A business operator working at a luminous project desk

NBW Field Guide · Websites · Measurement

GA4 tells you what happened. Clarity helps you investigate why.

Traffic data and behavior data work better together. The useful outcome is not another dashboard—it is a shorter path from “something changed” to “here is what we should inspect.”

Most small-business analytics problems are not caused by a lack of charts. They come from a gap between the question the operator has and the way the tools are organized. Google Analytics 4 is strong at describing acquisition, events, sessions, and key outcomes. Microsoft Clarity is strong at helping a team inspect how people actually interacted with pages through behavior-oriented tools such as recordings and heatmaps. Pairing the two creates a much more useful operating view than treating either as the complete answer.

The useful pairing

GA4 describes the measurable path. Clarity helps investigate the friction inside it.

One tells you where the pattern exists. The other helps you inspect what people experienced around that pattern.

What GA4 and Clarity are actually for

Google Analytics 4What happened, where traffic came from, and which events occurred.

Acquisition reports, events, sessions, user behavior summaries, key events, attribution views, audiences, and reporting dimensions.

Microsoft ClarityWhat the interaction looked like when people used the page.

Recordings, heatmaps, interaction patterns, and a GA-integrated investigation surface for drilling into behavioral context.

Microsoft’s current Clarity documentation explicitly describes a Google Analytics dashboard that lets teams analyze GA data with Clarity features such as recordings and heatmaps. That is the right mental model: the tools can complement one another because they answer different layers of the same website question.

GA4: design the event model around decisions

GA4’s measurement model is event-based. Google provides recommended and standard event patterns for actions such as page views, lead generation, sign-ups, purchases, and many other interactions. The temptation is to track everything because it is possible. The better approach is to define the small set of events that correspond to real customer or operating milestones.

For a service business, those might include a form start, form submission, phone-link click, appointment click, document download, contact CTA, or a thank-you-page confirmation. For an ecommerce business, the meaningful set is different. What matters is that the event names, parameters, and key-event decisions are documented so the report still makes sense six months later.

AcquisitionHow did sessions arrive?

Source, medium, campaign, referral, organic search, direct, paid channels, and other traffic dimensions.

EngagementWhat did people do?

Pages, screens, events, paths, and other measurable interactions in the GA4 event model.

Key eventsWhich actions matter to the business?

Define only outcomes that deserve elevated reporting; not every click should become a “conversion.”

AttributionHow is credit assigned?

Reporting attribution models affect how touchpoints receive credit. Label the lens instead of treating it as causal truth.

Traffic acquisition is not the same as first-user acquisition

Google’s current documentation distinguishes user acquisition from traffic acquisition. That difference is useful in practice: one view focuses on how users were first acquired, while the other focuses on how sessions arrive. An owner asking “where did this week’s traffic come from?” and an analyst asking “what initially brought these users into our audience?” are not asking the same question.

Clarity: behavior becomes an investigation surface

A traffic report can show that a landing page has plenty of sessions and weak completion. It cannot, by itself, show what a visitor did with a confusing form, whether people repeatedly clicked an element that was not interactive, how far users scrolled, or what sequence of visible behavior occurred before abandonment. That is where behavior-oriented investigation becomes useful.

Clarity should not be used as entertainment. Watching random recordings for an hour is not a measurement strategy. Use it after a signal tells you where to look: a drop in a key event, an unusual landing-page pattern, a high-exit page, a campaign that sends traffic but not meaningful actions, or a mobile/desktop discrepancy.

The website diagnosis loop

01DetectGA4 shows a meaningful change or weak path
02SegmentDevice, source, landing page, campaign, or event
03InspectUse Clarity recordings/heatmaps around that segment
04HypothesizeName the likely friction without pretending certainty
05Change & verifyFix the bounded issue and watch the relevant metric/path

Which tool answers which question?

Business questionStart in GA4Then inspect in ClarityWhat not to conclude too quickly
Which channels bring traffic?Traffic acquisition by source/medium/campaignInspect behavior for important source segmentsMore traffic does not automatically mean better leads
Why did form completion drop?Form-start and submit events; page/device/source segmentsRecordings/heatmaps around the form journeyA visual pattern is evidence for a hypothesis, not universal proof
Which landing page needs attention?Landing-page sessions, engagement, key eventsScroll/click patterns and recordingsHigh exit can be normal depending on the page’s job
Are mobile users struggling?Compare device segments and relevant eventsInspect mobile recordings and interaction patternsDo not infer technical root cause without reproduction
Did a campaign create useful behavior?Campaign/source → sessions → events/key eventsInspect the experience for that campaign’s landing trafficWebsite behavior is not the same as closed revenue

Build a measurement plan before adding more tags

A measurement plan is a small table connecting a business question to an event, source, owner, and expected use. If nobody can name the decision a metric supports, it probably should not dominate the dashboard.

QuestionWhat are we trying to decide?

Example: Are visitors reaching the quote form and completing it?

EvidenceWhich event or report answers it?

Form start, submit, landing page, source, device, error state if instrumented.

InvestigationWhat would make us open Clarity?

A meaningful drop, segment disparity, or unexpected path—not curiosity alone.

ActionWho owns the next move?

Content, design, dev, campaign, operations, or “no change; monitor.”

UTM governance still matters

GA4 cannot clean up a naming convention that never existed. Standardize how source, medium, campaign, and content parameters are created. A campaign called “Summer Promo” in one tool, “summer-promo” in another, and “SUMMER2026” in a third creates reporting fragmentation before analytics ever starts.

A good diagnosis separates signal, evidence, and inference

Three evidence levels

Observed → Investigated → Inferred.

Observed: form submits fell on mobile. Investigated: recordings show repeated interaction around one field. Inferred: the field may be creating friction. Reproduce and test before calling it the root cause.

This discipline prevents analytics work from becoming confident storytelling. GA4 and Clarity are powerful because together they reduce the distance between a numerical pattern and a visible experience. They still do not eliminate the need for QA, reproduction, user context, or business judgment.

Behavior analytics needs a privacy and sensitive-data boundary

Session-replay and behavioral tools can create privacy risk if implemented carelessly. The configuration should exclude or mask sensitive fields, respect the site’s consent posture, avoid collecting information that should never be replayed, and align with the organization’s privacy disclosures and policies.

The same rule applies to GA4: data minimization is useful. Do not send personally identifiable information into analytics fields that are not designed or permitted for it. Keep customer identity and sensitive business records in systems intended to handle them.

A weekly website visibility review can fit on one page

TrafficWhat materially changed?

Source mix, campaign traffic, landing pages, device mix, unusual rises or drops.

Key actionsDid important events move?

Forms, calls, bookings, signups, purchases, or other real milestones.

FrictionWhere should we inspect behavior?

Pick one or two high-value segments for Clarity—not the whole site.

DecisionWhat happens next?

Fix, test, investigate further, monitor, or explicitly do nothing.

AI can summarize the review; it should not invent the evidence

An AI-assisted reporting workflow can retrieve approved analytics data, compare it with prior periods, summarize material changes, and send a short brief so an owner does not have to remember to open multiple dashboards. That is useful when the automation is read-only and the source/freshness of each metric is visible.

The safe boundary is important: an AI summary should not claim that a UX issue caused a revenue loss unless the evidence supports that conclusion. It can say, “mobile form submissions fell 18%; recordings in the selected segment show repeated interaction around the date field; inspect the field implementation.” That is much more useful than pretending it proved causality.

Illustrative sample — not client work.

A concise weekly brief might say: “Organic sessions rose; quote-form completion stayed flat; mobile form starts increased but submits declined; Clarity inspection shows repeated field interaction on mobile; no production change recommended until the issue is reproduced.” The format demonstrates the standard, not an NBW customer result.

A practical 30-day setup

Week 1

Define business questions and audit existing tags

List key actions, traffic questions, current GA4 property/tag state, consent posture, and any duplicate or legacy measurement.

Week 2

Implement the event and campaign naming plan

Configure or clean up the events the business will actually use, document UTMs, and verify traffic-source reporting.

Week 3

Install or validate Clarity and sensitive-data controls

Confirm recordings/heatmaps work as intended, excluded content is protected, and the GA integration is configured where useful.

Week 4

Run a synthetic journey and create the operating review

Test important paths, validate events, verify reports, inspect one real behavioral segment, and document the weekly review template.

Sources & further reading

This guide uses current Google Analytics and Microsoft Clarity documentation for product mechanics and NBW recommendations for how to operate the combined system. Interfaces, reports, and integrations change; verify current documentation during implementation.

Need your website data made useful?

See the traffic. Investigate the behavior.

Nashville Business Works can scope a bounded Website Visibility setup using GA4 and Microsoft Clarity, including event planning, source hygiene, QA, and a practical handoff.

Explore Website Visibility →