How to Use Google Analytics
Most businesses have this installed and have never opened it. Others open it and look at one number. This guide is about building a rhythm you will actually keep, then about which figures mislead you before you have finished reading them.
Installed And Never Opened
Analytics installed and unread is the normal state of a small business website. Somebody added it during a build, it has been collecting quietly ever since and nobody has looked. That is not negligence. It is what happens when a tool arrives without anybody explaining what to ask it.
Why it goes unopened. It is intimidating.
The interface is built for large organisations with analysts. A plumber with a five page website opens it, sees an unfamiliar screen full of terms and closes it again, reasonably.
What that costs. Everything it could tell you.
The data has been accumulating. Which pages people actually arrive on, which of your services get attention and whether anybody ever gets as far as the contact page are all sitting there unread.
What this page is not. A tour.
We will not walk you through screens, because the interface changes and any such guide dates within months. Everything here is about what to ask and how often.
What it takes to be useful. Twenty minutes a month.
That is genuinely enough for most businesses. The value is in looking regularly with a question in mind rather than in looking hard once.
Where to start if you are new. One question.
Whether anybody who arrived last month did anything you wanted them to. If your setup cannot answer that, that is the finding. Block eight is where it leads.
The Three Questions Worth Asking
Almost every useful report answers one of three questions. Hold those in mind and the tool stops being a wall of figures and becomes something you interrogate. Everything else is detail you can reach for when a question needs it.
How many of the right people arrived. Not how many people.
The qualifier does the work. Visitors from the wrong country, the wrong search or the wrong intention are not an achievement. Counting them together with genuine prospects hides both.
Where they came from. The route in.
Search, another website, social, an advert or apparently nowhere. This is the question most directly connected to whether marketing is working. It is also the one with the most traps, which our traffic sources guide covers.
Whether anything happened. The only one that pays.
Did they enquire, ring, buy or book. A property that cannot answer this is reporting on activity rather than outcomes. setting up conversions is what fixes it.
Why three is the right number. You will remember them.
A framework of fifteen metrics gets abandoned. Three questions survive a busy month, which matters more than completeness.
How they fit together. As a sequence.
People arrive, from somewhere, then either do something or do not. Reading in that order tells you where the problem sits rather than merely that one exists.
A Working Rhythm
Three cadences. The discipline is as much about what you do not look at as what you do. Businesses that check daily generate anxiety without insight, then stop entirely after a fortnight.
Monthly. The main picture.
The three questions, against the same month last year where you have it. Twenty minutes, written down somewhere, with a note of anything that changed on the site that month.
Quarterly. Trends and decisions.
Whether the direction over three months differs from the direction over twelve, which pages consistently do the work and whether the mix of sources is shifting. This is where decisions belong.
On change. Verification only.
After a redesign, a migration or a new section, check within a few days that data is still arriving and arriving once. That is a fault check rather than a performance review.
What not to do. Watch it daily.
Daily figures on a small site are mostly noise. A quiet Tuesday is a Tuesday. Reacting to it is how businesses end up changing things that were working.
Why writing it down matters. Memory fails.
In six months nobody remembers what the figures were or what was changed when. A short monthly note is worth more than any dashboard. It costs nothing.
Who should do it. Somebody in the business.
Even where an agency reports to you, one person internally should look. A report you cannot check is a report you have to take on trust.
The Number Everybody Watches Is The Wrong One
Total visitors is the least useful figure in the tool and the one every business fixes on. It flatters and it misleads in equal measure. A great deal of reporting is built on it precisely because it is large and moves about.
Why it flatters. It counts everybody.
People from other countries, people who arrived by mistake, automated visits and people looking for something you do not sell. All of them make the number bigger.
Why it misleads. It hides direction.
A total can hold steady while the useful half collapses and the useless half grows. Nothing in the headline figure would tell you that had happened.
What businesses do with it. Judge the whole engagement.
Traffic went up so the work is succeeding. Or traffic went down so it is not. Both conclusions are drawn from a number that describes several unrelated things at once.
What to look at instead. The same figure, cut up.
Visitors from search, to the pages you sell from, who did something. That is three qualifiers on one number and it turns a vanity figure into a business one.
The version we watch for clients. Results, then sources.
How many enquiries, then where those specific people came from. Our material on measuring performance argues the wider case, including why traffic is not the objective.
Segment Before You Conclude
An unsegmented figure cannot support a decision. That is the single most useful habit in this guide and it takes seconds, because the tool will cut any number you are looking at by the things that matter.
By source. The first cut, always.
Search behaves differently from social, which behaves differently from an advert. Averaging them produces a figure describing nobody. The same is true of a conversion rate calculated across all of them.
By page. The second.
Which pages people actually arrive on is regularly not what the business expects. An old article can be doing more work than the service page everybody argues about.
By whether they did anything. The one that matters.
Comparing the people who enquired against those who did not tells you what the useful visitors have in common. That is where the actionable finding usually is.
By country, once. A sanity check.
A UK business with a substantial share of visits from elsewhere is looking at a distorted picture of everything, so it is worth knowing before reading anything else.
What segmentation reveals most often. A quiet success.
One source or one page performing far better than the average, invisible while everything is added together. That is a thing to do more of. It only appears when you cut the data.
Compare Against Something
Four hundred visitors is neither good nor bad. It is a number waiting for a comparison. Choosing the wrong one manufactures results that never happened.
The obvious comparison is the flawed one. Last month.
Calendar months differ in length, so February against January starts three days short before anything real occurs. That alone produces a fall somebody will explain with a theory.
Seasonality does the rest. Your trade has a shape.
A roofer in November and a roofer in June are different businesses. Comparing across that boundary measures the weather rather than the website.
The better comparison. The same period last year.
It handles length and season at once. The catch is that you need the data, which is why exporting monthly matters and why businesses that migrated without exporting lost the ability entirely.
The workable middle. Equal length periods.
Three months against the previous three. Long enough to be real, equal by construction and available to everybody regardless of history.
What to record alongside it. What you changed.
A comparison showing a difference is the start of a question. What was published, redesigned or removed in that window is what turns it into a finding.
Where The Reports Live
The tool groups its standard reporting around the same sequence as the three questions, which is convenient. Rather than describing screens that will be rearranged, here is which report answers which question and where each is covered properly.
Where people came from. The acquisition reporting.
Sources, with the caution that a source label is an inference rather than a fact. Our traffic sources guide covers why one of those labels is largely a bin for everything that could not be identified.
What they did. The engagement reporting.
Pages, time and actions. The useful reading here is which pages people actually arrive on rather than which the business considers important.
Whether it led to anything. The monetisation reporting.
Empty unless somebody configured it. Of limited use to a service business unless results carry a value. That is a configuration problem rather than a reporting one.
Whether they came back. The retention reporting.
Genuinely useful for subscription and repeat purchase businesses and largely skippable for a trade. The standard reports guide covers all four properly.
When to build something custom. Rarely, plus later.
Most businesses never exhaust the standard set. Building reports before you have used what exists is how people end up with a workspace nobody opens.
Reporting Is Not Fixing
A business that reads its reports every month while changing nothing has achieved the same as a business that never opened them. That sounds severe and it is simply arithmetic: nothing in a report alters a website.
Why the ritual persists. It feels like management.
Figures are reviewed, a summary is circulated and everybody has done something. No decision was required at any point, which is precisely why it survives.
What we have inherited. Two years of reports.
Accounts where monthly reporting was delivered faithfully and the website was not touched once in that period. The reporting was accurate and it produced nothing.
What acting looks like. Small and dated.
One page improved, one form simplified, one source pursued. Written down with the date so next quarter you can tell whether it worked.
The test to apply. What changed because of this.
Ask it after every review. If the answer is nothing three months running, the review is a habit rather than a management tool.
Where the argument lives in full. Our measuring performance material.
That covers what to report, to whom and why activity gets mistaken for progress. Everything on this tool sits on the Google Analytics guide.
Reading the
reports changes
nothing on
your website.
We have inherited accounts with two years of faithful monthly reporting and not one change to the site in that time. The reporting was accurate. It produced nothing, because nobody was ever asked to decide anything as a result of it.
What a useful monthly review covers:
If your property cannot tell you whether anybody enquired, that is the first thing to fix and it is a half hour job.
Every guide.
One tool.
Getting it installed and configured properly, reading the reports without being misled, measuring outcomes rather than activity, the advanced reporting and what to do when the numbers look wrong.
