Use impressions to learn what is true about your own site, and tool volume to size territory you do not yet rank for. They disagree because one counts what happened to you and the other models what happens to everybody.
What does each number actually count?
Impressions count the times a URL from your property appeared in a result a person saw. Search volume models how often a query was searched across all users, usually as a rounded twelve-month average.
That difference is not a matter of precision. Impressions are observed, bounded by your own ranking, and specific to your property. Volume is an estimate built from Google's own bucketed ranges plus panel data, then rounded to a presentable figure. One is a measurement with a narrow scope; the other is a projection with a wide one.
The rounding is worth dwelling on. Tools present volume as a single integer because a single integer is easy to sort a spreadsheet by, but the underlying figure is a band. Two keywords showing 1,000 a month may differ by a factor of two in reality, which is why ranking a keyword list by volume to three significant figures is a false precision that then gets built into a quarter's plan.
Where does each one mislead you?
Each has a known distortion, and knowing which one you are looking at is most of the skill.
| Signal | What it covers | Known distortion | Trust it for |
|---|---|---|---|
| Tool volume | All searchers, any site | Rounded, averaged over a year, close variants often grouped together | Relative sizing of territory you do not rank for |
| Keyword Planner volume | All searchers, any site | Reported in buckets and grouped across near-identical phrasings | Order of magnitude, and seasonality by month |
| Search Console impressions | Only queries where you appeared | Requires you to rank; rare queries are withheld for privacy | Whether what you shipped is being seen |
| Search Console clicks | Only queries where you appeared | Same withholding, plus deduplication by query and property | Which visibility converts into visits |
Why does a keyword with 5,000 volume show 40 impressions?
Because an impression requires you to appear. At position 40 you are several pages deep, where almost nobody scrolls, so the impression is never recorded even though every one of those searches happened.
A second mechanism compounds it. Search Console withholds queries that are rare enough to identify a person, so the sum of your query rows is always lower than the totals line above it. The gap is widest exactly where long-tail work lives, which means a report built by exporting the query table understates the thing the report was written to defend.
Grouping does the rest. If your tool reports one figure for a phrase and its three near-identical variants, while Search Console lists all four separately, you are comparing one grouped number to four ungrouped ones and concluding that reality fell short.
Volume answers "is this worth building?". Impressions answer "did what I built get seen?". Using either number for the other question produces work nobody can evaluate afterward.
Does rising impressions with flat clicks mean a problem?
Usually not. It most often means you gained visibility on queries you rank poorly for, which adds impressions at positions that were never going to produce clicks.
The sitewide line hides this because average position is an average across impressions. Add 20,000 impressions at position 30 and the line moves against you while nothing about your existing pages got worse. The way to tell is to split the query set: compare the queries that existed last month against the ones that are new, and read the click-through rate within each group rather than across both.
The same split matters for value. A set of newly visible queries with no buying intent will never show a conversion rate worth reporting, and that is a fact about the queries, not a failure of the pages. Reporting both groups as one number is how a good month gets described as a bad one.
Dividing your impressions by tool volume and calling the result your share of the market. Volume groups variants, impressions exclude withheld queries, and the two cover different populations, so that percentage is not a quantity that exists.
Which number should decide the next piece of work?
If you are deciding whether a page is worth writing, use volume, treat it as a band rather than a figure, and sanity-check seasonality in Google Trends before committing a quarter to a spike that only occurs in November.
If you are deciding whether the page you already shipped is working, use impressions, and only for the queries that page is meant to serve. Then the question in front of you is narrower than the dashboard implies: either you are not being seen, which is a ranking problem, or you are being seen and not chosen, which is a title and snippet problem. Organic search reporting goes wrong when those two get one chart, and that is the chart most teams are looking at.