Danmarks-Posen
-18%
Cost per click, Shopping ads
- -14%Cost per click, whole campaign
- +21%Impression share
- +11%Clicks, whole campaign
Danmarks-Posen left Google's CSS. Their Shopping clicks got 18% cheaper.
How to read the numbers
Danmarks-Posen is a Danish webshop that sells bin bags and bins. In March 2026 they moved their Google Shopping ads from Google's own CSS to The CSS Partner. Same products, same ads, same campaign. Here are the numbers from their own Google Ads account, comparing the 30 days before the switch with the 30 days after.
- -18% Cost per click, Shopping adsThe ads built from the product feed, where the CSS applies. 30 days after the switch compared with 30 days before.
- -14% Cost per click, whole campaignAll ad formats in the Performance Max campaign together.
- +21% Impression shareFrom 23.8% to 28.8%, for the whole campaign. Google does not split impression share by ad format.
- +11% Clicks, whole campaignWith 4% lower spend. Performance Max moved part of the budget from the Shopping ads to other formats.
Across the campaign, Danmarks-Posen got into more auctions and paid less for each click, while total spend went slightly down. Impression share lost to rank fell from 72.5% to 65.6%: with the margin gone, their bids reached further.
Impression share is the share of the auctions their products could appear in where they actually appeared. Without the margin, the full bid enters each auction, so the same money can win more of them. Here, impression share rose while spend fell: the money went further, not up.
This is one account over 30 days, not a controlled test. A CSS switch moves the whole Merchant Center account at once, so there is no unchanged half to compare with. Impression share also moves for reasons that have nothing to do with the CSS. Treat it as an example of what the margin is worth on a real account, not as a promise of the same figures.
Source: Danmarks-Posen's Google Ads account, their Performance Max campaign. The Shopping figure covers the ads from the product feed in the same campaign. Periods: the 30 days before (15 February to 16 March 2026) compared with the 30 days after (20 March to 18 April 2026). Retrieved 28 October 2026.
The full method, and why we are careful with it, is on our measurement page.