Open GA4 once a month, look at five reports, and close it again. That is the whole routine, and it will tell you more than the dashboard someone built you in a hurry. The reason GA4 feels like hard work is not that the reports are bad. It is that there are far too many of them, the metrics you want are switched off by default, and two settings buried in the admin screens decide how much history you will still have in a year. Sort those out first and the rest of the tool becomes quite reasonable.
Five reports, thirty minutes a month
Here is the short list, in the order worth reading. Reports → Acquisition → Traffic acquisition tells you which channels brought sessions and how many key events each one produced. Reports → Engagement → Pages and screens tells you what people actually read. Reports → Engagement → Events is where you check that your tracking still works. Reports → Retention matters if you sell anything repeatable. And one saved exploration — a funnel from landing page to enquiry — answers the question everyone actually asks.
Five is not a compromise. It is the number of reports a small team can read carefully enough to notice a change. A twenty-tile dashboard gets skimmed, and skimming a dashboard is the same as not looking at it. If someone on your team wants a sixth report, make them delete one.
- Traffic acquisition — is the mix of channels changing, and is any channel producing enquiries rather than just sessions?
- Pages and screens — which pages earn attention, and which ones people leave immediately?
- Events — did a form, a phone-tap or a download stop firing after the last site change?
- Retention — do people come back, or is every month a fresh start?
- One funnel exploration — where in the path from arrival to enquiry do people give up?
Fix the settings before you read anything
Two settings do more damage than any reporting mistake. The first is data retention. Google’s help page is specific: for a Google Analytics property you can set the retention period of user-level data to 2 months or 14 months, and that period also applies to key events data. For all other event data the standard options are the same 2 or 14 months, with 26, 38 and 50 months available only on Analytics 360. Age, gender and interest data is always held for 2 months no matter what you choose, and Google signals data is capped at 26 months regardless of the setting.
The default on a new property is the short one. So if nobody touched it, the funnel you want to build in March can only see back to January. Google also notes that increasing the retention period is applied to data you have already collected, which is a small mercy, and that Large and XL properties are limited to 2 months, which is not.
| Data type | Standard property | Analytics 360 | Notes |
|---|---|---|---|
| User-level data (includes key events) | 2 or 14 months | 2 or 14 months | New properties start at 2 months |
| All other event data | 2 or 14 months | 2, 14, 26, 38 or 50 months | Large and XL properties are limited to 2 months |
| Google signals data | Up to 26 months | Up to 26 months | Capped at 26 months whatever you set; a shorter setting wins |
| Age, gender and interest data | 2 months | 2 months | Always 2 months, regardless of the setting |
The second setting is the attribution model, and it is covered further down because it changes numbers you have already reported. Do retention first. It takes one minute and it is the only one of the two that is pure gain.
Key events replaced conversions, and the wording matters
GA4 does not have conversions in the old sense. It has key events. Google’s definition is plain: a key event is an event that measures an action particularly important to the success of your business, and you create or identify the event first, then mark it as a key event. Until you mark something, GA4 is collecting page views and scrolls and nothing that looks like a result.
This catches people out constantly. A client rings to say their enquiries have fallen off a cliff, and what has actually happened is that the form plugin changed its success URL, so the event that was marked as a key event no longer fires. The Events report is where you catch that, which is why it earns a place in the five.
Mark three things at most for a small business site: a submitted enquiry form, a tap on a phone number or email link, and one meaningful content action such as a pricing page view or a document download. Google’s help page on reporting key events notes that you can add the key events metric to any report and narrow a report to a single key event from a drop-down, so three well-chosen events give you a lot of reporting reach.
Google Analytics Help, “Key events”A key event is an event that measures an action that’s particularly important to the success of your business.
Reports and explorations are not the same data
This is the single most confusing thing about GA4, so it is worth being blunt. Standard reports and explorations do not draw on the same pool of data. Google’s retention page states that the retention setting does not affect standard aggregated reports, including primary and secondary dimensions, and that it only affects explorations and funnel reports.
In practice: your Traffic acquisition report will happily show you eighteen months of sessions by channel, while an exploration built on the same property refuses to go back more than two months. Nothing is broken. One is aggregated and kept, the other is built from user-level rows that have been deleted on schedule.
There is a second trap in explorations. Google’s configuration limits page says ad-hoc queries processing beyond 10 million events will be sampled, while standard queries are not affected by that limit. So an exploration over a busy twelve-month window can return a sampled estimate, and a colleague running a similar exploration with a slightly different date range can get a different answer. If a number is going in front of a board, take it from a standard report.
Engagement rate and bounce rate are hidden on purpose
Engagement rate and bounce rate exist in GA4, but Google’s own help page says that by default most reports do not include them, and that to see them you need to customise the report — something only an Editor or Administrator can do, and only on detail reports. That is not a bug, it is a nudge. GA4’s bounce rate is simply the inverse of engagement rate, and engagement is defined by session duration, page views or whether a key event fired.
Add engagement rate to the Pages and screens report once and leave it there. It is the fastest way to spot a page that attracts traffic and gives nothing back. Do not chase the number itself. A pricing page with a low engagement rate and a healthy key event count is doing its job; a blog post with high engagement and no onward clicks is a dead end with good manners.
Traffic acquisition, without the channel-name confusion
Traffic acquisition is where most of the useful month-to-month signal lives. Read it by session default channel group, add the key events metric, and resist the urge to interpret week-level wobbles. Three things reliably confuse people here.
- Direct is a bucket, not a channel. It collects everything with no usable referrer: bookmarks, pasted links, some app clicks, some email clients.
- Organic Social and Referral overlap. A link shared in a private message often lands in Direct rather than either.
- Unassigned means your tagging slipped. It usually points at a campaign URL missing its source or medium parameter, not at a GA4 fault.
If you run paid campaigns alongside organic work, the comparison is less obvious than it looks, and we have written separately about how to split a budget between the two in SEO vs paid ads. The short version: GA4 will tell you what happened after the click, and almost nothing about what happened before it.
Attribution settings quietly rewrite your history
Now the setting that needs a warning label. Google’s help page on attribution settings says the reporting attribution model determines how credit is assigned to touchpoints, and — this is the part people miss — that changing the reporting attribution model applies to historical and future data. Change it on a Tuesday and last quarter’s channel numbers move.
That is not a reason to avoid the setting. It is a reason to decide once, write down the date you changed it, and tell whoever reads your reports. If a channel’s contribution jumps 30% with no change in spend, the first question should be whether someone touched the model.
- Export the current month’s Traffic acquisition report to a spreadsheet first, so you have the before picture.
- Note the change date somewhere your future self will find it — a comment in the property name field works.
- Expect paid and organic search credit to shift in opposite directions.
- Do not change it in the middle of a campaign you are being judged on.
When GA4 stops being the right tool
GA4 is free, generous and genuinely good at describing web behaviour. It is poor at two things small businesses care about: joining a web session to a deal that closed four months later, and keeping history beyond its retention window. Both have the same fix, and it is not a different analytics tool.
Put the enquiry into your CRM with its source attached, and treat GA4 as the top of the funnel rather than the whole of it. If you need raw history, the BigQuery export keeps event-level rows outside the retention window, though you are then paying for storage and query time and you have taken on a small data engineering job. For a business doing a handful of enquiries a week, a tidy spreadsheet and an honest CRM field beat both.
The content side of this matters more than the measurement side. If your blog is not earning attention there is nothing for GA4 to report, which is the argument we make in content marketing that actually works, and the technical groundwork is covered in SEO fundamentals every business should know.
Frequently asked questions
Do I need Analytics 360?
Almost certainly not. The jump from a standard property is mostly longer retention windows, higher limits and a service agreement, and it is priced for enterprises. If your real problem is losing history, the BigQuery export solves that for a fraction of the cost. Buy 360 when sampling genuinely blocks a decision you are making weekly, not because a report once showed an estimate.
Why do GA4 and my CRM disagree about enquiries?
They are counting different things. GA4 counts key events fired in a browser, with no idea whether a human on the other end was real or whether the enquiry was duplicated. Your CRM counts records a person created or accepted. Expect GA4 to run higher. Treat the CRM as the number of record and GA4 as the explanation of where those records came from.
Should I change data retention today?
Yes, and it takes a minute. Admin, then Data collection and modification, then Data retention. Google confirms that increasing the period applies to data already collected, so you are not only helping future reports. There is no analytics downside; the only consideration is whether your privacy notice and retention policy describe the longer period honestly.
Is the BigQuery export worth setting up for a small site?
If you have fewer than a few thousand sessions a month, probably not yet. It is worth it when you need history beyond 14 months, want to join web events to order or CRM data, or keep hitting sampling in explorations. Budget a day of setup and a small monthly cloud bill, and make sure someone on the team can write SQL before you commit.
What single report should I check if I only have five minutes?
Traffic acquisition, with the key events metric added and the comparison set to the previous period. It shows whether your channel mix is holding, whether anything produced an enquiry, and whether Unassigned traffic has crept up, which is usually the first sign that campaign tagging has broken somewhere.
If your GA4 property has never been configured properly, or nobody can say which events are marked as key events, we can audit it remotely and leave you with the five reports that matter and a one-page explanation of each. Our digital marketing services cover measurement setup alongside the content and search work it is meant to prove. Get in touch with Eudora Technology to talk about your project.
.jpg){kind=link}
