GA4 implementation services for numbers you can defend
Installing the tag is the first hour. Not the job.
GA4 implementation services cover planning, configuring and validating Google Analytics 4 so the data can be trusted: a measurement plan, an event and parameter schema, key events tied to revenue, consent handling, and validation against a second source. In your own accounts, for ecommerce, subscription and iGaming brands.
- 01Measurement plan before any tag is touched
- 02Built in your accounts, documented at handover
- 03Validated against a second data source
Start with a free GA4 review
FreeThe review checks whether your property is collecting, whether key events fire and fire once, whether consent is blocking the tag, and whether cross-domain and referral exclusions are set. Findings within 24 hours.
Prefer to talk it through? Book a free 30-minute intro call →
GA4 implementation services: a tag install is the smallest part
GA4 stands for Google Analytics 4. Pasting its tag takes ten minutes. Everything that decides whether the numbers mean anything happens before and after.
A tag install
Pageviews and a few automatic events. It confirms traffic arrived, not what it did or what it was worth.
An implementation
A measurement plan tied to how you make money, an event and parameter spec, key events, consent handling, validation.
Your account, not ours
The property sits in your Google account, the container in your Tag Manager. Access stays with you.
GA4 fails quietly. A wrong currency parameter, a purchase event firing twice, a consent banner blocking the tag: none throws an error. They produce a plausible number that is wrong, and someone budgets against it. This page owns the measurement layer. The container that delivers the events is Google Tag Manager work, and dashboards, attribution and reporting sit on the marketing analytics hub.
Audit, setup or migration: which one you need
Nothing is rebuilt before it is diagnosed. The audit comes first even when the answer turns out to be a rebuild, because you cannot migrate history you have not understood.
GA4 audit
For when GA4 runs but nobody trusts it. We review the property, container, events and consent, then report what is broken and what to fix first.
GA4 setup
For when nothing is collecting, or what exists is beyond repair. Built from a measurement plan, with events named to a spec so reports stay readable.
GA4 migration
For moving off a legacy property, or where several properties have collected overlapping data and nobody knows which to trust.
Not sure which? The free review answers that. And if you run a small site with one contact form and no ad spend, you need none of the three: install the tag, mark the form as a key event, and spend the budget where it moves revenue.
Events, parameters and the measurement model
GA4 records events carrying parameters. That is the whole model, and it is why a loose naming habit turns into an unreadable report six months later.
Named to a spec
Recommended event names where they exist, because Google's reports and Ads integrations expect them. Custom events only where nothing standard fits, each written down before it is built.
Parameters get registered
A parameter not registered as a custom dimension is collected but not reportable, and registration does not backfill. Data from before that decision stays out of the report, which is why the schema is agreed up front.
Kept low in cardinality
Parameters carrying a near-unique value per session flood a dimension and get rolled into a catch-all row. Identifiers belong in a warehouse export, not a reporting dimension.
The measurement plan drives all of it. Each business question maps to a decision someone actually makes, then to the metric that answers it, then to the events that produce it. What survives gets built; what does not is dropped, because every extra event is one more thing to maintain, QA and explain.
Key events, and where conversion counting goes wrong
A key event is an ordinary event marked as one of the outcomes that matter. Marking it is trivial. Defining it correctly is where most setups fail.
The classic fault is a conversion firing on arrival at a thank-you page rather than on a confirmed transaction. Refreshes count again. The back button counts again. A bookmarked URL counts forever. The second is counting method: once per event and once per session produce different totals from identical traffic.
Then there is what happens downstream. Key events imported into Ads become the signal the bidding algorithm optimises towards, so an inflated one does more damage than a missing one and paid media pays twice. Deduplicating across platforms needs a shared transaction ID, agreed at the spec stage.
Purchase, as it is defined
Ecommerce and revenue tracking
The full funnel
view_item, add_to_cart, begin_checkout, add_payment_info, purchase, each carrying item-level parameters: ID, name, category, variant, price, quantity.
Why item detail matters
Parameters you never sent cannot be recovered later. Send them from day one and you can segment by product or margin band afterwards.
Cross-domain and redirects
Where most setups quietly break. A redirect out to a payment gateway and back starts a new session unless both domains are declared, and the sale gets credited to the gateway as a referral.
GA4 and your payment provider will never match exactly. Refunds, retried payments, consent-blocked sessions and ad blockers all remove buyers from GA4 while the money still lands. We do not promise alignment. We narrow the gap and document the expected variance, so a small difference reads as normal and a persistent one reads as a fault. Store-side work runs with our WooCommerce team, and the funnel feeds ecommerce CRO.
The faults that make a property untrustworthy
None of these throws an error. Each one is found by looking for it deliberately.
Self-referrals
Your own payment gateway or login subdomain showing up as a traffic source. Fixed with cross-domain configuration and an unwanted-referral list.
Internal and developer traffic
Office and agency sessions counted as customers, staging events landing in the production property. Both filtered, both re-checked after an office move or VPN change.
Personal data in URLs
An email address in a thank-you URL is personal data sent to Google, which breaches the terms and puts the property at risk. Query strings are stripped before collection.
Page paths that explode
Session IDs and tracking parameters left in the page location turn one page into thousands of rows, long before anyone calls the report broken.
Duplicate collection
Two properties, or a container tag and a hard-coded tag both firing. Traffic looks healthy and every engagement metric is wrong.
Timezone and currency
A reporting timezone that does not match the business, or a currency left at the default. Both make reconciliation impossible until corrected.
Consent, modelling and what GA4 cannot tell you
A good implementation is honest about its ceiling. These limits go into the handover rather than being left for you to discover.
Consent sets the floor
Visitors who decline are not observed. Consent mode models part of the gap, but a model is an estimate drawn from consenting traffic and needs volume to run at all. Smaller properties may never qualify.
Attribution is a model
Credit is assigned under assumptions, not measured. Lookback windows are settings, and GA4 and Ads use different ones by default, which alone is enough to make two accurate systems disagree.
Retention and thresholds
Event-level data is kept for a limited window set in Admin, after which explorations stop reaching back. Low-volume rows can also be withheld to protect identity.
The answer to the last is a BigQuery export, switched on early rather than after the first month you need history: unsampled event-level data on your own retention terms, billed by Google to your cloud account and never marked up. Reconciling the platforms against each other belongs to the analytics and reporting hub.
How the implementation runs
Discovery and plan
Access review, a walk through the site and checkout, measurement plan agreed.
Build in staging
Built in a staging Tag Manager container against the spec. Nothing published live.
QA and publish
Every event checked in DebugView, consent granted and denied. Versioned publish, rollback available.
Validation window
Live data compared against a second source until the variance is stable and explained.
The validation window is the step most implementations skip, and it is why teams stop trusting GA4 six months in. Data that looks fine on day one drifts once real refunds and consent rates hit it, so we hold the project open through it. The blockers are rarely our build time: they are developer availability, your consent platform, and whether checkout sits on somebody else's domain.
What we need: Editor access on the GA4 property, access to the Tag Manager container, the Ads link if conversions are imported, and an export from the payment provider or CRM to reconcile against. Admin rights only if retention or data streams must change.
What the handover contains
The measurement plan mapping business questions to events. The event and parameter dictionary, so nothing gets invented ad hoc later. A record of the property configuration and why each setting was chosen. The QA evidence, and the documented variance against your second source.
Written for whoever inherits it, not for whoever built it: an event dictionary in plain language rather than a screenshot of an admin screen.
Everything lives in your own accounts. We do not show prospective clients another company's property or dashboard. What we can show you is your own, in the free review.
What is included
What a GA4 implementation costs
Two things get confused here: the platform and the work. Google Analytics 4 is free at the standard tier, within the limits Google publishes. GA4 360 is a separate enterprise product most businesses never need. The tool does not cost money. The implementation does.
An audit or a defined build is fixed scope, quoted before it starts, so you never get an open-ended bill. Ongoing work runs as a monthly retainer. Four things move the figure: domains and data streams, whether ecommerce events are in scope, which consent platform you run, and whether we build in your container or your developers ship the dataLayer.
Included: the plan, the build, QA, the validation window, the handover. Billed elsewhere: BigQuery storage and queries, charged by Google to your cloud account and never marked up. An exact quote comes back with the free review.
Why a senior-only team
The person who writes your measurement plan builds the container and takes the call. Nothing is delegated on your property, and no account manager relays questions. It matters here because GA4 misconfiguration is silent: the mistake surfaces months later, when someone compares two numbers.
GDPR and consent are how we build by default. Everything downstream leans on this layer: a conversion rate optimisation test is only as trustworthy as the event underneath it, and so is the reporting on the wider SEO service. Send the property ID, your platform and the conversions you rely on.
Fair questions, straight answers
What do GA4 implementation services include?
More than installing the tag: a measurement plan, event and parameter naming, key events, consent configuration, and validation against a second source before anyone reports the numbers. The tag itself is the first hour.
Does GA4 cost money?
The standard tier is free, within the limits Google publishes. GA4 360 is a separate paid enterprise product most businesses never need. The implementation work is quoted as fixed scope or a retainer. If BigQuery export is on, Google bills storage and queries to your cloud account.
How long does a GA4 implementation take?
A single-domain site is usually a few weeks from plan to publish; multi-domain or ecommerce takes longer. The main variable is developer availability, not our build time. Then add the validation window before the data drives decisions.
What is the difference between a GA4 audit and a GA4 setup?
An audit reviews an existing property and reports what is wrong and what to fix first. A setup builds or rebuilds the configuration from a measurement plan. Most people who think they need a setup need an audit first, and the free review tells you which.
Do I need Google Tag Manager to use GA4?
No. GA4 runs fine on a direct gtag install, which is cleaner for a simple site. Tag Manager earns its place once you have multiple tags, custom events, consent logic or more than one person making changes, and that side of the build is Google Tag Manager consulting. Semrush has a fair primer.
Why doesn't GA4 revenue match my payment processor?
Because the two count different things: refunds, failed payments, consent-blocked sessions, ad blockers and offline orders all create gaps. A good setup narrows the gap and documents the expected variance. A gap nobody can explain means a broken implementation.
Why are some GA4 numbers marked as modelled?
Because visitors who declined consent are not observed, and Google estimates part of that gap from the behaviour of consenting traffic. Modelled figures are estimates, not recovered data, and modelling needs volume before it runs at all. We report which numbers are modelled rather than presenting them all as measured.
Can you migrate us off an old or duplicated Analytics setup?
Yes, including cases where several properties have collected overlapping data for years. We document what each was measuring, so history stays interpretable instead of being written off. Old properties keep collecting until the new setup is validated.
What access do you need from us?
Editor level on the accounts you already own, listed in full under how the implementation runs. We create nothing in our own name, and nothing is published without your sign-off.
Can you implement GA4 for a casino or iGaming brand?
Yes, including registration, deposit and multi-brand event structures that standard templates do not cover, a vertical we already work in through our iGaming SEO work. Before launch we flag what must be checked against Google Analytics terms and your licence. Final sign-off stays with your legal team.