What GA4 Actually Is (and Whether It Still Fits an SMB in 2026)

GA4 is Google Analytics 4, Google's current analytics platform, an event-based product with a free standard tier that is the default analytics install on most small business sites in 2026. A regional distributor installed it under deadline pressure when Universal Analytics stopped processing data on July 1, 2023, kept the default events the setup wizard wired up, and three years later still cannot answer the one question that comes up in every Monday meeting: which marketing channel actually brings in customers who order more than once. The platform is fine. The capture is the problem. The interesting question is the gap between owning a working GA4 property and getting an honest answer out of it.

What GA4 actually is

GA4 is the analytics product Google ships today, free for the standard tier, with a paid 360 tier most SMBs never need and never will. It records events on your site (a page view, a click, a form submission, a purchase) and ties those events to anonymous users so you can see how people move through your site over time. It runs in a browser via a small JavaScript snippet, or through Google Tag Manager, or through one of the integrations your CMS already ships. The free tier comes with a generous monthly event quota, a BigQuery export almost no other free analytics product offers, and the same model-driven attribution Google uses inside its own ad products.

The thing being given away is access to Google's analytics engine, the one most agencies and most ad platforms already know how to read. That ubiquity is the underrated feature. Whatever else you can say about GA4, every freelancer, every contractor, and every ad platform you work with knows what a GA4 property is and how to look at one. A boutique analytics tool may be technically better in some narrow dimension; almost none of them have that gravity.

The one structural shift from Universal Analytics

If you used Universal Analytics, the data model in GA4 is genuinely different and the difference is the source of most "GA4 is confusing" complaints on the surface internet in 2026. Universal Analytics counted sessions and pageviews as the primary units. GA4 counts events and users. A pageview in GA4 is just one kind of event among many; a button click, a video play, a scroll, a form submission, and a purchase are all events of equal structural standing. The session is still computed, but it is derived from the event stream rather than being the unit the platform is built around.

The practical effect is that the old "average session duration on the homepage" question maps onto GA4 awkwardly, and the new "how many people who triggered the contact-form-submit event in the last 90 days came back and triggered the purchase event" question is the kind of thing GA4 answers more naturally than Universal Analytics ever did. For an SMB starting fresh in 2026, this is fine. For a team that built years of UA reports around the old model, the migration was real work, and that real work is the reason the contrarian "GA4 is bad" framing got loud for a while. Three years in, the data model is the right one for the questions a business actually has; the migration pain has receded; the platform is fine.

Is GA4 still the right analytics for a small business in 2026?

Yes for most SMBs. The standard tier is free, the platform is the default in the ad-and-agency ecosystem, and the data model fits modern business questions. The real work is not picking the tool; it is deciding which events the business needs to capture and writing that decision down.

What an SMB actually has to decide

The defaults GA4 wires up at install time give you page views, basic engagement, scrolls, outbound clicks, file downloads, and a few other generic events. None of those events answer the question the regional distributor brings up in every meeting. The events that answer her question are specific to her business: a quote-request submission, a customer-service form, a repeat-order purchase, a particular product line being viewed. GA4 records what it is told to record. It does not know which of those events matter to the business until somebody decides and wires the capture.

This is the entire game. An SMB that installs GA4, leaves the defaults running, and never has anyone sit down to decide which events the business actually needs is going to look at the panel in three years and feel exactly what the distributor feels. The panel will be full of numbers and none of them will be the one she is trying to read. The deep instrumentation discipline (what an honest tracking plan actually looks like in practice) is one pillar over; the bridge guide on the working discipline of writing a tracking plan covers the events-and-properties model, the naming convention, the verification pass, and the small failure modes that bite when an SMB tries to do this without a tracking plan in hand. The narrower point for the tool-choice question is this: the question is not which analytics product to install. The question is what to measure.

The consent reality in 2026

Consent Mode v2 is in market and most ad platforms now expect you to send a consent signal alongside any analytics or ad event. The practical reality for an SMB site in 2026 is that some percentage of your visitors will decline analytics cookies, and GA4 records less detail (or, depending on your configuration, only modeled aggregate data) for that group. This is not a GA4 quirk. It is the regulatory shape of the web in 2026 and every analytics tool you would consider as an alternative has the same problem. An SMB needs a consent strategy regardless of which platform it picks, and a small business with EU traffic needs a Consent Mode v2 setup whether it runs GA4 or anything else.

Treat the consent layer as a separate object from the analytics tool. The consent banner decides which visitors GA4 is allowed to record in detail; GA4 decides what to do with the signal it gets. Most "GA4 is broken because of consent" complaints in 2026 are really "we shipped a site with a consent banner that defaults to opt-out and we are surprised our analytics numbers dropped". Both halves are real and both halves need to be looked at by name.

The cookie story the readers usually have wrong

The most common reader-side prior in 2026 is that GA4 is somehow killing cookies, or will, or has. The actual shape is different. Browsers and regulators are slowly reducing what third-party cookies can do on the open web. That story affects every analytics tool, every ad platform, and every retargeting product on the market. It is not a GA4 feature and it is not a GA4 bug. GA4 still uses cookies where consent allows; it uses modeled aggregates and first-party signals where consent does not; the broader cookie-deprecation story is something an SMB owner should be aware of as a property of the web in 2026, not as a property of which analytics product to pick.

Where GA4 is genuinely good, briefly

The model-driven attribution Google ships inside GA4 is better than what Universal Analytics had and better than most ad-platform reporting an SMB would compare it to. The BigQuery export on the free tier is a real feature; almost no other free analytics product gives you raw event data to query as SQL. The predictive metrics (purchase probability, churn probability) work well enough on accounts with enough volume to be useful for ad-platform audiences, and on accounts without enough volume they quietly stay gray, which is honest. The Google Signals layer, where users are signed into their Google account, gives you a more complete cross-device picture than tools without that hook can produce.

None of these features matter much if the SMB has not decided which events the property should record. They are good features built on top of an event stream the owner has to define.

The "who am I measuring" question, briefly

The question GA4 cannot answer for you is which buyer the business is even trying to win. A site can capture every event a buyer triggers and the report will still feel hollow if the underlying customer picture is muddy. The companion question (who the SMB is selling to, how to describe that buyer in language that lets the business decide what to measure for them) is a marketing one, and the customer-profile post covers it. The two posts read in sequence; this one names the tool and the considerations, that one names the buyer the events are supposed to describe.

What to do next

If GA4 is already installed and capturing the defaults, the next move is not to switch tools. It is to sit down once and write the tracking plan, which the bridge guide on this page handles in depth. If GA4 is not installed yet and the question is whether to pick something else, the honest answer for most SMBs is no; pick GA4, get it installed, and spend the time you would have spent comparing products on deciding what to capture instead. If the in-house bandwidth to wire and maintain the capture against a plan does not exist, Iron Goo's foundation service is where the wiring and the upkeep of a GA4 property against a real tracking plan can live; it is not a setup-by-the-hour pitch, it is the ongoing instrumentation discipline that keeps the capture honest as the site changes.

The measurement question lives in the marketing tab where the owner spends most of their time, but the disciplined answer to it is an instrumentation problem that lives one pillar over. Read the tracking-plan guide next and turn the GA4 property you already have into one that answers the question that comes up in every Monday meeting.