Budget allocation review
Channels weighed on marginal return rather than blended averages, with money moved when the next dollar has a better home.

Operating discipline
Platform expertise is table stakes. The work that changes results is allocation, testing cadence, and refusing to let reporting flatter a channel that is not earning its place.
Every platform page in this pillar covers how a specific channel works. This page covers the part that sits above all of them: how budget gets divided, how tests are run, and how a program is judged when several channels are influencing the same purchase.
That distinction matters because most underperforming paid programs are not failing on execution inside a platform. They are failing on allocation. Budget sits where it has always sat, a channel keeps its share because it reports well, and nobody asks whether removing it would actually change revenue.
Incrementality is the question underneath all of it. A channel that reports conversions is not necessarily producing them. Branded search and retargeting are the usual suspects, since both frequently take credit for demand that another channel created and would have closed anyway.
A management engagement is the ongoing practice of asking those questions on a schedule, in writing, with the authority to move money when the answer changes.
What it is
Allocation, testing, hygiene, and an honest account of what measurement can and cannot tell you.
Allocation is treated as a recurring decision rather than an annual plan. Channels get reviewed against marginal return, meaning what the next increment of spend is likely to produce rather than the blended average of everything spent so far. A channel can look efficient overall and still be the wrong place for the next thousand dollars.
Creative testing runs on a cadence because fatigue is predictable. Different platforms decay at very different speeds, so the production calendar is set per channel rather than uniformly. Tests are structured so that a result can be read, which usually means changing one meaningful variable rather than shipping a new concept and calling the difference a learning.
Exclusion and negative hygiene is unglamorous and consistently one of the largest available savings. Negative keyword lists, placement exclusions, audience suppressions for existing customers, and geographic trims all leak quietly. We review them on schedule rather than when something looks wrong.
Attribution gets discussed honestly, including its limits. In-platform reporting counts generously and every platform claims the same conversion. Analytics applies a different model and produces a third number. Rather than pretending one is truth, we agree on a primary source, watch total revenue against total spend, and use holdout or geo tests when a channel's real contribution is genuinely in question.
Fit
We would rather say no early than sell a program that cannot work.
Deliverables
The work that happens above the platforms.
Channels weighed on marginal return rather than blended averages, with money moved when the next dollar has a better home.
Structured questioning of which channels create demand and which capture it, using holdout or geo tests where the stakes justify them.
A per-channel production and rotation schedule set against real fatigue rates, with tests built so results can actually be read.
Negative keywords, placement exclusions, customer suppressions, and geographic trims reviewed on schedule instead of after a bad month.
One report reconciling in-platform numbers against analytics, with the gaps explained rather than quietly averaged away.
A short document stating what we are changing, what we are not, and what we would stop funding if the decision were ours alone.
How we run it
A repeating loop, not a quarterly scramble.
Platform reporting and analytics are compared first, so the month's discussion starts from an agreed set of figures.
Each channel is assessed on what additional spend would likely return, not on how well it reported last month.
Negatives, exclusions, suppressions, and geography are worked through as a checklist so the quiet leaks stay closed.
Planned creative and structural tests go live on the cadence each channel needs, with the read date set in advance.
Changes made, changes declined, and the honest call on any channel that is not earning its share.
Budget allocation across channels, creative testing on a set cadence, negative and exclusion hygiene, reconciled reporting between platforms and analytics, and a written monthly recommendation on where spend should move.
It is the layer above individual platform work. Execution inside a platform matters, but allocation between platforms is usually where the larger result sits.
Each platform claims a conversion it touched, using its own window and model, so the same sale is counted several times. Analytics applies a single model across all of them and produces a lower, more conservative number.
Where this connects
Allocation decisions only make sense against the channels being allocated.
Search usually anchors the plan, so allocation starts by sizing what is left on the table in Google Ads before any budget moves toward less proven channels.
Account-level social structure and measurement sit in Meta ads which is where most cross-channel creative volume questions surface first.
Video is the channel most often judged unfairly by last-click reporting, which is why YouTube ads gets measured on downstream demand rather than its own click.
The channel most likely to overstate its contribution is retargeting so it is the first place we test whether reported conversions are actually incremental.
Reconciled numbers depend on the measurement foundation built in conversion tracking since inconsistent definitions make every allocation argument unwinnable.
Questions
Allocation reviewed monthly, in writing, including what we would cut.