Impressions are not traffic, and our own two numbers show the gap

An impression is Google showing your page to somebody. A click is that person choosing it. They sit in adjacent columns of the same Search Console report and they are not versions of one another, which is why a supplier quoting impressions as traffic is quoting the wrong column. The cleanest demonstration we can offer is our own: the export on our proof page, taken 2026-10-14, records 5,965 impressions and 3 clicks on aiknowsus.com across the window 2026-10-04 to 2026-10-12. Same report, same nine days, same site. One number is in the thousands and the other is 3.

What each column actually counts

An impression is recorded when a link to your page appears in a result set a person saw. It needs the page to be indexed and judged relevant enough to be served, and nothing more. A click is recorded when that person chose your result over everything else on the screen. Between the two sits position, and position is the variable that converts one into the other. The export puts a figure on that: 2,685 impressions on 11 October 2026, the best day in the window, at an average position of 41.6. The pages were being counted. They were not yet standing anywhere a searcher was likely to reach.

That is the whole mechanism. Impressions measure whether Google will show you. Clicks measure whether a human will pick you. A number made of the first and labelled as the second describes nothing that happened.

The substitution, and where it happens in a report

Nobody writes impressions and calls it traffic in those words. The swap happens in three quieter places, and all three are easy to spot once you know the shape of them.

• The axis label is missing. A rising chart with no unit named on the vertical axis is usually impressions, because impressions are the number that rises first and furthest.
• The word used is visibility or reach. Both are real words and neither is a column in the report. Ask which column the figure came out of.
• A multiple replaces the pair. A growth multiple computed on impressions and presented beside a sentence about customers has quietly changed subject between the chart and the caption.

The three questions that separate the columns

When somebody hands you a figure, these three settle what it is in under a minute.

Ask this	What the answer tells you
Which column is this, impressions or clicks	Whether the figure is about being shown or being chosen
What is the average position behind it	Whether clicks could plausibly follow. 41.6 means not yet
What is the other column for the same window	The pair is the claim. One number alone is a selection

Our own answers to all three are published rather than described, which is the only reason this page is worth reading. The third question is the one most reports cannot survive, because the second column is usually the one that was left out.

Why we publish the pair rather than the better of the two

5,965 impressions on a domain that was registered days before the window opened is a real result, and it is the result we are proud of. Printing 3 beside it is what stops the first number being read as something it is not. Publish the impressions alone and you have claimed an audience. Publish both and you have claimed discoverability, which is narrower, true, and the harder of the two things to do from a standing start.

It is also the version that holds up when somebody checks. A single column can be quoted back at you with a different meaning attached; a pair with a window on it cannot, because the second number settles what the first one was.

What a reader may compute, and what they may not

With both numbers and the window, a reader can work out the share of impressions that became clicks for themselves, which is the point of giving them both. What they cannot do is infer anything about query intent, device mix, or the split between brand and non brand searches, because a performance total carries none of that. Nor is the share from one site transferable to another, which is why the position by position version for your own queries is worth the hour it takes: your own click share curve, grouped by position, answers a question an industry average cannot.

Questions about impressions, clicks and which column a figure came from

Is an impression worth anything on its own?

Yes, as a stage result. It proves the page is indexed and being served for real queries, which is the gate everything else sits behind. Treat it as evidence about indexing and relevance, and keep the business conversation on clicks and what follows them.

Why would impressions rise while clicks stay flat?

Because the pages are being served and not selected, and what decides selection is where they sit. Sort your queries by impressions, keep the ones whose average position sits below the first screen, and work those pages before touching anything else.

Does a high impression count mean the keywords are right?

Not by itself. Impressions accumulate on any query a page is judged relevant to, including ones nobody buys on. Check which queries produced them before reading a rising impression line as commercial progress.

Should a monthly report show both columns?

Both, with average position beside them and the window printed on the chart. Three numbers and a date fit on one line and remove the main way this figure gets misread.

What does 3 clicks in nine days tell a buyer about our product?

That we publish the column that does not flatter us. We sell AI visibility measurement and the page writing that follows it, so a number we report against ourselves is the only kind worth anything from a company with an interest in the answer.

What to do with a report that only shows you one column

Ask for the other one before you discuss the first. In practice that is a single line in a reply: send me the same window with clicks, impressions and average position together, and the date the export was taken. A supplier who has the figures will send them in a minute, because all three sit in one row of one report. A supplier who sends a longer explanation instead has told you which column the chart was. That request costs nothing and it settles more than any amount of reading the slide, and it is the same request we invite on our own figures, which is why both of our columns are printed on one page with the window across the top of them.