IMAGE EXTENSIONS GOOGLE ADS: HOW THEY ATTACH, WHEN THEY LIFT CTR

The first thing worth saying about image extensions google ads is that the name is half wrong now. Google renamed extensions to "assets" back in 2022, so what advertisers still call image extensions is officially the image asset. The label changed; the behaviour did not. A thumbnail attaches to your Search text ad, and when it shows it sits beside your headline, description, and URL on the results page. In the accounts I manage I have learned to treat this feature as quietly powerful and quietly frustrating in equal measure: the upside is real, but the system fails silently in ways that are easy to miss if you only look at whether the upload succeeded. This article goes straight to what I actually observe, how a google ads image extension attaches, when it earns its place, and the requirements that get rejected without a useful warning.

How image extensions attach to a Search ad

An image asset is not part of the ad you build. It is uploaded separately at the campaign or ad group level and then paired with your text ad at auction time. When someone searches and your ad qualifies, the system makes a real-time decision about whether appending an image is likely to increase the click. If it judges yes, the searcher sees your image extensions google ads thumbnail next to the copy; if not, the ad shows text-only. You never control which of your uploaded images serves, nor which format combination appears, because that is auction-dependent. The thumbnail is small, a square or rectangle off to the side, never a large banner.

The cost model is the part most advertisers misread. A click on the image is billed exactly like a click on your headline, the same CPC, with no surcharge for the image itself. Uploading images adds no fixed cost; you only pay when someone clicks. This matters because it removes the usual "is it worth the spend" hesitation: there is no separate spend. The only cost is the production effort and the risk of an off-theme image dragging relevance down. If you are still mapping how visual assets fit your Search structure, the broader picture of image sizing across Google Ads formats is worth having in mind before you commit a creative set.

When image assets actually lift CTR

Google's own figure is an average 6% increase in click-through rate when image assets show with Search ads, per their official documentation on image assets for Search campaigns. That number is an average, and averages hide everything interesting. In my accounts the direction is consistent, the image rarely hurts CTR when it is relevant, but the magnitude swings hard by context.

The lift is largest in two situations. First, when the product is genuinely visual: furniture, apparel, physical equipment, anything where seeing the thing answers part of the buyer's question before they click. Second, when competitors on the same query run text-only ads. On a results page that is a wall of blue links, the only ad carrying a thumbnail draws the eye disproportionately. That is the inference I keep returning to: the image's value is partly relative to the page, not absolute. Where every competitor already shows images, the advantage compresses.

Where the uplift is small or invisible: commodity queries, pure price comparison, and informational searches where the searcher wants a fact, not a picture. A user typing "what is X" is not stopped by a product photo. I have run image assets on B2B software accounts where the CTR change was within noise, simply because the buying decision was not visual. The honest expectation is a modest, reliable gain when the image is relevant, never a campaign transformation. Treating it as a CTR lever to pull when the visual matters, rather than a default to bolt onto everything, is closer to how I actually use it, and it sits alongside the rest of a visual approach to Google image ads rather than replacing it.

The requirements Google rejects in silence

This is the section I wish someone had handed me earlier. The image spec is strict, and a surprising amount of the enforcement happens quietly, an image that is technically uploaded but never serves, or a relevance limitation that no banner shouts at you. Knowing the rules up front is the only reliable defence.

The hard specs first. A square 1x1 image is required, with 1200x1200 recommended and 300x300 the floor. The landscape 1.91x1 ratio is optional but recommended, 1200x628 ideal, 600x314 minimum. Accepted formats are PNG and JPG, and the file must stay under 5120 KB. Keep your important content inside the central 80% safe area, because the system can crop and the edges are the first to go.

The rejection rules are where accounts quietly bleed potential. Google rejects images with text or graphic overlays, including brand logos placed on the image. It rejects excessive whitespace, collages, blurry or distorted images, and crops that make the subject hard to identify. The frustrating pattern is that these do not always surface as a loud disapproval, sometimes the asset simply sits there approved-looking yet rarely serving. The full image policy and format requirements spell out the edge cases, and I treat that page as a pre-flight checklist rather than something to consult after a rejection.

There is also an account-level gate that catches new advertisers off guard. To use image assets at all, your account must be open at least 60 days, carry a clean policy history, have active campaigns and active text ads, and show Search spend above zero in the last 30 days, plus sit in an eligible vertical. Sensitive verticals are excluded outright. If the option to add images is missing from your interface, the answer is almost always one of these gates, not a bug.

My practical rule: never trust the upload confirmation as proof the asset is working. I check the asset report after a week of impressions. If an image was approved but is barely serving, that is the signal, either it is failing a relevance check against the ad group, or a competing image is winning every auction. Inferring the difference from serving data, not from the upload screen, is what separates an image set that performs from one that just exists.

Campaign level or ad group level: the relevance trade-off

You can attach images at the campaign level or the ad group level, and the choice is a straight trade between speed and relevance. Campaign level is faster, one set of images covers every ad group, which is fine only when every ad group genuinely sells the same thing. Ad group level takes longer but gives the tightest relevance, and in my experience the better serving rate, because the image can match the specific theme of the keywords in that group.

My default is ad group level for any account with distinct product lines, and campaign level only for tightly themed single-offer campaigns where one image set carries over cleanly. The reason traces back to the silent-rejection problem: relevance is the lever the auction uses to decide whether to serve, so a generic campaign-level image applied across mismatched ad groups is the most common cause I see of "approved but never showing." Aim for at least four unique, relevant images per group, including a square, and add a landscape to widen the format combinations the auction can build. Past four, add more only when you have genuinely different on-theme visuals, padding the slot with off-topic images quietly drags down the relevance score that decides serving. The discipline is the same one I apply to every visual asset: relevance and correct ratios first, volume a distant second.