Ad Rank.

Ad Rank is not a stored property of a keyword but a value recalculated for every single search query. The same ad can therefore be in position one in the morning, position four in the afternoon and not appear at all for a slightly different wording. Once you understand this, you stop treating 'the position' of a keyword as a fixed quantity. Ad Rank itself is not visible in the Google Ads interface; you only see its proxies: the Quality Score, the average CPC and the impression share.

Several factors feed into the calculation. The bid is only one of them and merely states the maximum that would be paid. Added to it are the expected click-through rate, the relevance of the ad to the search query and the landing page experience, the three components that Google also summarises as the Quality Score. Further factors are the context of the query (location, device, time of day, wording), the expected impact of assets such as sitelinks or call extensions and, finally, minimum thresholds: if you do not reach them, you do not appear at all, even with a high bid.

An example from practice: two providers bid on 'emergency plumber Zurich'. Provider A bids CHF 12 per click but sends visitors to a general home page without a phone number in the visible area. Provider B bids CHF 8 and sends visitors to a page that describes exactly the emergency service, loads on mobile in under a second and shows the number as click-to-call. B achieves the higher Ad Rank in this auction, appears above A and pays less per click, because the actual cost per click is derived from the Ad Rank of the advertiser placed below, not from your own bid.

This leads to an uncomfortable order of priorities: landing page and ad text before budget. Buying visibility solely through higher bids permanently compensates for a weakness that would be cheaper to fix once. At the same time, Ad Rank is not a goal in itself. For expensive terms, position one is often the least profitable placement, because the premium over position two or three is rarely covered by more conversions. It is useful as a diagnosis: if delivery drops, it shows whether the cause is the bid, the quality or a minimum threshold that has not been reached.