The problem this solves
Most channel budgets are last year's budget with a percentage adjustment, defended by a mix story that nobody has audited against the data in quarters. Meanwhile the actual mix drifts: a paid channel's costs creep up, organic quietly compounds, a channel that earned its budget two years ago now coasts on incumbency. The gap between the story and the reconstruction is where budget goes to underperform, and it persists because reconstructing the truth is a genuinely annoying analysis.
Annoying because the inputs disagree by design. GA4 attributes by its rules, each ad platform attributes generously to itself, and spend lives in yet another set of exports. Anyone who joins these sources honestly has to make assumptions, and the difference between a defensible mix analysis and a misleading one is whether those assumptions are written down. The usual spreadsheet, built under deadline, buries them, which is why channel debates so often end in dueling numbers instead of decisions.
The reality check does the reconstruction with its assumptions on the surface: real spend and real outcomes per channel with every figure traceable to its source pull, acquisition cost compared on one stated basis, and the reallocation question answered as a model with three scenarios rather than a single prescriptive claim. The budget conversation gets a floor of shared numbers to stand on.
How the mission runs
- Pull outcomes by channel from GA4. GA4 provides sessions, conversions, and revenue signals by channel over the analysis window, with the channel grouping and conversion definitions pinned in the output. This is the outcomes side of the ledger, stated on one consistent attribution basis that the model discloses.
- Pull true spend from the ad platforms. Google Ads and Meta Ads provide actual spend and platform-reported results per campaign, mapped to the same channel groupings. Platform-attributed conversions are kept as labeled reference columns beside the GA4 basis, so the model never silently blends two attribution worldviews.
- Reconstruct the real mix. Spend and outcomes join into the actual mix: what each channel truly costs, what it truly returns on the stated basis, and how that compares to the mix story the budget assumes. The gaps between assumed and actual mix are called out explicitly, because those gaps are the findings.
- Compare acquisition cost honestly. Cost per acquisition is computed per channel on the single stated basis, with the soft spots flagged: channels whose volume is too thin for a stable read, and organic's cost treated explicitly rather than pretended to be zero. An honest comparison with caveats beats a clean one that collapses under questioning.
- Build the three-scenario model in Sheets. The reallocation model lands in Google Sheets: current mix as the baseline, then three scenarios - such as consolidation into proven channels, a measured bet on an underfunded one, and a balanced shift - each with projected outcomes and the assumptions driving them, adjustable directly in the sheet.
The prompt
This is the exact objective the agent receives. Swap the obvious placeholders for your own domain, segment or channel and run it as-is from the console, Slack, or the API.
What comes back
A Google Sheets model holding the reconstructed truth about your channel mix: real spend and outcomes per channel on a stated attribution basis, acquisition cost compared with its soft spots flagged, and three reallocation scenarios with adjustable assumptions and projected outcomes. Every figure traces to the pull that produced it, so the budget meeting argues about which scenario to fund instead of whose spreadsheet to believe.
Make it yours
- Extend the outcome line into HubSpot pipeline data when your tracking supports it, so scenarios project pipeline impact rather than stopping at conversions.
- Run it quarterly with definitions held constant and add a drift tab, showing how the real mix moved against the budgeted mix since last quarter.
- Ask for a sensitivity view on the winning scenario: which assumption, if wrong, changes the recommendation, so leadership knows exactly what they are betting on.
Frequently asked questions
Whose attribution numbers does the model believe?
It commits to one stated basis for the comparison and keeps platform-reported figures as labeled reference columns beside it. The point is consistency you can inspect: every channel measured the same way, disagreements between sources visible rather than blended, and the basis itself swappable if your team prefers another.
Can the scenarios be adjusted after delivery?
Yes, that is why the deliverable is a model rather than a slide. Assumptions sit in editable cells with projections recalculating from them, so the budget meeting can test its own ideas live. The delivered scenarios are starting positions, and the sheet is built to be argued with.
How does this stay honest about organic's cost?
By refusing the zero-cost fiction. Organic carries real content and program costs, and the model represents them explicitly using inputs you provide, flagged as your figures. Treating organic's cost honestly is what makes its acquisition cost comparable to paid channels instead of automatically, misleadingly winning.