Understanding Impressions Clicks and Position Changes in SEO
These three numbers are always read separately and they only mean anything together. A fall in one alongside a rise in another is usually good news. Almost every alarmed email about search performance is somebody reading one of them alone.
Three Numbers That Only Work Together
How often you appeared. How often somebody chose you. Where you typically sat. Read any one of those alone and you will reach the wrong conclusion more often than not, because each one moves for reasons the other two explain.
Why they are inseparable. They describe one event.
A search happened, your page was among the results, somebody either chose it or did not. Splitting that into three figures and reading one is like judging a conversation from a single word.
The commonest error. Position alone.
Somebody sees an average position get worse and concludes things are deteriorating. Block five explains why that reading is usually backwards.
The second commonest. Impressions alone.
A large rise in appearances treated as growth. Block two explains why appearing more is not the same as being seen more.
Where these numbers live. Search reporting.
They come from the search side rather than from your website, which is why they will not match your site analytics. Our performance report guide covers where to find them.
What this page does not cover. The tool.
Which buttons to press belongs with our Search Console guides. This page is about what the numbers mean once you are looking at them.
What An Impression Actually Counts
An impression is your page appearing in a set of results. That is all it is. It does not mean anybody saw it, read the heading or scrolled far enough down to reach it, which is why this figure can rise enormously without a single useful thing happening.
What it is counting. Presence.
Your page was among the results a search returned. Whether the person ever laid eyes on it is a separate matter the number knows nothing about.
Why that matters. Position changes everything.
An appearance near the top and one several screens down count identically. Two sites with the same impression figure can be having completely different months.
What makes impressions rise. Usually breadth.
Appearing for more different searches rather than appearing better for the same ones. That is genuine progress and it is the earliest sign work is taking effect.
What else makes them rise. Things you did not do.
A subject becoming topical, a seasonal peak or a competitor disappearing. All produce more appearances without anything changing on your site.
The reading to avoid. Impressions as an achievement.
A report leading with a large rise in appearances is reporting the easiest number to move. On its own it says almost nothing.
The reading that works. Alongside clicks.
Appearances rising and clicks rising with them is progress. Appearances rising alone is a question rather than an answer. Block six covers it.
A Ratio, Not A Score
Click through rate is clicks divided by impressions. It is a relationship between two other figures rather than a measurement of anything in its own right, which means it can move without either underlying number doing anything interesting.
Why that matters immediately. Two ways to change it.
The rate improves if more people click. It also improves if you appear less often. One of those is progress and the other is the opposite. The ratio alone cannot distinguish them.
What actually drives it. Position, mostly.
Where you sit affects how many people choose you more than anything you write. A ratio compared across different positions is comparing two different situations.
The second driver. The kind of search.
Somebody searching your business name behaves completely differently from somebody researching a subject. Mixing those together produces a figure describing neither.
Why industry averages are useless here. Nothing is comparable.
A published average combines every kind of search across every kind of business. Comparing your ratio against one tells you nothing. We publish no such figures.
What it is genuinely useful for. Yourself, over time.
The same searches, on your own site, across periods. That comparison holds the other variables still and it can tell you whether a change to how you appear did anything.
The other legitimate use. Spotting a mismatch.
Appearing well for something and almost nobody choosing you usually means the search wanted something other than what you offer.
An Average Of Averages
Average position condenses every appearance you made, across every search, every device and every location, into one figure. Once you understand what is being averaged, most alarm about this number disappears.
What goes into it. Everything, equally.
A prominent appearance on your most valuable search and a distant one on something irrelevant contribute the same weight. The figure does not know which you care about.
Why it moves without you. The mix changes.
Start appearing for a batch of new searches you rank poorly for and the average worsens, while nothing about your existing placements changed at all.
The location effect. Frequently overlooked.
For anything local, results differ across a town. The average blends places where you appear well with places you barely serve.
What it cannot tell you. Any specific placement.
You cannot work out where you sit for a particular search from this number. It was never trying to answer that question.
Why it gets reported anyway. One number, one arrow.
It condenses beautifully into a report, which is precisely the property that makes it misleading.
How to use it properly. Narrowly.
Look at the average position for one specific search over time, rather than at a site wide figure. Narrowed that way it becomes informative.
The rule worth remembering. Worse can mean better.
A worsening average frequently accompanies genuine improvement, which is block five.
The Combination That Looks Bad And Is Good
Impressions up sharply. Average position down. Clicks up. Almost everybody who sees that pattern reads the middle figure and worries. It is one of the healthiest things that can appear in a search report and it means the work is doing exactly what it should.
What is happening. Breadth arriving.
The site has started appearing for a much wider range of searches. Many are new, most are specific and you rank modestly for a good number of them.
Why the average worsens. Arithmetic.
Every new low placement enters the average. Add enough of them and the figure drops, even though not one of your existing placements got worse.
Why clicks rise anyway. More chances.
Appearing for far more searches produces more visitors even at modest placements. The extra breadth more than covers the lower average.
Why this pattern is so valuable. It is how growth starts.
Breadth almost always comes before height. A site being understood better appears for more things first and improves on the important ones later.
What happens if you react to it. Real damage.
Businesses see the average fall, conclude something broke and change things that were working. That is how a genuinely good month becomes the start of a bad quarter.
How to confirm it is this. Check the searches.
Look at what the new appearances are for. If they are relevant to what you sell, this is the pattern and it should be celebrated rather than investigated.
How to report it. Explain it once.
Say plainly why the average moved. A client told in advance will not panic when they see it.
The Combination That Looks Fine And Is Not
Position holding steady. Impressions flat. Clicks falling. Nothing in that looks alarming and it is the pattern most worth investigating, because something has changed in how your results are being presented rather than in where they sit.
What it usually means. The page changed around you.
New features appearing above the ordinary results push everything down the screen without altering anybody's position. You are where you were and fewer people reach you.
The second explanation. The answer is being shown.
Where a search is satisfied directly on the results page, people no longer need to click. Common on anything answerable in a sentence.
The third. Somebody more appealing arrived.
A competitor changed how their result reads. Or a well known brand entered the space. Your placement held and the choice changed.
The fourth. How you appear got worse.
A rewritten heading or description that reads less usefully. Rare, easy to check and simple to correct.
How to tell which. Search it yourself.
Look at what the results page for that search actually contains now. Two minutes of looking usually explains the whole thing.
Why it gets missed. Nothing looks wrong.
No figure in the report went red. Only the relationship between them changed, which is why the three have to be read together.
Where the diagnosis continues. Its own guide.
Interpreting a traffic drop covers working out what happened and in what order to check.
Reading It Properly
Reading this data well is a small set of habits rather than a skill. Four of them, applied together, prevent most of the wrong conclusions people draw from search reporting.
Compare like periods. The first.
The same span against the same span a year earlier. Comparing a short month against a long one produces confident nonsense, as does comparing a quiet season against a busy one.
Segment before concluding. The second.
By search type and by page. A flat total frequently hides one section growing while another falls away. Only the split reveals it.
Know the data is imperfect. The third.
Some of it is sampled and some is rounded. Very rare searches are held back to protect privacy. That underrepresents exactly the small specific searches a local business depends on.
Read the three together. The fourth.
Never form a view from one figure. The whole argument of this page in one habit.
What to record alongside. What you changed.
A dated note of every change to the site. Without one, no movement in any of these numbers can be explained afterwards.
What not to do. Check it daily.
These figures move constantly for reasons nobody controls. Watching closely produces reaction to noise, which our measurement guide covers.
Branded And Non-Branded Are Different
Searches for your business name and searches for what you sell behave so differently that they are effectively two separate reports mixed into one. Splitting them apart reveals more than any other single action available here.
How your own name behaves. Strong everywhere.
You appear at the top, almost everybody clicks and the figures look excellent. That is expected, since those people were already looking for you.
How everything else behaves. Considerably worse.
Lower placements, fewer people choosing you and far more variation. That is the genuinely competitive half and it is what the work is aimed at.
What mixing them does. Hides both.
Your name lifts every combined figure, which makes reach look healthier than it is and masks any decline in it.
The trap in a good month. Recognition rising.
Other marketing, a busy season or local attention drives people to search your name. Every combined number improves and reach did not move at all.
What each half tells you. Two businesses.
Branded searches measure whether people who know you can find you. The rest measures whether people who do not know you are finding you. Only the second is bought.
How to do the split. Filter by your name.
Divide the searches into those containing your business name and those that do not, then read each set separately. That is the whole exercise.
What to do with the result. Report both.
Two figures, always labelled. Never one combined number, since that is where the misleading happens.
Do Not Report These Alone
These three belong in a report as context rather than as the result. They describe what happened in search, which is a step towards the thing a business is paying for rather than the thing itself.
Why they end up leading. Availability.
They can be gathered without asking the business for anything, which makes them the easiest content for a report. Ease is not a reason to lead with something.
What they cannot tell you. Whether anybody bought.
Every figure here stops at the click. What happened afterwards lives in your analytics and in your own records.
Where they belong. Below the outcome.
After enquiries, as the explanation for why enquiries moved. That is the position they earn and they earn it easily.
How to present them. As a pair, at least.
Never one alone. Appearances and clicks together, with any comment on average position accompanied by the reason it moved.
What to say about the average. Once, in advance.
Explain that it can worsen while things improve. A client who knows that will not be alarmed later.
Where the full argument sits. The reporting guide.
What belongs in a monthly report covers it. Everything sits on the measuring performance guide.
Your average
position got
worse because
things are
going well.
Appearing for a much wider range of searches means new low placements entering the average, so the figure drops while not one of your existing placements got worse. Clicks rise anyway, because breadth produces more visitors even at modest positions. Businesses see that middle number fall, conclude something broke and change things that were working.
How we read these three:
Position holding, impressions flat and clicks falling is the pattern nobody investigates. Nothing in the report went red and something changed all the same.
Every guide.
One question.
What each number actually means, how to read the data without being misled, what belongs in a report and what does not, then how to connect any of it to the enquiries the business is actually paying for.
