Chapter 28: Google Analytics 4 — Measuring Organic Traffic, User Behavior, and Conversions

"Not everything that can be counted counts, and not everything that counts can be counted." — William Bruce Cameron, Informal Sociology (1963) — the line the whole internet misattributes to Einstein, which is itself a small lesson in checking your sources.

Overview

In the last chapter we stood up Google Search Console (GSC) and learned to read what happens on Google — what you rank for, what you're almost ranking for, which pages are indexed. But Search Console's story stops at the click. It can tell you that 900 people saw your water-heater page in the results and 40 of them clicked. It cannot tell you what those 40 people did next: whether they read the page or bounced in disgust, whether they found the booking form, whether any of them became a phone call, a lead, a paying customer. The moment someone clicks your result, they leave Google's property and arrive on yours — and to see what happens there, you need a different instrument.

That instrument, for most of the web, is Google Analytics 4 (GA4). This is the chapter where SEO stops being about visits and starts being about value — where "we grew organic traffic 30%" becomes the far more important sentence, "organic search generated 52 booking requests last month, and here's what that's worth." For a business owner deciding whether to keep funding SEO, that second sentence is the entire conversation. Traffic is a means; the phone ringing is the end.

But there is a second, quieter theme running through everything here, and this book would be dishonest to hide it: the numbers are not as solid as they look. GA4 will hand you figures accurate to the decimal — 1,284 sessions, a 61.3% engagement rate — and a large fraction of the industry treats those decimals as gospel. They are not gospel. They are a modeled, consented, ad-blocked, cross-device-fragmented approximation of reality, and a professional reads them as a compass, not a ledger. Learning GA4 well means learning both halves at once: how to extract real, decision-grade insight from it, and exactly how much to trust what it tells you.

In this chapter, you will learn to:

• Think in GA4's native language — events and parameters — instead of the pageviews-and-sessions model you may remember from the old Analytics.
• Isolate organic search from every other traffic source, and read its behavior separately.
• See which landing pages actually earn and satisfy organic visitors.
• Read engagement metrics for what they are — and refuse the myth that they feed your rankings.
• Define and configure the conversions that matter for a service business: forms, calls, and bookings.
• Understand attribution well enough to know why the last click lies about SEO.
• Measure honestly in a privacy-first world — respecting consent, and knowing the limits of the count.

Learning Paths

📊 Strategist: this is a core chapter — the whole discipline of proving SEO's value lives here, and §28.5 (conversions) and §28.6 (attribution) are your center of gravity. 🏪 Local Business: focus on §28.5 (calls and form fills are your conversions) and the honest limit in §28.7 — and note the hard truth in §28.5 that a call placed from your Google Business Profile never reaches GA4 at all. 🛒 E-Commerce: §28.5's purchase event and revenue tracking are your bread and butter; skim the lead-gen framing and read the e-commerce notes. 📝 Content Creator: §28.3 (landing pages) and §28.4 (engagement) tell you which content actually holds people; attribution (§28.6) matters less if you don't sell directly. 🔧 Developer: you'll often be the one implementing events and Consent Mode — §28.1 (the event model) and §28.5 (configuration, and the no-PII rule) are where you earn your keep, and §28.7 is the compliance conversation you'll be dragged into.

28.1 GA4's model: everything is an event

If you used Google Analytics before roughly 2023, you learned a model built around pageviews and sessions, with "events" as a separate, secondary kind of data you bolted on for clicks and downloads. That product — Universal Analytics (UA) — is gone. Google stopped processing data in standard Universal Analytics properties on July 1, 2023, and replaced it with Google Analytics 4, a ground-up redesign with a fundamentally different data model. This isn't a cosmetic reskin, and pretending it is will confuse you at every turn. The single most important thing to understand about GA4 is this:

In GA4, everything is an event. A page view is an event (named page_view). A scroll is an event (scroll). A click on a link that leaves your site is an event (click). A video play, a file download, a search on your site, a form submission, a purchase — every one of them is an event, a discrete record that something happened, carrying parameters that describe it. There is no longer a special "pageview" hit type sitting above the others. The pageview got demoted to just another event, and once that clicks into place, the whole system makes sense.

THE GA4 DATA MODEL                                        [schematic — not to scale]

   USER  ──has──▶  SESSION  ──contains──▶  EVENTS  ──described by──▶  PARAMETERS
  (a person,     (a visit; starts       (page_view, scroll,        (page_location,
   tracked via    with session_start,    click, form_submit,        page_title, link_url,
   a client ID)   times out after        video_start, purchase…)    value, currency…)
                  30 min idle)                  │
                                                ▼
                              some events you MARK as a  ★ KEY EVENT
                              (a "conversion" — the thing that matters to the business)

Walk that chain from right to left and you have the mental model that makes every GA4 report legible. A user (a person, or more precisely a browser/device Google recognizes by a random client identifier) has sessions (visits). Each session contains events. Each event carries parameters that give it detail — which page, which link, how much a purchase was worth. And a handful of those events you will single out as key events: the ones that represent real business value. That's the spine of §28.5.

Because everything is an event, GA4 arrives already collecting a lot without any work from you. It's worth knowing the four tiers, because they determine how much setup a given thing needs:

The practical upshot for an SEO: a freshly-installed GA4 with enhanced measurement switched on (it's on by default) is already capturing pageviews, scrolls, outbound clicks, site searches, and — increasingly — some form interactions, with no code. That covers a surprising amount of what you need. The work in §28.5 is mostly about the last two tiers: telling GA4 which of these events count as conversions, and occasionally adding one it doesn't capture on its own.

🔎 How Search Sees It Here's what's happening under the hood when someone lands on Rivertown's site. A small piece of Google's tracking code (the Google tag) sits on every page. When the page loads, it fires a page_view event and bundles in parameters — the URL, the page title, the referring source, a random client ID stored in a cookie. As the visitor scrolls, clicks, or submits a form, more events fire, each timestamped and tied to that same session. GA4's servers collect this stream and assemble it, after the fact, into the sessions and users you see in reports. Two things follow from this that trip people up. First, the "session" is a reconstruction, stitched together from events by rules (a new session after 30 minutes of inactivity) — not a thing that objectively exists. Second, if the code doesn't fire, the event doesn't exist — an ad blocker, a declined cookie banner, or a visitor who leaves before the script loads is simply invisible. GA4 doesn't estimate what it missed by default; it just never knew. Hold that thought hard; it is the whole of §28.7.

And now the single most important myth to kill before we go one step further, because it distorts how half the industry uses this tool.

🚫 SEO Myth: "Google uses my Google Analytics data to rank my site — so a low engagement rate or high bounce rate is dragging down my rankings." No. Google's Search representatives have stated, repeatedly and on the record over many years, that Google Search does not use Google Analytics data as a ranking signal. Think about why for a second: the vast majority of sites don't run GA4 at all, or run a competitor, or block it — Google can't build a ranking system on a signal it only has for some sites and that site owners fully control and could game trivially by faking events. Analytics is your instrument for understanding your visitors; it is not a wire feeding your numbers back to the ranking systems. This matters practically: do not "optimize your bounce rate to rank higher." That is optimizing a gauge you installed on your own dashboard in the hope the road will change. Improve the page so it genuinely serves people — and the engagement metric will improve as a symptom of that, not as a cause of ranking. (The related idea — that Google measures how long you linger on a result, "dwell time" — is unconfirmed and belongs in Chapter 2's evidence bin, not in your GA4 workflow.)

28.2 Isolating organic traffic: finding SEO in a sea of channels

GA4 shows you all your traffic — organic search, paid ads, social, email, referrals, direct. For SEO, the first skill is surgical: pull organic search out of that mix and look at it alone. Everything you want to know — did SEO grow, which organic pages perform, do organic visitors convert — requires first drawing a clean boundary around organic traffic.

GA4 sorts every visit into a channel using its Default Channel Group, and the one you care about is Organic Search — sessions that arrived from the unpaid results of a search engine (Google, Bing, and the rest). Note the two boundaries baked into that definition. It excludes Paid Search (clicks on ads — a different channel), and it lumps all search engines together under one label. It also sits beside Organic Social, Referral, Direct, and the others. Your job is to filter the report down to Organic Search and, when you need to, down to Google specifically.

The primary report for this lives under Reports → Acquisition → Traffic acquisition. One distinction here is worth getting exactly right, because choosing the wrong report quietly corrupts your SEO numbers:

TWO ACQUISITION REPORTS — CHOOSE DELIBERATELY           [schematic]

  USER ACQUISITION            TRAFFIC ACQUISITION   ◀── use THIS one for SEO
  "Where did this user        "Where did this
   FIRST come from, ever?"     SESSION come from?"
        │                            │
   user-scoped                  session-scoped
   (first touch, sticky)        (each visit attributed to its own source)
        │                            │
   good for: "how do people     good for: "how much did ORGANIC drive
   first discover us?"           THIS month, and how did those visits behave?"

Traffic acquisition is session-scoped: each visit is credited to the source that brought that visit. User acquisition is user-scoped: it credits every one of a person's visits to how they first arrived, ever. For measuring ongoing organic performance — "what did organic search do this month?" — you want Traffic acquisition, filtered to Session default channel group = Organic Search. (User acquisition is useful for a different, first-touch question we'll meet in attribution, §28.6.) Confuse the two and you'll credit organic with visits it didn't drive, or rob it of ones it did.

🛠️ Try It on Your Site Open GA4 → Reports → Acquisition → Traffic acquisition. Find the primary dimension (it defaults to "Session default channel group") and confirm Organic Search is one of the rows. Click it, or use the "Add comparison / filter" control at the top to filter to Organic Search only. You're now looking at pure SEO traffic. Note three numbers for the last 28 days: Sessions, Engaged sessions, and (once you've done §28.5) Key events. Write them down — that's your organic baseline, and next month's version of this exact view is your first honest answer to "is SEO working?" The tool is free; you already have it if you did Chapter 27.

For anything beyond a quick look — comparing organic against another channel, slicing organic by landing page and device at once, building a reusable view — GA4's standard reports get cramped, and you graduate to the Explore section (Reports' more powerful sibling). There you can build an organic segment: a saved, reusable definition of "sessions where the channel is Organic Search," which you can drop into any exploration so every metric and breakdown is automatically scoped to SEO. Building that segment once and reusing it everywhere is the small habit that separates fumbling through GA4 from working it. (Defining an "organic segment" is our term for exactly this reusable filter.)

Two refinements make that organic view sharper. First, add Session source / medium as a secondary dimension to split "Organic Search" into google / organic, bing / organic, and the rest — useful when you want Google alone, or when a suspicious spike turns out to be a single scraper engine. Second, mind the channel next door: Direct. In GA4, "Direct" is less a real source than a junk drawer — it collects visits GA4 couldn't attribute at all (someone typing the URL, yes, but also untagged links, some app referrals, and sessions where the referrer was stripped by privacy tooling). For an SEO this matters in a specific way: a rise in branded organic — people searching "Rivertown Home Services" after seeing a truck — often shows up partly as Direct, because a returning fan may bypass search entirely. So Direct and branded organic tend to move together, and a chunk of what looks like Direct is really the downstream echo of SEO and brand-building doing their job. Don't dismiss Direct as "not SEO"; read it as SEO's shadow.

Now the honest limit, and it's a big one — the single most important thing to understand about GA4 for SEO:

⚖️ Evidence Check Claim: "GA4 tells me which keywords bring my organic traffic." Flatly false, and the reason is historically important. — What GA4 shows: that a session came from Organic Search, and often the search engine (google / organic, bing / organic). Which page they landed on, how they behaved, whether they converted. Rich post-click detail. — What GA4 never shows: the search query — the actual words the person typed. Since Google moved search to HTTPS and secured its query data (beginning in 2011, the era of "(not provided)"), analytics tools have been blind to the organic keyword. GA4 inherited that blindness by design. It knows that organic search sent someone; it does not know what they searched. — The resolution: this is precisely the seam where GA4 and Search Console fit together. GSC knows the query and the click (pre-click, on Google); GA4 knows the behavior and the conversion (post-click, on your site). Neither tool alone tells the whole story, and no tool joins a specific query to a specific person's on-site behavior. You reason across the two. Anyone selling you "your exact organic keywords and what they earned, per visitor" is selling a fiction the platforms deliberately prevent.

🔗 Connection The pre-click half of the picture — impressions, clicks, click-through rate, average position, and the all-important query — lives in Chapter 27 (Google Search Console). GA4 is the post-click half. Chapter 29 (SEO Reporting) is where we combine the two into a coherent story for a stakeholder, and Chapter 30 is where third-party tools try (imperfectly) to fill the keyword gap GA4 leaves.

28.3 Landing-page performance: which pages earn organic visitors

Once you can isolate organic traffic, the most SEO-useful cut of it is by landing page — the page where a visit began. For search, the landing page is almost always the page that ranked and got clicked, which makes the organic landing-page view a direct readout of your SEO at the page level: which of your pages are actually pulling people in from search, and what those people do once they arrive.

Find it under Reports → Engagement → Landing page (or build it in Explore with the "Landing page" dimension), then scope it to organic with your segment or a channel filter. The result is a ranked list of your entry pages with, beside each, the metrics that tell you whether the traffic is any good: sessions, engaged sessions, engagement rate, average engagement time, and key events. This is where "traffic" becomes "traffic that does something."

📄 Read the Report — organic landing pages, read like a diagnosis

text FIGURE 28.1 — "Four organic landing pages, four different problems" [constructed teaching example] THE REPORT GA4 Landing page report, scoped to Organic Search, last 28 days (Rivertown). WHAT'S THERE PAGE SESS. ENG.RATE AVG ENG.TIME KEY EVENTS /water-heater-replacement 520 67% 1m 55s 9 /blog/why-furnace-smells-burning 880 71% 2m 30s 1 /locations/cedar-hills 240 34% 0m 22s 2 /book 90 58% 1m 05s 38 WHAT IT SHOWS /book converts hard (38 key events from 90 sessions) — it's the finish line, not a discovery page. The furnace blog pulls the most organic sessions and holds them (2m30s) but almost nobody converts — classic top-of-funnel informational intent. The Cedar Hills location page is the alarm: 34% engagement, 22 seconds — people arrive and leave. Something is wrong with that page or its intent match. WHAT IT DOESN'T It doesn't say WHY Cedar Hills fails (thin content? slow? wrong intent? — that's a Chapter 25 / Chapter 9 investigation), and it can't tell you what those furnace-blog readers searched (that's GSC). It shows symptoms, not causes. THE MOVE Add a clear next step (a "book a tune-up" call to action + internal link) to the high-traffic furnace blog to capture some of that held attention; open a real investigation into the Cedar Hills page. Leave /book alone — it's working. THE LESSON Read landing pages as a portfolio: discovery pages earn attention, money pages earn conversions, and a low-engagement page with real traffic is a leak to plug.

Notice what that reading refuses to do: it does not treat "most sessions" as "best page." The furnace blog wins on traffic and loses on conversions, and that's not a failure — it's an informational page doing an informational job, and the fix isn't to mourn its conversion rate but to build a gentle bridge from it toward the money pages. The /book page, by contrast, would look like a failure on a traffic-only report (a measly 90 sessions) and is in fact the most valuable page in the set. This is why traffic-only reporting misleads, and why we spend the rest of the chapter on behavior and conversions rather than raw counts.

The report gets more diagnostic the moment you add a secondary dimension. Landing page × device category often reveals that a page holding desktop visitors fine is hemorrhaging mobile ones (a Core-Web-Vitals or layout problem — Chapters 16, 17). Landing page × new vs. returning tells you whether a page wins strangers or only serves regulars. Landing page × session default channel group confirms a page's traffic really is organic and not, say, an email blast you forgot about. Each split turns a flat ranking into a why, and none of them costs anything but a click. The discipline is to let a surprising number in the top-level report — a page that under- or over-performs its neighbors — drive which second dimension you reach for, rather than slicing at random.

One technical caution particular to landing pages in GA4. Early GA4 was genuinely bad at landing-page reporting (the dimension was missing at launch and returned later), and you may still hit rows labeled (not set) — sessions where GA4 couldn't determine the landing page, often because the tag didn't fire on the true entry page or the session was reconstructed oddly. A handful of (not set) is normal noise; a large share of it means a tracking problem worth fixing before you trust the report. As always: verify the instrument before you trust its readings.

28.4 Engagement metrics: what "engaged" actually means

Universal Analytics measured (dis)engagement mainly through bounce rate, defined as the share of visits that saw one page and did nothing else. It was a crude, much-abused metric — a person could read your entire 2,000-word article, get exactly what they needed, and leave, and UA would count that satisfied reader as a "bounce." GA4 threw that definition out and rebuilt engagement around a positively-defined idea. You need its exact definitions, because they are specific and widely misquoted.

An engaged session is a session that did at least one of three things:

1. lasted longer than 10 seconds (the threshold is adjustable up to 60), or
2. had 2 or more page or screen views, or
3. included at least one key event (a conversion).

The engagement rate is simply the share of sessions that were engaged: $\text{engagement rate} = \dfrac{\text{engaged sessions}}{\text{total sessions}}$. If 1,000 organic sessions produced 620 engaged sessions, the engagement rate is $620 / 1000 = 62\%$ (illustrative numbers). And here's the tidy part that resolves a lot of confusion: GA4's bounce rate is now just the inverse of engagement rate — the percentage of sessions that were not engaged, or $1 - \text{engagement rate}$. So GA4's bounce rate and UA's bounce rate are different animals wearing the same name; do not compare them, and do not import old bounce-rate intuitions into GA4.

Two more engagement metrics round out the picture. Average engagement time is the average time your content was actually in the foreground of the user's screen (GA4 measures focus, not just an open tab) — a genuinely better "did they read it?" signal than UA's easily-inflated "time on page." And engaged sessions per user tells you whether people come back and dig in. Together these are your best on-site read of "is this content satisfying the people SEO sends it?"

UA → GA4: THE ENGAGEMENT TRANSLATION                     [schematic]
  UNIVERSAL ANALYTICS (retired)        GOOGLE ANALYTICS 4 (now)
  Bounce rate (single-page = bad)  ──▶ Engagement rate (engaged = good); bounce = 1 − engagement
  Time on page (inflated, gappy)   ──▶ Average engagement time (foreground focus only)
  Pageviews (the headline metric)  ──▶ Just one event among many (page_view)
  Goals                            ──▶ Key events (see §28.5)

But before you start chasing these numbers, sort out what they are and aren't — because the temptation to treat engagement as a ranking lever is exactly the myth §28.1 warned about, wearing a subtler disguise.

⚖️ Evidence Check Claim: "A higher engagement rate improves my Google rankings." Sorted onto the honesty scale: — Confirmed by Google: nothing here. Google does not use your GA4 engagement rate as a ranking signal (see §28.1). Full stop. — Plausible but unconfirmed (the interesting part): Google's systems may use some aggregate behavioral signals from its own data — not your Analytics — as part of quality evaluation. The 2023 U.S. antitrust proceedings surfaced internal references to Google search signals with names like "Navboost" that appear to draw on click and interaction data. That is Google's interaction data, at Google's scale, not the engagement rate in your GA4. What weight any of it carries is unknown, and inferring "so I should juice my GA4 engagement rate" is a category error. — The honest read: engagement metrics are a diagnostic for you, not a dial for Google. A page with a genuinely high engagement rate is probably satisfying its visitors — which is correlated with the things that do help rankings (good content, matched intent) because they share a cause. Use engagement to find pages that disappoint readers and fix them. Do not use it as a scoreboard you think Google is reading.

🚫 SEO Myth: "GA4 killed bounce rate because a high bounce rate was hurting everyone's SEO." Two errors in one sentence. GA4 didn't kill bounce rate — it redefined it as the inverse of engagement (it's still there, in the metric picker). And bounce rate never directly hurt your SEO, because, again, Google isn't reading your Analytics. GA4 changed the metric because UA's version was a genuinely bad measure of user satisfaction — it punished pages that answered a question so well the visitor had no reason to click further. The lesson isn't "bounce rate is now safe"; it's that any single engagement number is a starting point for a question, never a verdict on your SEO.

28.5 Conversions: turning visits into forms, calls, and bookings

Everything so far has been prologue to the number that actually matters. Traffic, engagement, landing pages — a business owner tolerates these as context, but what they pay for is conversions: the moments a visitor does the thing your site exists to make them do. For Rivertown that's a booking request, a contact-form submission, or a phone call. For a store it's a purchase. Configure these well and GA4 stops being a curiosity and becomes the tool that justifies the whole SEO program. Configure them badly — or not at all, which is Rivertown's Chapter 1 diagnosis, "no idea what's working" — and every other number is trivia.

First, the vocabulary, because Google changed it and left a trail of confusing older documentation. A conversion is the business concept: a completed action you value. In Universal Analytics these were Goals. In early GA4 they were called conversions. Then, in 2024, Google renamed the GA4 concept to key events — reserving the word "conversions" for the closely-related, ads-oriented metric — so that today, in the GA4 interface, you mark an event as a key event to tell GA4 it represents business value. Same idea, three names across the product's history. Throughout this book we'll use key event for the GA4 mechanism and conversion** for the concept; when you see "conversions" in an old tutorial, read "key events."

FROM RAW EVENT TO MEASURED CONVERSION                    [schematic]

  a form_submit event fires ─▶ you MARK "form_submit" as a KEY EVENT ─▶ it now appears as a
  (raw signal, one of many)     (Admin → Events, or Admin → Key events)   conversion in every report,
                                                                          scoped to organic and beyond

Here is how you'd define Rivertown's three conversions, in plain terms — and, just as important, what each one honestly does and does not prove:

• Form fills. The contact and "request service" forms. Modern GA4 enhanced measurement can capture some form submissions automatically as form_submit events; in the real world, many forms (custom plugins, multi-step flows) still need a small amount of setup — a thank-you-page pageview marked as the key event, or an event configured through a tag manager. Mark it, and give it the recommended name generate_lead where it fits. What it proves: someone asked to be contacted. What it doesn't: that anyone answered, that the lead was real (spam forms exist), or that a job resulted.
• Phone calls. On the site, a phone number should be a tappable tel: link (the click-to-call fix from Chapter 17). A click on that link can be captured as an event and marked a key event. What it proves: someone tapped to call. What it emphatically doesn't: that the call connected, that anyone picked up, or — the big one — anything at all about calls placed from your Google Business Profile. A furnace-panic searcher who taps "Call" in the local pack (Chapter 25) never touches your website; that call is invisible to GA4 forever. It shows in your Business Profile insights (Chapter 25) or in dedicated call-tracking software, not here. This is the most common way local businesses badly undercount their real conversions — treating GA4 tel-clicks as their total calls.
• Bookings. Rivertown's online booking form is the highest-value on-site action — an actual scheduled job. The clean way to measure it is a key event on the booking-confirmation step ("your appointment is requested"). One honest complication: if the booking runs through a third-party widget in an iframe, the confirmation may happen on another domain and evade your GA4 unless cross-domain tracking is set up — a real, common gap worth checking rather than assuming.

And the fourth, for completeness and for the e-commerce readers: purchases. A store implements the GA4 purchase event carrying a monetary value and the items bought, which lets GA4 report actual revenue by channel — the cleanest conversion measurement there is, because the value is a real dollar figure, not an estimate. Rivertown's site is lead-generation, not a store (its light parts/filters commerce is a Chapter 31 footnote), so its "conversions" are leads, not transactions. That distinction drives a useful trick we'll set up for the business case later:

Because a lead isn't a sale, you can assign a key event an estimated value so GA4 can total it. Suppose Rivertown's team estimates an average booked job is worth about \$450, and that roughly 1 in 5 booking requests becomes a booked job. Then a booking request is "worth," as a planning figure, $450 \times 0.20 = \$90$ (entirely illustrative — these are constructed numbers you'd replace with the business's real close rate and job value). Assigning that value lets you approximate the dollar return of organic search — but it is an estimate built on the business's own assumptions, and honest reporting labels it as such. The rigorous version of this math is Chapter 39's job, not this chapter's.

🛠️ Try It on Your Site In GA4, go to Admin → Events. You'll see the events GA4 is already collecting. Find one that represents value (say form_submit or a generate_lead you've set up) and toggle "Mark as key event." That's it — no code — and it will now count as a conversion in every report going forward (it isn't retroactive). Before you trust it, open Admin → DebugView, then go to your own site and perform the action; watch the event appear in DebugView in real time. Testing that the event actually fires when the action happens is the step 90% of people skip and then wonder why their numbers are wrong.

Two GA4 features make most of this achievable without a developer, which matters because the person doing SEO is often not the person who can deploy code. Create event (Admin → Events → Create event) lets you mint a brand-new event from an existing one when a condition is met — for example, "when a page_view fires with page_location containing /booking-confirmed, create an event called booking_complete" — and then you mark that as a key event. Modify event lets you rename or adjust events after the fact. Between enhanced measurement, these rules, and a tag manager for anything stubborn, a surprising share of conversion tracking needs no engineering ticket at all. The genuinely code-dependent cases — a booking widget on another domain, a custom checkout — are the exception, and worth flagging to a developer early rather than discovering broken later.

One more capability closes the loop between an online lead and its real-world outcome: GA4 (and Google Ads) can import offline conversions — feeding back, from your CRM or scheduling system, the fact that lead

4471 eventually became a booked, paid job. This is how a mature setup finally answers "which organic

booking requests turned into revenue," and it's the technically-clean version of the estimated-value trick above. It requires plumbing between systems (and, done right, a consent and privacy review), so we only name it here as the destination; Chapters 29 and 39 are where connecting leads to revenue becomes the point.

Two guardrails on conversion setup that matter more than any tactic. The first is a hard legal-and-policy line; the second is the honesty this whole book runs on.

🚫 SEO Myth: "Capture the customer's phone number and email as parameters on the form-submit event, so I can match conversions to people in GA4." Do not do this. Google Analytics's own policy prohibits sending personally identifiable information (PII) — names, emails, phone numbers — into GA4, and doing so can get your data deleted or your account terminated, on top of the privacy-law exposure in §28.7. It is a genuinely common and genuinely serious mistake, often made innocently by capturing form-field values. Measure that a lead happened and what it was worth as a category; identify who the lead is in your CRM (customer relationship management system) or booking software, which is built to hold personal data lawfully. GA4 is for aggregate behavior, not a back-door customer database.

🔄 Check Your Understanding Rivertown's owner says, "GA4 shows 30 phone-call conversions from organic last month, but the office logged 140 calls. Your tracking is broken." Give two reasons the GA4 number can be genuinely lower than the real call count without anything being broken.

Answer

(1) Calls that never touched the website — anyone who tapped "Call" on Rivertown's Google Business Profile / local pack (Chapter 25), or who found the number elsewhere and dialed it, is invisible to GA4; those calls show in Business Profile insights or call-tracking software, not here. (2) A tap is not a connected call — GA4 counts tel: link clicks, not answered conversations, and it only sees clicks from consented, non-blocked visitors (§28.7). The gap between 30 and 140 is mostly measurement scope, not a bug. The fix isn't to distrust GA4 but to combine it with Profile insights and the office's call log.

28.6 Attribution: who gets credit, and why the last click lies

A single conversion often has more than one cause. Someone discovers Rivertown through an organic search for "why is my furnace short cycling," reads the blog, leaves. Three days later, cold again, they search "Rivertown Home Services" by name, click through, and book. Which channel gets the "credit" for that booking — the organic informational search that introduced them, or the branded search that closed them? That question is attribution: the set of rules for assigning credit for a conversion to the touchpoints along the path that produced it. And how you answer it changes, dramatically, how good SEO looks.

Three attribution models frame the whole debate:

ONE BOOKING, THREE STORIES ABOUT WHO EARNED IT          [constructed teaching example]

  Day 1: ORGANIC search (furnace blog) ─▶ Day 3: DIRECT/BRAND search ─▶ BOOKING (a scheduled job)
         the discovery touch                  the closing touch

   LAST-CLICK  says:  Direct/Brand earns 100%.  Organic earns 0%.   ← SEO looks worthless
   FIRST-CLICK says:  Organic earns 100%.        Direct earns 0%.    ← SEO looks all-powerful
   DATA-DRIVEN says:  ~split, e.g. Organic 0.6 / Direct 0.4         ← closer to the messy truth

Stare at that diagram, because it contains the most important political fact about SEO measurement: under last-click attribution, SEO's early-funnel work is systematically invisible. SEO is very often the first touch — the informational search that introduces a stranger to a business — and last-click hands all the credit to whatever channel happened to be there at the finish (frequently "direct" or a branded search, which is itself often the fruit of an earlier organic discovery). A strategist who reports SEO's value on a last-click basis is quietly arguing for their own budget cut. This is not a reason to cherry-pick first-click instead (which lies in the opposite direction); it's a reason to understand the model your numbers are using and to say so out loud.

GA4's own answer to this has shifted, and honesty requires noting it. GA4 now defaults to data-driven attribution and, in 2023, retired most of the old rules-based models (linear, time-decay, position-based, and standalone first-click) from its attribution settings — leaving data-driven and last-click as the main options. So the "first-click vs. last-click vs. data-driven" framing you should carry is partly conceptual: you'll reason with all three to understand credit, even though the GA4 interface now steers you toward data-driven. Which raises the obvious question about that default.

⚖️ Evidence Check Claim: "Data-driven attribution tells me the true value each channel contributed." Sorted: — What's genuinely better about it: DDA distributes credit across all the touchpoints instead of crowning one, using patterns in your own conversion data. For a multi-touch journey it is almost always less wrong than last-click, and it's a reasonable default. — What it is not: truth. It is a model — a machine-learned estimate whose inner workings Google does not fully expose, that needs sufficient conversion volume to train (a low-traffic site may get thin or unstable results), and that you cannot independently audit. It can't see offline influence (the neighbor who recommended Rivertown), and it only knows the touchpoints GA4 actually captured — which, per §28.7, is an incomplete set. — The honest posture — the statistician's proverb: all models are wrong; some are useful. Prefer data-driven to last-click, report which model you used, and never present any attribution number as the objective contribution of a channel. It is the best available estimate, presented honestly as an estimate.

Two settings quietly shape every attribution number, and you should know they exist even if you rarely touch them. The lookback window is how far back GA4 looks for touchpoints to credit before a conversion (a booking today can credit an organic visit from weeks ago — but only within the window). And GA4's Advertising → Attribution section holds the conversion paths report, which lays out the actual early/mid/late touch sequences leading to your key events — the closest GA4 comes to showing you the multi-step journeys that the single-number models flatten. For an SEO arguing that organic does unseen early-funnel work, that conversion-paths view (showing how often Organic Search appears as an early touch) is often the most persuasive screen in the whole tool. It won't end the attribution debate — nothing does — but it replaces assertion with the platform's own picture of the path.

🔗 Connection Attribution gets harder, not easier, for long, considered purchases. Chapter 32 (SaaS & B2B SEO) takes on multi-touch attribution across sales cycles measured in months, where the gap between "last click" and "what actually earned the deal" becomes a business problem in its own right. And the reason attribution matters at all — turning credited conversions into a defensible dollar figure and an ROI case — is Chapter 39 (The Business Case for SEO). This section gives you the model literacy those chapters build on.

28.7 Privacy, consent, and the honest limits of measurement

We end where the chapter's second theme has been heading all along: the numbers are an approximation, and the gap between the number and reality is growing. This isn't cynicism; it's the professional posture the whole book argues for — theme 3, evidence over folklore, turned on your own tools. An SEO who presents GA4 figures as precise truth is committing the exact sin the book spends forty chapters warning about in others.

GA4 was, in part, Google's response to a world where measurement is legally and technically constrained. Several design choices reflect that. GA4 does not log or store IP addresses (it uses them momentarily for coarse geolocation, then discards them). It offers data-retention controls (user-level data kept for 2 or 14 months). And it's built around consent rather than assuming it. Understanding why requires a brief, careful tour of the rules — conceptually, and explicitly not as legal advice.

• GDPR (the European Union's General Data Protection Regulation, 2018) requires, broadly, a lawful basis to process personal data, meaningful disclosure, and — for analytics/marketing cookies — usually consent obtained before tracking. It's why European sites greet you with a cookie banner.
• CCPA/CPRA (California's Consumer Privacy Act and its amendment, the Privacy Rights Act) take a somewhat different, opt-out-oriented approach — a right to know, to delete, and to opt out of the "sale" or "sharing" of personal information.
• These are not abstract. European data-protection authorities in Austria, France, and Italy ruled in 2022 that certain uses of Google Analytics could violate GDPR (over data transferred to the U.S.), before the 2023 EU-U.S. Data Privacy Framework shifted the ground again. The point for you isn't the legal detail — ask a qualified lawyer for that — it's that how you may lawfully measure depends on where your users are, and it changes.

Google's technical bridge to this world is Consent Mode: a mechanism by which your site tells Google whether a given visitor has consented to analytics and advertising cookies. When consent is granted, GA4 measures normally. When it's denied, GA4 sends only cookieless, non-identifying pings (or nothing), and Google models the resulting gap — statistically estimating the conversions and behavior it wasn't allowed to directly observe. (A 2024 update, "Consent Mode v2," tightened what's required for sites using Google's advertising and audience features in the European Economic Area.) Sit with the consequence: a portion of what GA4 reports for a consent-banner site is modeled, not counted.

THE FUNNEL OF LOSS: real behavior → what GA4 reports     [schematic — not to scale]

  EVERYONE who actually visited & acted
     │  − visitors who blocked the tag (ad blockers, privacy browsers, extensions)
     │  − visitors who declined the cookie/consent banner (measured only via modeling, if at all)
     │  − actions where the tag failed to fire (slow loads, quick exits, script errors)
     │  − cross-device journeys split into "different users" (phone then laptop = two people to GA4)
     │  − small numbers hidden by DATA THRESHOLDING (to protect anonymity)
     ▼
  WHAT GA4 SHOWS YOU  ◀── directionally useful; NOT a complete census

That last item deserves a name because it surprises people. Data thresholding is GA4's practice of withholding data — hiding rows, or a whole breakdown — when the numbers are small enough that an individual could theoretically be re-identified (it kicks in most often when Google Signals or demographic data is enabled). You'll open a report, filter to a narrow segment, and find rows simply missing, with a small icon explaining that data was withheld. Nothing is broken; GA4 is refusing to show you data it judges could compromise anonymity. It's a privacy feature that feels like a bug.

⚖️ Evidence Check Claim: "GA4 reported 1,284 organic sessions, so 1,284 people visited from organic search." Sorted, as the capstone of the chapter's honesty: — Roughly true and useful: as an approximate magnitude and, above all, as a trend — 1,284 this month vs. 900 last month is a real, meaningful rise — GA4 is trustworthy and decision-grade. — Not literally true: the real number is almost certainly higher (blocked, declined, and unfired visits are missing) and fuzzier (some sessions are modeled, some "users" are one person on two devices, some data is thresholded away). The decimal precision is false confidence. — The professional habit: trust GA4 for direction, comparison, and relative magnitude; distrust it as a precise census. Compare like-for-like periods, watch trends over single readings, round when you report, and never build a claim on the third significant figure. "Organic grew meaningfully and drove dozens of bookings" is honest; "organic drove exactly 1,284 sessions and 52.0 conversions" is precision theater.

There's an ethical line here too, and it's short: measure to serve users, honor their consent, and collect only what you need. Don't deploy tricks to defeat the privacy protections users chose (evading consent, smuggling PII, fingerprinting around cookie refusals) — that's the analytics version of the black-hat tactics this book rejects everywhere else, and it invites the same legal and reputational blowback. The durable posture is the same one that wins at SEO: work with the system's genuine intent, not against it. A slightly incomplete but honest and lawful measurement setup beats a "complete" one built on data you weren't permitted to take.

📈 The Strategy File

Rivertown's Chapter 1 diagnosis included a line we've carried for twenty-seven chapters: "no idea what's working." Chapter 27 fixed half of it by standing up Search Console — Rivertown can now see its impressions, clicks, and queries. This chapter fixes the other half: defining and configuring Rivertown's conversions in GA4, so that organic traffic can finally be tied to the things that make the Delgados money. We are not building the owner's report yet (that's Chapter 29), and we are not doing the ROI math yet (Chapter 39). We are defining what counts as success and wiring GA4 to capture it — honestly, and within the rules.

Here is Rivertown's conversion configuration as it enters the Strategy File:

FIGURE 28.2 — "Rivertown's GA4 conversion setup"                      [the Strategy File]
  CONVERSION (KEY EVENT)   HOW IT'S CAPTURED                    WHAT IT PROVES / DOESN'T PROVE
  Booking request          Key event on the booking-            The highest-value on-site action.
   (the priority)           confirmation step; check for         Does NOT prove the job was booked,
                            iframe/cross-domain gaps.             kept, or paid — that lives in the
                                                                  scheduling system, not GA4.
  Contact / service form   form_submit (enhanced measurement)   A raised hand. Does NOT prove the
   fill                     or thank-you-page key event; named   lead was real (spam happens) or
                            generate_lead. NO PII in parameters.  answered. Confirm quality in the CRM.
  Phone-link (tel:) click  click event on the tappable number   Someone tapped to call. Does NOT count
                            (the Ch 17 click-to-call), marked     GBP/local-pack calls (invisible to
                            a key event.                          GA4 — see GBP insights, Ch 25) or
                                                                  whether the call connected.
  Estimated lead value     Assign an illustrative value to the  Lets GA4 approximate organic's dollar
   (planning only)          booking key event (e.g., ~\$90 per   return. It's an ESTIMATE on the
                            request, from a constructed \$450     business's own close-rate assumptions,
                            job × 20% close). Label it as such.   NOT measured revenue (rigor = Ch 39).

Now the honest scorecard, because a Strategy File that overclaims is exactly what this book exists to replace.

What this component settles. Rivertown can now, for the first time, answer "did organic search produce business value?" in concrete terms — booking requests, form fills, and tel-clicks, isolated to the Organic Search channel, page by page. Paired with Chapter 27's query data, the Delgados have a real, two-sided measurement system where a month ago they had a shrug. That is the foundation every later argument (the report, the budget, the ROI case) stands on.

What it explicitly does not settle. Three honest gaps, stated plainly so no one is misled:

• GA4 undercounts Rivertown's real calls, badly. The furnace-panic customer who taps "Call" in the local pack — the exact high-intent moment Chapter 25 is built around — never appears here. Rivertown's true conversion picture is GA4 plus Business Profile insights plus the office's call log, never GA4 alone.
• A conversion is not a booked, paid job. GA4 sees the request; the scheduling system and the CRM see the outcome. Tying organic leads to organic revenue requires joining GA4 to those systems, and the clean ROI math is deferred to Chapter 39.
• The numbers are approximate. Consent banners, ad blockers, and thresholding mean Rivertown's measured conversions are a floor, not a census. We'll report trends and magnitudes, never false precision.

The Strategy-File task for your own site (Appendix C's worksheet): list the two or three actions on your site that represent real business value. For each, write (a) how you'd capture it as a GA4 key event, (b) one sentence on what it proves, and (c) one sentence on what it doesn't — the invisible conversions, the "request ≠ sale" gap. Then mark at least one as a key event in GA4 and verify it fires in DebugView. You will have done, honestly, what most sites never do at all: decided what success means before measuring it.

Conclusion

We set out to move from "how many visits" to "did it make money," and we got there — with both hands open about the limits. GA4's world is events and parameters, not pageviews and sessions; you isolate SEO by scoping to the Organic Search channel in the Traffic acquisition report, knowing GA4 will show you everything about those visits except the query that drove them — the seam where Search Console (Chapter 27) completes the picture. You read landing pages as a portfolio of discovery and money pages, and engagement (engaged sessions, engagement rate, average engagement time) as a diagnostic for you, never a dial Google reads. You define conversions as key events — forms, calls, bookings, purchases — and you now know the two rules that matter most: never send Google personal data, and never mistake a tel-click for a booked job. You can reason across first-click, last-click, and data-driven attribution, and you understand the political fact buried in there: last-click quietly makes SEO look worthless because SEO is so often the first touch. And you finished by refusing the field's favorite lie — that analytics numbers are truth — in favor of the honest read: directionally trustworthy, precisely fictional, increasingly constrained by consent and privacy by design.

What remains genuinely unresolved is worth naming. We measured leads, not revenue; we captured on-site conversions while a chunk of Rivertown's real calls happen off-site; and every count is a floor beneath an unknowable true number. None of that makes the measurement worthless — it makes it honest, which is the only kind worth reporting.

Which is exactly the next problem. Having collected the data, you have to communicate it — to an owner who doesn't care about engaged sessions and does care about the phone ringing, without drowning them in metrics or overselling what the numbers prove. In Chapter 29, we build the report: the KPIs that matter, the vanity metrics that don't, the cadence, and the honest narrative that ties organic traffic to leads to revenue for someone who will never open GA4 themselves.

→ Continue to Chapter 29: SEO Reporting — Dashboards, KPIs, and Communicating Value.

Key Terms

• GA4 (Google Analytics 4) — Google's current, event-based web and app analytics platform; successor to Universal Analytics, which stopped processing data in standard properties on July 1, 2023.
• Event — in GA4, the atomic unit of measurement: any recorded interaction (a pageview, scroll, click, form submission, purchase), described by parameters. In GA4, everything is an event.
• Engaged session — a session that lasted longer than 10 seconds, had 2 or more page/screen views, or included at least one key event; GA4's positively-defined replacement for the old "non-bounce."
• Engagement rate — the share of sessions that were engaged (engaged sessions ÷ total sessions); GA4's bounce rate is its inverse ($1 - \text{engagement rate}$).
• Conversion — the business concept of a completed, valued action (a lead, a booking, a sale); called a "Goal" in Universal Analytics and, in GA4 since 2024, a "key event."
• Key event — GA4's term (from 2024) for an event you have marked as representing business value, so it is counted as a conversion in reports; the mechanism you configure for forms, calls, and bookings.
• Attribution — the set of rules for assigning credit for a conversion to the touchpoints along the path that produced it (e.g., first-click, last-click, data-driven).
• Consent Mode — Google's mechanism for a site to signal whether a visitor consented to analytics/advertising cookies; when consent is denied, GA4 sends limited signals and models the gap rather than directly measuring it.
• Organic segment — a saved, reusable GA4 definition scoping data to sessions from the Organic Search channel, dropped into explorations so every metric is measured for SEO traffic alone.

Spaced Review

Retrieval practice. Try each before revealing the answer. (Mixing this chapter with Chapter 27.)

1. In GA4's data model, what is a "pageview," and why does calling everything an "event" change how you think about the tool?
2. You want to know whether a specific conversion came from an organic search for the words "emergency plumber." Which tool can get you closest, and what is the hard limit that stops any tool from joining that exact query to that exact conversion?
3. (Chapter 27) A page shows high impressions but a low click-through rate in Search Console. What does that pattern suggest, and which on-page elements would you look at first — and why is this a GSC question rather than a GA4 one?
4. (Chapter 27) Name two things Search Console shows you that GA4 fundamentally cannot, and one thing GA4 shows that GSC cannot.
5. Explain why reporting SEO's value on a last-click basis tends to make a good SEO program look like a poor investment.