A GA4 Implementation Agency That Finishes and Hands It Over
A GA4 implementation agency delivers a defined project: an audit of what is tracking today, an agreed event taxonomy and data-layer spec, GA4 and Google Tag Manager built or corrected against that spec, conversions tied to revenue rather than clicks, consent handling that holds up, BigQuery export switched on, and documentation your next developer can actually read. ULEY runs this as a fixed-price Automation Build with an end date and an acceptance test, typically 3-6 weeks — not an open-ended retainer. We hold no Google certification or partner status, and none is required to do this work.
What GA4 & Google Tag Manager is
GA4 and Google Tag Manager are two separate Google products, usually implemented together, that do genuinely different jobs — and confusing them is the source of most of the confusion around both. Google Analytics 4 is the measurement and reporting product. Its data model is event-based: a page view, a purchase, a form submit, a video play are all events carrying parameters, which replaced Universal Analytics' session-and-page-view model when Google shut that product down in 2023 and 2024. Google Tag Manager is not analytics at all. It is a tag-management container: a snippet on the site that decides which third-party tags — GA4, Google Ads, Meta, LinkedIn, and others — fire, on which trigger, with which variables, so tracking changes do not require a developer deployment each time. The practical consequence is that most problems people describe as GA4 problems are Google Tag Manager problems: the data model is behaving correctly and the trigger is wrong. GA4 is free below a high event volume, with GA4 360 as the paid enterprise tier; Google Tag Manager is free. Both include a raw event export to BigQuery, which GA4 offers at no cost and which is not retroactive — it only captures data from the day it is enabled. Because both are configuration products rather than code you own, implementation quality lives almost entirely in decisions — what an event is called, what counts as a conversion, what happens when consent is denied — rather than in the tools, which is why two correct-looking installations can produce numbers that disagree.
What we do with it
- Implementation audit: what is firing today, what is firing twice, what stopped firing after a site change nobody connected to the drop. Written findings, naming what is already correct and should be left alone.
- Event taxonomy and data-layer spec agreed in writing before any tag is built or rebuilt — the naming convention, the event dictionary, and which system is the source of truth for each conversion. This is the deliverable that outlives the engagement.
- GA4 and Google Tag Manager built or corrected against that spec, including trigger logic, variables, and the cleanup of legacy tags nobody has dared delete.
- Conversion definitions tied to revenue or a booked job rather than to a click — defined once, in writing, so marketing and sales stop reporting different numbers from the same week.
- Ecommerce tracking where it applies: the full purchase funnel with item-level data structured so it reconciles with the store, not just a purchase event that fires.
- Consent Mode configuration alongside your consent platform, so measurement behaves correctly and legally when a visitor declines — decided deliberately rather than left at whatever the banner installed.
- Server-side tagging where the case for it is real, with an honest read on the added infrastructure cost and maintenance burden it brings.
- CRM and ad-platform connections, including offline conversion import, so the ad account optimizes toward outcomes that closed rather than forms that submitted.
- BigQuery export enabled and a Looker Studio reporting layer built around the decisions it has to drive.
- Documentation, a QA pass against the spec, and a handover session — the acceptance test for this project is your team being able to read and change it without us.
Who this is for
Businesses with a live site, real spend, and a GA4 property whose numbers nobody quite believes. The most common shape: GA4 was installed in a hurry before Universal Analytics shut off, it has technically worked ever since, and no one has read it since. Also a fit: a site replatform or redesign that is about to break tracking, where the sensible move is to specify measurement before the launch rather than reconstruct it after; a tracking setup inherited from a departed agency or contractor with no documentation; and a team about to buy attribution or a data warehouse who need the event layer underneath it to be trustworthy first. Across the four lead verticals in strategy/ICP.md this is foundational rather than vertical-specific — multi-location home services groups needing calls tied to booked jobs, DTC brands needing item-level ecommerce data that reconciles with the store, professional services firms needing attribution that survives a six-month consideration window, and course creators needing to see which source produces an enrollment rather than a registration.
Who this isn't for
You do not actually need this. Stated first because it is the case an implementation agency has every incentive not to raise: if you run a single-page site with one contact form, the form tool's own notification email already tells you everything a GA4 build would, and a tracking project would be ceremony. The same is true if you are pre-launch with no traffic and no spend — there is nothing to measure yet, and a measurement plan written now will not survive contact with the product you actually ship. Hardest and most common version: if nobody on your team will change a decision because of a number, better measurement will not produce better decisions. It will produce a dashboard nobody opens. That is worth naming before you spend, not after. Also not a fit: your question is who maintains tracking as consent rules and platforms keep changing — that is the ongoing capability, and the web analytics service page owns it, not this one. You want a checklist to run the implementation yourself — that is a different and legitimate need, and this is a commercial page, not a how-to. You are shopping on lowest price alone. You need enterprise analytics — a warehouse team, contractual SLAs, GA4 360 administration at scale — which sits above what a 13-person remote team should take on. You are in a heavily regulated vertical such as healthcare or financial services requiring compliance infrastructure ULEY has not published or verified.
What GA4 & Google Tag Manager connects to
| Integration | What it does |
|---|---|
| Google Tag Manager | The container where the implementation actually lives. Whether tags run through it or are hardcoded into the site is the single decision that determines what every future tracking change costs you. |
| Google Ads | Conversion import and audience sharing, plus offline conversion import from the CRM so bidding optimizes toward deals that closed rather than forms that were filled in. |
| Google Search Console | Organic query and landing-page data surfaced inside GA4, with its real limitations stated: the Search Console reports are sampled and dimension-limited, and they will not reconcile exactly with GA4 sessions. |
| BigQuery | Raw, unsampled event export — free from GA4 and the thing that makes your history portable. It is not retroactive, so it captures nothing before the day it is enabled. We turn it on early in every build for that reason alone, whether or not anyone plans to query it yet. |
| Looker Studio | The reporting layer, built around the decisions each report has to drive rather than around every metric GA4 exposes. |
| Consent management platform | Consent Mode wired to your banner so measurement and ads personalization behave correctly when consent is declined. For advertisers serving users in the EEA and UK, Consent Mode v2 signals are a condition of using audience and remarketing features, not an optional refinement. |
| CRM (HubSpot, Salesforce and similar) | Offline conversion import in both directions, so a lead in the CRM and a conversion in the ad account are provably the same event rather than two plausible numbers. |
| Server-side tagging | A second container running in your own cloud, which improves signal durability and moves data handling under your control — at the cost of real infrastructure spend and ongoing maintenance. We will tell you when that trade is not worth making for your volume. |
| Call tracking | Call events tied to a source and to a CRM record, so a phone-forward business can attribute a booked job rather than a ring. |
Migrating on or off
Moving on, from Universal Analytics: the honest fact is that Universal Analytics data does not migrate. The two products have incompatible data models, and nothing carries over. What you can do is archive the old data — export to BigQuery or Sheets before access lapses — and annotate the changeover date so future reports show a documented seam rather than an unexplained cliff. Anyone offering a merged historical view across both is offering something the products do not do. Moving on, from another platform such as Adobe, Matomo, Piwik, or Plausible: same story, plus a specific expectation to set. The numbers will not match, in either direction, because the tools define sessions, users, and attribution differently. Plan a parallel-run window and a short written explanation of why the two disagree, because the alternative is spending your first quarter on GA4 defending it. Moving off GA4 later: this is where the architecture you choose now decides the cost. If tracking runs through a Google Tag Manager container against a documented data layer, replacing GA4 with something else is a container change — the site keeps emitting the same events and a different tag consumes them. If tags are hardcoded into the site templates, the same swap becomes a developer project across every page. That difference is the strongest practical argument for the container-plus-data-layer pattern, and it is the part most rushed implementations skip because it costs more on day one and nothing on day thirty. BigQuery export is the other half: it is what makes your history yours rather than the vendor's, and it captures nothing retroactively. When migrating is a bad idea: switching analytics platforms because you do not trust the numbers. In almost every case the definitions were never agreed, not the tool wrong — marketing counts one thing as a conversion, sales counts another, and no document says which is correct. Move to a new tool with the same undefined conversions and you get new numbers you also do not trust, minus your history. Fix the taxonomy first. If the numbers are still wrong afterward, that is a real finding and worth acting on.
What ULEY Actually Charges
Where most GA4 and Google Tag Manager implementations sit: a fixed-scope project — audit, event taxonomy and data-layer spec, build or correction, conversions, consent, BigQuery, dashboard, documentation and handover — priced and agreed before we start. Trade-off: a defined end date and an acceptance test, which is the point; this is a project that finishes rather than a retainer that renews. For orientation only, 2026 US market research puts GA4 and Google Tag Manager implementation with dashboards at $3,000-15,000+. Market range, not ULEY's price — our published rate card is the only ULEY number on this site.
For a property you inherited and cannot diagnose: full access, a written read of what is firing, what is double-counting, and what silently stopped, plus a prioritized plan. Trade-off: findings and a roadmap, not a rebuilt implementation — and handing that roadmap to your own developer is a completed engagement, not a failed sale. Credited toward a Growth Retainer if you continue.
For after the project ends, if you want it maintained rather than owned in-house: tracking checked against platform and consent changes, reporting kept current as the questions change. Trade-off: an ongoing line item instead of a one-time cost. This is the ongoing measurement capability the web analytics service page covers in full — noted here so the boundary is explicit rather than blurred. 2026 US market research puts advanced analytics and attribution programs at $15,000-30,000/month at the enterprise end, which is above ULEY's band and named here so nobody mistakes our tier for it.
For a narrow, bounded job: one tag that broke after a site change, a consent configuration review, or a second opinion on an implementation someone else delivered. Trade-off: hours against a task list you define, with no owned outcome attached.
GA4 & Google Tag Manager — Common Questions
No to both, and we will not imply otherwise. The useful version of this answer is what those credentials actually measure. The Google Analytics certification is a free, open-book, multiple-choice exam that a motivated person passes in an afternoon; it establishes familiarity with the interface, not the ability to design an event taxonomy. Google Partner status is an Ads program, earned on ad spend under management, certifications, and performance thresholds — it says nothing about analytics implementation quality, and it is common for a badged Ads agency to have no analytics engineering depth at all. Both are statuses granted by the vendor, and neither is independent evidence that a specific build will be correct. What to check instead, of us and of anyone certified: ask to see a real event taxonomy document and data-layer spec from a delivered project, and ask them to explain, out loud, what happens to your conversion data when a visitor declines consent. Those two questions separate implementers from installers faster than any badge.
Time-boundedness, and it is a real distinction rather than a naming preference. This page is a project: it has a start, an end, an acceptance test, and a handover document, and when it is done it is done. The web analytics service is the ongoing capability — measurement that keeps working as consent rules, platforms, and business questions change, plus the reporting and attribution work that sits on top. Many clients buy the project and never buy the capability, and that is a complete outcome, not an unfinished funnel.
Everything the numbers need in order to be trustworthy, and nothing that exists only to pad a scope. Concretely: an audit of what is tracking today; the taxonomy and data-layer spec agreed before anything is built; the GA4 and Tag Manager build itself; conversions defined against revenue or a booked job rather than clicks; ecommerce tracking where it applies; Consent Mode matched to your banner; CRM and ad-platform connections including offline conversion import; BigQuery export; a Looker Studio reporting layer; a QA pass; and a handover session with documentation your next developer can read. Server-side tagging is in when the case for it holds up, and we will tell you when it does not.
No. We audit what is live first and put the findings in writing, including the parts that are correct and should not be touched. Most inherited properties are perhaps two-thirds sound with a specific set of gaps — a rebuild that discards the working two-thirds is more expensive, more disruptive, and easier to sell, which is exactly why it is worth resisting.
No, and nobody can make it. The two products have incompatible data models and there is no supported path that merges them into one historical view. What is worth doing is archiving the old data before access lapses and annotating the changeover date, so future reports show a documented seam rather than a cliff no one can explain.
Sometimes, and less often than it is sold. It genuinely improves signal durability against browser and consent restrictions and moves data handling under your own control. It also adds real cloud infrastructure cost and a maintenance burden that does not go away. For a lot of businesses the browser-side implementation, done properly with a clean data layer and correct consent handling, closes most of the gap for none of the recurring cost — and we will tell you when that is your situation.
Because they are counting different things, on purpose. Attribution windows, the credit model, and how each platform treats consent and modeled conversions all differ, so the numbers should not match and a build that makes them match is usually hiding something. The work is deciding, in writing, which system is the source of truth for which decision — then reporting from that, rather than reconciling two tools forever.
ULEY has one published rate card and this maps onto it — most implementations run as a fixed-scope Automation Build, with the Audit Sprint as the diagnostic entry point. Figures are on the pricing page. For market context only: 2026 US market research puts GA4 and Google Tag Manager implementation with dashboards at $3,000-15,000+, and enterprise analytics and attribution programs at $15,000-30,000/month. Those are market ranges cited for orientation, and the enterprise band is above what we take on.
That is the acceptance test, not a bonus. You get the event taxonomy, the data-layer spec, the container documentation, and a handover session, and the project is finished when your team or your developer can read the setup and change it without calling us. If you would rather we kept maintaining it, that is the Growth Retainer and it is a separate decision you make after the project, not a condition of starting it.
Ready to see where your budget leaks?
Free 30-minute audit, written roadmap included. No contracts.
Get My Free Growth Audit