One warehouse. MMM-calibrated attribution, contribution-margin P&L, predictive LTV, and monitoring that keeps every number defensible — from Shopify, your ad platforms, marketplaces, email and GA4.
Meta reports one ROAS. GA4 reports another. Shopify reports something else entirely. Your CMO picks whichever number looks best for the board deck — which means nobody actually knows what’s working.
Reconciling those numbers is step one, and most teams never even get there. But it isn’t the whole problem. Even once every source agrees, last-click still can’t see the brand and prospecting spend that creates the demand — it only credits the click that harvests it. And ROAS still counts revenue, not profit: it ignores the COGS, shipping, fees, discounts and returns that decide whether the sale made you any money.
Marketing Data Foundations fixes both layers. In three to four weeks your Shopify, ad platform, marketplace, email and GA4 data lives in one warehouse — reconciled, modelled on contribution margin, and calibrated with marketing mix modelling so demand-creating spend finally shows its true worth.
Meta, GA4 and Shopify all report different revenue. Foundations reconciles every source into a single blended CAC and MER figure — the one that goes in the board deck, not the three that contradict each other.
First-touch, last-touch and linear all share last-click’s blind spot: they can only credit a tracked click. We add marketing mix modelling (MMM), so brand, prospecting and view-through spend that creates demand finally shows its real contribution — and iOS and cookie loss stop quietly eroding your numbers. Reported ROAS becomes incremental ROAS: what the spend actually caused, not what the platform took credit for.
ROAS ignores fulfilment, returns and cost of goods. We rebuild CAC, MER and every channel on contribution margin — plus CAC payback by cohort — so you optimise toward the customers and products that make money, not the ones that merely make revenue.
90-, 180- and 365-day LTV, plus predicted LTV at the moment of acquisition, by channel, cohort and SKU. Then we push that value back into Meta CAPI and Google as the conversion signal — so the platforms' own bidding optimises toward customers worth keeping, not just orders worth counting.
Freshness checks, source reconciliation and anomaly alerts to Slack — “Meta CPA up 38% versus its 7-day average” lands before your CMO asks. The layer stops being a monthly report and becomes infrastructure your team relies on daily.
Every model is built in dbt with full documentation, tests and CI, and self-serve dashboards in the BI tool your marketers already use. When you hire your next analyst, they inherit a codebase they can read and extend — not a maze of legacy SQL no one remembers writing.
The core package is fixed-scope and yours to own. When you’re ready to go further, each layer below builds on the same warehouse — no re-plumbing, just more leverage.