GA4 & Looker Studio Consultant

Related case study

What a GA4 consultant actually does

A GA4 consultant is not a reporting clerk. The job is to make the numbers inside Google Analytics 4 agree with what actually happened in your business, and then make those numbers legible enough that somebody acts on them.

Most engagements start the same way. A marketing lead opens a report, sees a conversion count, compares it against what the sales team recorded, and finds the two disagree by thirty percent. Nobody can say which one is wrong. Budget decisions stall, because no one wants to defend a channel using a figure they cannot vouch for.

Fixing that is digital analytics work, not reporting work. It means auditing how data collection actually happens on your site, rebuilding the event structure so it maps to outcomes the business already cares about, and only then wiring up the reporting layer. The best Google Analytics experts spend most of their hours upstream of the dashboard, in the tagging and the taxonomy, because that is where the errors are born. A dashboard built on a broken GA4 property is just a faster way to be wrong. This is the part that surprises people who expected a Google Analytics consultant to arrive with a template.

Two things separate a working setup from a decorative one. First, every number has a defined origin you can trace back to a specific event on a specific page. Second, someone can explain in a sentence what each metric is supposed to change. If a KPI cannot pass that second test, it is decoration.

Seven signs your setup is quietly lying

You rarely get an error message. Bad measurement fails silently, and the reports keep rendering. These are the patterns that show up most often in a GA4 audit:

• Duplicate transactions. The purchase event fires on the confirmation page and again on refresh, so revenue reads high and nobody trusts the e-commerce tracking.
• Conversions with no source. A large direct bucket usually means lost referral data, a cross-domain gap, or a redirect stripping parameters, not a surge of people typing your URL from memory.
• Events that fire on page load. A form submission tracked on arrival rather than submission inflates conversion tracking by every visitor who bounced.
• Google Ads and GA4 disagreeing. Different attribution windows and conversion definitions produce different totals. That is expected. A gap nobody can explain is not.
• Internal traffic counted as demand. Your own team, your developers, and your staging environment all landing in production reporting.
• Reports nobody opens. If the standard reports go unread for a quarter, the problem is relevance, not access.
• A property still shaped like the old tool. Sessions and pageviews carried over from Universal Analytics, with none of the event model that makes GA4 useful.

Any one of these is fixable in isolation. Together they mean the measurement layer needs a rebuild rather than a patch, and troubleshooting one report at a time will not get you there.

The GA4 audit comes before the rebuild

Every engagement opens with a GA4 audit, and any GA4 consultant who skips it is guessing. I inventory what is currently collected, what fires it, and whether it can be trusted. That covers the data streams on the property, every tag and trigger in Google Tag Manager, the data layer your developers push to, consent handling, filters, channel groupings, and the conversion definitions that feed Google Ads.

The audit also documents what is missing. Most sites measure the last step and nothing before it. You know how many demo requests came in, but not how many people started the form and abandoned at the phone field, which is the part you could actually fix.

Two constraints shape the work. Google Analytics 4 replaced Universal Analytics on July 1, 2023 for standard properties, and Google has since deleted the historical data, which means no exported archive means no true year-over-year baseline for anyone who skipped the export. If a GA4 migration was rushed, you inherited whatever the automated setup guessed at. That is the usual starting point, and it is worth knowing before anyone quotes you a number.

The output is a prioritized fix list, ranked by what it unblocks rather than by how broken it looks. A miscounted micro-conversion can wait. A revenue figure feeding a bidding strategy cannot. This mirrors how I run a technical SEO audit, and the two often run together, because crawl problems and tracking problems tend to share a root cause.

A measurement plan, then the tagging

Before a single tag changes, we write a measurement plan. This is the deliverable to ask any Google Analytics consultant for first, because it is the one that constrains everything after it. It is a short document, usually two pages, that lists the decisions your team makes on a recurring basis and the evidence each one requires. Every KPI on the list has to trace to a decision. Anything that does not gets cut, which is normally about half the original list.

Only then does GA4 implementation start. The plan translates into a naming convention, a set of custom events with consistent parameters, and a data layer specification your developers can build against without guessing. Consistency matters more than cleverness here: a parameter named three different ways across three templates will cost you more analysis time than the event was ever worth.

Implementation details depend on the stack. A WordPress site usually needs a clean GA4 setup through Google Tag Manager rather than a plugin that hard-codes its own tags. Shopify needs the checkout events reconciled against the platform's own reporting before anyone trusts the e-commerce tracking. Lead generation sites need the form, the call, and the CRM stage stitched into one path, because a lead that converts offline is invisible otherwise.

Where ad blockers, browser restrictions, or consent rules are eating a meaningful share of the data, server-side tagging is worth the added infrastructure. Server-side tracking is not a default recommendation. It adds cost and a container to maintain, so I only propose it when the audit shows the loss is large enough to change a decision.

Everything gets validated end to end before it is called done: fire the event, watch it arrive, confirm it lands in the right report with the right parameters, and check the number against a source outside GA4.

Looker Studio dashboards people actually open

Clean data that nobody reads changes nothing. The reporting layer is where measurement turns into data-driven decisions, and it is where most analytics projects quietly fail, because the default instinct is to build one dashboard containing everything.

I build one page per audience instead. An owner or executive gets revenue, cost per acquisition, and pipeline by channel, and nothing else. A channel manager gets the working view: campaigns, landing pages, funnel-stage drop-off, the numbers they can move this week. Looker Studio dashboards are free, they connect natively to GA4, Google Ads, and Search Console, and they let people self-serve without a login to the analytics property itself.

Custom reports inside GA4 still matter for the ad-hoc questions, and the standard reports are fine for spot checks. But data visualization is a persuasion problem as much as an accuracy problem. A chart that takes forty seconds to interpret will not survive a leadership meeting, and the decision gets made on instinct instead.

For higher volumes or longer retention windows, BigQuery is the answer. GA4 exports to it natively, which removes the sampling and cardinality limits and lets you join marketing data against CRM and financial records. Teams already standardized on Power BI can be served from the same BigQuery layer, so the pipeline does not need rebuilding to match the tool. That matters more than tool preference: whether the final chart renders in Looker Studio or Power BI, the modeling logic lives in one place and stays auditable. Choosing BigQuery early is cheaper than retrofitting it after two years of data you cannot query.

Attribution, and telling the truth about channels

Attribution modeling is where analytics stops being a technical exercise and starts being a political one. Every channel owner prefers the model that flatters their channel, and every model is a simplification.

My position is that you pick a primary model, document why, and hold every channel to it consistently. Then you check it against something outside the model: incrementality tests, geo holdouts, or the blunt instrument of turning spend off in one region and watching what happens. Data analysis that never gets tested against reality is just a well-formatted opinion.

This is also where measurement earns its keep across the rest of a marketing strategy. Reliable user behavior data is what makes CRO work possible, because you cannot prioritize a marketing funnel analysis without knowing where people actually leave. It is what separates PPC waste from PPC investment. It is what lets full-stack SEO consulting be argued in revenue rather than rankings, and it is increasingly what a GEO consultant needs in order to prove that AI-assistant referrals are converting at all, since those sessions often arrive with weak or missing source data.

The end state is one number every channel reports to. Not because a single metric explains everything, but because it forces the arguments to happen in the same units.

What honest measurement changed

Clean Air Columbia is the clearest example. The program covered SEO, PPC, and email, and the reason those channels could be managed as one budget is that all three reported into a single revenue view rather than three separate tool dashboards. Revenue grew 78.3 percent year over year, capped by a record month above $70,000, and paid conversions improved 49 percent month over month once the search terms and negatives were pruned against conversion data we trusted.

The measurement work is rarely the headline in a case study, which is exactly why it gets skipped. It shows up instead as the thing that made the rest defensible. When paid spend was cut on underperforming terms, nobody had to argue about whether the conversions were real. That is the return on hiring a Google Analytics consultant early rather than after the first budget dispute.

The pattern repeats across the portfolio. A rebuild of conversion tracking is what lets an e-commerce SEO consultant prove product-page work paid for itself, and it is the first thing to verify after a replatform, since website migration SEO work is judged entirely against a benchmark that has to be trustworthy before launch. Across 30-plus GA4 migrations and rebuilds, the pre-work has never been the part clients regretted paying for.

Read the Clean Air Columbia case study for the full channel breakdown.

GA4 setup checklist

If you take one thing from this page, take the order. Most failed analytics projects did the right steps backwards.

• Audit first. Inventory the property, the container, the data layer, and every conversion definition before changing anything.
• Write the plan. Decisions first, then KPIs, then events. Never the reverse.
• Name things once. One convention for events and parameters, documented, applied everywhere.
• Validate end to end. Fire it, watch it land, reconcile it against a source outside GA4.
• Report by audience. One page per stakeholder, built around the decisions they own.
• Export to BigQuery. Set it up on day one, even if you do not query it for a year.
• Hand over the keys. Documentation and GA4 training so the setup survives without me.

That last one is deliberate. A good GA4 expert should make themselves unnecessary for the routine work and available for the hard questions. Bringing in a Google Analytics consultant who leaves you dependent on them for a monthly report has solved the wrong problem, and it is worth asking any candidate how the handover works before you sign anything. It is the question most Google Analytics experts hope you will not ask, and the one that best predicts how the engagement ends. Measurement is also the natural companion to a SEO content strategy engagement, since a content program without reliable data is a publishing schedule with no scoreboard.

Metrics this work moves

FAQ