Google Analytics Consulting
Google Analytics consulting is help configuring, correcting, and reading GA, the default measurement tool on most websites. The default status is the thing to manage: GA's numbers appear in most reports, and many teams treat them as ground truth. GA measures what its script can observe, estimates part of what it cannot, and its coverage has narrowed year over year as browsers, regulators, and bots change the web underneath it. It works as an instrument when its blind spots are named and its numbers are checked against financial records.
How it actually works
GA4 records events sent by a script on your pages: page views, clicks, purchases, each tied to an identifier stored in a cookie. From those events it assembles sessions, users, and conversion paths, and it applies its own attribution model to distribute credit. Everything it reports derives from what that script successfully observed and sent.
The blind spots are structural, not configuration mistakes. Safari's Intelligent Tracking Prevention caps cookie lifetimes, so a returning Safari visitor counts as a new user and a multi-week journey reads as several unrelated visits. Ad blockers and privacy tools stop the script for a meaningful share of traffic. Bots inflate visits and can fire conversion events. Consent banners remove more visitors from observation, and GA fills part of that gap with modeled estimates. And no tag fires for the phone order, the store visit, or the sales call, so journeys that end offline are invisible to it.
Calibration is what makes GA workable. Compare GA-reported revenue to the order database and the financial records, month by month, and learn your property's gap. The gap is normally stable, and a stable gap makes GA sound for direction and comparison: better campaigns versus worse ones, this month versus last, even when the absolute numbers run low. Trouble starts when absolute GA figures flow into board decks or bidding decisions as if they were complete. Because every GA-based CAC and ROAS inherits the understatement, the gap should run as a standing adjustment with a named owner and a stated tolerance: the owner reconciles GA revenue to the order database monthly, publishes the current factor, and investigates whenever the gap moves outside tolerance, since a moving gap means the tracking or the traffic changed. That is the same control finance applies to any account it reconciles, applied to a reporting system that otherwise runs unaudited.
In practice
A common finding in our audits: GA showing a double-digit percentage less revenue than the order system, with the gap concentrated in Safari and mobile traffic, and a team pricing its acquisition decisions on the GA number. The spend was being judged against understated revenue, which made working campaigns look marginal. Correcting the read changed the budget allocation before any ads were changed.
Where we come in
We configure GA to capture what it can capture, document what it cannot, and reconcile it on a schedule against financial records. GA becomes one instrument inside a measurement system, alongside controlled experiments and portfolio-level models, instead of the single source of record for performance. The blind spots stop being surprises and become known quantities your team plans around.
Start a Revenue Health Pre-Assessment →See it in action
Related terms
- GA4
- The current version of Google Analytics, built on an event model rather than sessions. Its reports blend observed data with modeled estimates, and the blend is not always labeled.
- Intelligent Tracking Prevention (ITP)
- Safari's restriction on tracking cookies, which caps their lifetime. It splits one returning visitor into several new ones, understating retention and overstating acquisition.
- Bot traffic
- Automated visits from crawlers and scripts. GA filters some of it; the remainder inflates traffic and can fire conversions, and it is not evenly distributed across channels.
- Consent mode
- Google's mechanism for respecting visitor privacy choices. Declined visitors go unobserved, and GA models estimates of their behavior into some reports.
- Data sampling
- Reporting on a subset of data and extrapolating, which GA applies to some complex queries. Sampled reports can differ from the full data, and the sampling is easy to miss.
- Offline conversion
- A sale that closes where no script runs: a phone call, a store visit, a signed contract. GA can only see it if it is imported, and most setups never import it.