Local Inventory Ads Built on Verified Store Stock

The number this depends on: Store-level inventory accuracy. We do not launch a store whose cycle count accuracy sits below 95 percent, and we measure it per store rather than accepting a company average, because the average hides the two stores that will generate the complaints. Google verifies the local feed against a sample of items at store level, and an ad that sends a buyer to an empty shelf costs the store more than the click cost. What we check before anything is built: Cycle count cadence and results per location. Shrink by category. The gap between POS on-hand and sellable on-hand, which is where damaged units, reserved stock and apartado quietly live. Whether counts are corrected in the system or on paper. If the assessment says the accuracy is not there, the recommendation is to fix counting first, and we would rather deliver that finding than a launch that burns the channel's credibility internally. The feeds and the store link: Primary product feed plus the local product inventory feed, with store codes matching Business Profile location codes exactly, the location group linked in Merchant Center, and every store in the feed verified with correct hours. Hours are where local inventory ads fail quietly, particularly around Semana Santa and December, when a store shows open in the profile and closed in reality. Pickup later and pickup today go live only for stores that can genuinely hold an item, and on-display-to-order is used where the unit is showroom stock rather than sellable stock. POS and ERP reality in Mexico: What actually sits behind the counter is usually Aspel SAE, CONTPAQi Comercial, Microsip, SAP, Oracle Xstore or Shopify POS. Most of them can drop a delta file on SFTP hourly; few push near real time without middleware someone has to maintain. We size the refresh interval to what the system can honestly sustain and set the availability buffer to cover the remaining gap, rather than promising a fifteen-minute refresh that nobody in the building can deliver. Measurement: Store visits or store sales, and which one applies to you is stated before the build, not after. Matched transaction upload from the POS with hashed identifiers where the transaction data supports it; modelled store visits where volume qualifies for them. This matters because it decides what the channel is allowed to claim internally, and a claim that outruns the measurement is the fastest way to lose a budget line the following year. Not included: Business Profile creation and verification for unverified stores. POS integration development, we specify the file and the interval, your team or your vendor builds it. Cycle count program design and staffing, we set the cadence, your operations team runs it. In-store merchandising and signage. Ongoing campaign and budget management past the initial structure. Who this is not for: Retailers without store-level inventory in a system at all. Retailers unwilling to commit to counting, since everything here rests on that. Single-store businesses, where the build cost is larger than local demand can return and a well-maintained Business Profile with free local listings does the same job for nothing. Chains that want a national average pushed to every store feed, which is a way of being wrong at every location simultaneously.