Every marketing leader has stood in a room and presented a number they were fairly sure was directionally right and privately unsure was exactly right. That gap between confidence and correctness is where most attribution problems live, and it rarely gets closed by adding another dashboard. It gets closed by understanding the specific ways measurement quietly lies, because they tend to repeat across almost every account.
Last-click takes credit for work it didn't do
Last-click attribution hands 100% of the credit to whatever touchpoint happened right before conversion, usually a branded search term or a retargeting ad. That touchpoint didn't create the demand. It caught demand that upper-funnel channels, content, or word of mouth had already built. When leadership sees branded search as the top performer quarter after quarter, they're often looking at a harvesting channel dressed up as a demand-generation one.
The fix isn't switching to a different single-attribution model, since every single-touch model has its own version of this bias. It's running the occasional incrementality test, holding a channel back for a defined period and measuring what actually happens to conversions, to check whether the credit a channel receives matches the value it's actually creating.
Platforms grade their own homework
Meta's reported conversions and Google's reported conversions are each measured inside their own walled garden, using their own attribution windows, and they routinely overlap in ways that inflate the combined total when added together naively. Add in a CRM reporting its own version of the same pipeline, and it's common for the sum of all platform-reported results to exceed total actual company revenue for the period, sometimes by a wide margin.
This is not fraud. It's structural. Each platform is built to make its own contribution look as complete as possible, because that is what keeps budget flowing to it. Leadership needs one source of truth outside any single platform, usually the CRM or a data warehouse, to reconcile against.
View-through conversions get counted like they're real
A view-through conversion means someone saw an ad, didn't click it, and converted anyway within some attribution window, sometimes as long as 28 days. Counting these at face value assumes the ad caused the outcome. Often it didn't; the person was already going to convert, and the ad simply happened to be shown to them at some point along the way.
View-through metrics aren't worthless, but they should never carry the same weight as a click-through conversion in a leadership report without a clear label distinguishing the two. Quietly blending them into one combined number is one of the fastest ways to overstate a channel's real impact.
Offline and sales-assisted revenue rarely makes it back into the model
For any business with a sales team, a long consideration cycle, or an offline component, a meaningful share of revenue traces back to marketing touchpoints that never get reconnected to the original campaign. The lead came from a paid channel, but by the time it closed three months later through a sales conversation, the CRM record has lost the thread back to the original ad.
This isn't a measurement nuance, it's often the single largest blind spot in the entire model. Fixing it usually means instrumenting the CRM to capture original source at the lead level and building the reporting to close that loop, not adding another marketing dashboard.
What a trustworthy attribution setup actually requires
- A single source of truth for revenue, independent of any ad platform's self-reported numbers
- Occasional incrementality or holdout testing on major channels, not just correlation-based models
- Clear separation between click-through and view-through conversions in any leadership-facing report
- A closed loop between CRM source data and original marketing touchpoints, especially for sales-assisted revenue
None of this is exotic. It's disciplined, and it takes longer to set up than trusting whatever number the platform shows by default. The businesses that do it stop arguing about whether marketing is working, because the number in the room finally matches the number in the bank account.