Journal · 21 September 2025
Product analytics asks what people did after they arrived. Marketing analytics asks which story, ad, or partner is allowed to claim the arrival. Those questions share a warehouse only in fantasy. They use different clocks, different identity rules, and different incentives. When the two dashboards disagree, the failure is not that someone cannot “align the numbers.” The failure is that nobody wrote the disagreement on paper.
A paid network may credit a download that product never sees as a first open, because the person never completed the ATT prompt, or because the SKAN window closed, or because a family iPad signed into the wrong Apple ID. Product may count a funded account that marketing cannot touch because the click happened in an email. Both can be true. The board pack that picks the larger number is doing politics, not measurement.
Attribution Without Guesswork is our room for this mess. We do not crown a model. We list claims. “This campaign may have influenced installs in this window, under these identity constraints.” That sentence is too long for a banner. It is the right length for a finance partner. Students who hoped for a single source of truth leave slightly disappointed, which we take as a sign the teaching landed.
Identity is the usual fracture. Marketing tools are built to stitch aggressively; product tools, if you are careful, are built to doubt. Identitycachegrid keeps drawers. We join them with a dated rule. Until that rule exists, comparing “users” across the two stacks is like comparing a postcode to a passport.
A practical truce: one shared definition of first value, owned by product, referenced by marketing as a conversion when — and only when — the identity is in the logged-in drawer. Paid reports can still show their own click metrics in an appendix. The meeting then argues about spend against funded accounts, not about which vendor is lying.
If your company needs the two graphs to match before anyone can work, you will wait forever. Write the gap. Date it. Revisit it when ATT, a new partner, or a rebrand changes the join rule. That is less comforting than a blended dashboard. It is also how you stay employable when the blended dashboard is later found to be a costume.