Most purchases still happen offline. If your ads run online and your sales close in a store, a showroom, or a call centre, your ad platforms are optimizing on the minority of your revenue they can see. This playbook closes that gap, in the sequence we use on real implementations.
Step 1: Find where offline truth lives
Retail: POS exports, usually files landing somewhere nightly. Auto: the dealer management system and test-drive bookings, often behind a CRM. Add call-centre logs and, increasingly, a warehouse table that already consolidates all of it. You need event time, value, and customer identifiers (email, phone). You do not need loyalty-card completeness; partial identifier coverage still moves match rates meaningfully.
Step 2: Decide the matching strategy
Two mechanisms, best together. Click IDs (GCLID for Google, fbclid for Meta) captured at the lead or online order and carried through to the offline record. And hashed identifiers, which platforms match against their logged-in users. Auto funnels are click-ID friendly because a test drive starts with a form. Retail leans on identifiers because walk-ins never clicked anything.
Step 3: Pick destinations
Meta Offline CAPI, Google Store Sales for in-store transactions and Offline Conversion Import for click-based outcomes, Snap and TikTok offline events where those channels matter. Send to all of them from the same source; the formatting differs, the data does not.
Step 4: Handle privacy before anyone asks
Hash at source with SHA-256 after normalization. Carry consent flags per record. Send match keys, never raw PII. If your legal team wants the pipeline inside your own cloud account, that is a deployment choice (Datahash Core), not a blocker.
Step 5: Read the feedback and iterate
Every platform reports match results per batch. First upload sets the baseline; the fix list is almost always identifier normalization (phone formats, casing before hashing) and coverage (capture email at POS more often). Expect the second month’s match rate to beat the first by a distance.
Know the Google Store Sales thresholds before you plan around them
Retail chains, restaurant groups, and automotive OEMs and dealer groups usually want Store Sales specifically, because it matches in-store transactions back to the ads that preceded them: purchases at the till, reservations, test drives that turn into a car sale. It does that well, but Google gates it by volume. Plan on uploading at least 30,000 transactions within a 90-day window before store sales reporting appears in Google Ads, and some accounts need more. The account also needs around 300,000 reported store visits and 500,000 ad interactions over the same 90 days. Cadence matters as much as volume: upload at least weekly, because irregular uploads make the data harder for Google’s systems to process, and a long gap can cost you the reporting entirely.
Below those thresholds, nothing else in this playbook changes. Offline Conversion Import has no comparable volume gate, and neither does Meta Offline CAPI. Send Store Sales batches weekly from day one and let the volume accrue while the other destinations produce results.
What changes when it works
Platforms start optimizing toward customers who buy offline, which usually means older, higher-value, less cookie-visible buyers your campaigns were structurally underweighting. Retail chains attribute in-store revenue to YouTube and Search. Auto teams see which campaigns produce showroom visits rather than form fills. Budgets follow, and this time they follow evidence.