Keyword Metrics
Keyword metrics fall into two groups: the ones Google publishes and the ones vendors invented. The difference matters — decisions can rest on the first, while the second only sorts a list. This lesson covers what each number actually means per the documentation, and which metrics do not exist at Google at all.
It rests on the Keyword Planner metrics reference and the Search Console API documentation.
Search volume: an approximate yearly average
The metrics reference defines the average searches field word for word: "approximate number of monthly searches on this query, averaged for the past 12 months". Two words there matter more than the rest: approximate and averaged.
The consequence that breaks planning: seasonality vanishes from that number. A query with 12,000 searches in December and none the rest of the year reports a thousand a month — exactly like flat year-round demand. So the average is read next to the monthly breakdown, which the same source returns as a separate per-month field, and checked against Google Trends.
"Competition" is about ads, not SEO
The most common substitution. The competition index in Google's own data is computed like this: "the level of competition from 0-100 is determined by the number of ad slots filled divided by the total number of ad slots available". That describes the ad auction. It says nothing about how hard the organic top is to reach.
The bids deserve precision too: the high and low ranges are the 80th and 20th percentiles of top-of-page bids, not "the cost of a click". As an indirect sign that a topic carries money, useful; as a difficulty metric, no.
What that means for decisions
| Metric | What it actually is | How to use it |
|---|---|---|
| Average monthly searches | An approximate 12-month average | Order of magnitude, not a traffic forecast |
| Monthly breakdown | Approximate values per month | Seasonality and publication timing |
| Competition index | The share of ad slots filled | A sign of commercial demand |
| Bid ranges | The 20th and 80th percentiles of bids | How much money the topic holds |
| KD / difficulty | A vendor's own estimate | Sorting within one tool, nothing more |
Your own data: what the Search Console docs state
Here the numbers are not estimates — but not as simple as they look either. Three facts from the API documentation:
- Position is an average. The field is defined as "average position in search results". "Position 8.4" does not mean the page sits eighth; it means impressions are spread across places.
- Fresh data is incomplete. The data state parameter separates finalized from fresh data: when recent data is requested, some of the rows returned may represent incomplete data. A drop "over the last three days" is often just unfinished counting.
- Aggregation by site and by page differ. The documentation points straight at the explanation of how data is calculated by site versus by page. The sum over pages need not match the site total.
Pulling all of it out is easiest through Search Console export, comparing periods through wins and losses, and lost clicks on retained impressions through lost clicks.
The order of evaluating a query
- The task. What the person wants to do and which format the results show — see the lesson on search intent.
- The business tie. Whether we have an answer and whether the topic leads to money.
- The order of demand. The approximate average plus the monthly breakdown.
- Competition as it is. Who holds the top — the composition of the results, not KD.
- Value. What traffic of this kind brings: conversions by query and the link between positions and revenue.
- The decision. Take it, merge it into an existing page, or decline.
Mistakes
- Treating volume as a traffic forecast. It estimates demand, not clicks.
- Comparing KD across different vendors.
- Reading Keyword Planner's "competition" as SEO difficulty.
- Drawing conclusions from the last few days in Search Console, where data is not finalized.
- Hunting for the cause of a drop in metrics rather than in the documented order for debugging traffic drops.
- Judging a query by numbers without opening the results: format beats any metric — see the lesson on what a keyword is.
What comes next
Next the metrics turn into money: valuing a query and setting priorities are covered in the following lessons of this module. The pages overview that a check usually starts from is the pages report, and the base definitions of impressions and clicks are in the performance report documentation.