Ad Asset Architecture for Search and Performance Max

On mobile, assets can occupy more vertical space than the ad they belong to. In most accounts we audit, that space is filled by four account-level sitelinks written once, in a hurry, and never touched since. The audit first: We start from asset-level impression data — what is actually serving, at which level, and how often. Accounts routinely have thirty assets attached and four that ever appear. Anything that has not served in ninety days is either misassociated, outranked by an automated asset, or ineligible, and each of those has a different fix. Automated assets: Dynamic sitelinks, automatically created assets and dynamically generated descriptions write copy from your website without anyone approving it. For a pet supplement brand or a retailer running financing offers, that is a claim serving before a reviewer has seen it. We turn them off in those accounts and report what the change costs in traffic, because it is not free and pretending otherwise would be dishonest. In accounts with no claims exposure we leave them on and monitor them. The association matrix: Assets are mapped to campaign intent rather than bolted to the account and forgotten. Brand campaigns get deflection sitelinks — support, login, order status, pricing — because the person searching your name has usually already bought. Non-brand prospecting gets proof and comparison: the specification page, the returns policy, the trial terms. Conquesting campaigns get the one sitelink that answers the switching objection, and nothing else. Formats, as they actually behave: Sitelinks need four before the format serves, carry two description lines each on desktop, and appear as a swipeable row of up to six on mobile, so upload order matters. Structured snippet header choice matters more than the values underneath it. Image assets are cropped from source at 1200 pixels or better into 1.91:1, 1:1 and 4:5, composed for the square crop rather than letterboxed into it, with logos at 1:1 and 4:1. Price assets read from the same values as the Merchant Center feed so a promotion does not create a mismatch between the ad and the cart. Promotion assets carry real start and end dates and a discount type that matches the page. What we refuse to attach: Lead form assets on B2B software with a sales-qualified motion. The cost-per-lead improvement they show is a reporting artifact and the pipeline conversion rate on those leads is not close. Call assets without call tracking that reports a conversion at a minimum duration, sixty or ninety seconds depending on the sales motion — without that threshold, bidding learns to buy hang-ups. Location assets on ecommerce accounts with no store to visit. Delivery: An asset library with a naming convention, the association matrix as a sheet your team maintains, bulk upload through Google Ads Editor, image crops delivered as files, and a rotation calendar for anything seasonal so nothing expires quietly in the middle of a peak week. Not included: Landing page creation, product photography, video production, Merchant Center feed work, and Performance Max asset group strategy beyond the assets themselves. That last one touches budget and audience signal and is scoped separately. Who this is not for: Anyone who wants extensions added as a checklist item before a quarterly review. Accounts under forty thousand dollars a month in search spend, where the surface area does not justify the fee. And brands that cannot produce approved product imagery on a schedule — the image half of this work stalls without it, and we would rather scope around that at the start than watch it stall in week three.