions dashboards must support
Week 2
- Release of taxonomy v1 and GTM changes to the test environment
- QA pass across the 5 most important flows
- Dashboard v1 + stakeholder review
What we need from you
- Admin access or a shared GTM workspace
- List of regulated data categories
- Feature flag or experimentation tool (if you have one)
- An owner for ongoing dictionary updates
You might be better off starting elsewhere
If you don’t have stable measurement yet, cleaning up SEO or CMS content before heavy experiments may pay off more.
FAQ
Analytics and experiments
Tools, access, and how reporting stays reliable after we leave.
Ask a specific questionYes. Where possible, we use separate test views or a staging GTM, then promote changes with a clear log.
No - we make definitions and pipelines reliable so in-house analysts spend their time on insights, not on firefighting broken events.
We implement it when client-side reliability or ad-block resilience justifies the operating cost - and we document hosting and failover.
We coordinate with SDK owners (Segment, Firebase, etc.) so mobile and web events line up in the dictionary.
A statistical power check, guardrail metrics, segments registered in advance, and a written recommendation - flat results are still a useful learning.
Analytics and experimentation - all of Serbia
Our studio is in Novi Sad, and we work with companies across Serbia - Belgrade, Niš, Kragujevac, Subotica, Novi Pazar, Čačak, Pančevo and beyond. The way we work is the same whatever the city: video calls for the weekly rhythm, in-person visits when they make sense.
Tell us what you’re building
Tell us which decision is waiting on data - funnel, feature adoption, or campaign ROI - and we’ll propose a measurement plan or test design.