Frameworks

Operating manuals for AI-era marketing.

The Thinking section is the argument. The Frameworks are the architecture under it — built to be downloaded, calculated, and forwarded into a leadership meeting. Each one converts an abstract claim into a number a CFO can put in a deck.

Flagship · 14-page PDF

The 2026 Signal Architecture Framework

How the Google ecosystem, Meridian-grade measurement, and agentic AI fit together as one system. Three connected tools — the Signal Flow Map, the Meridian Readiness Checklist, and the Cost Per Decision Worksheet — designed to be used inside a real organization, by a real team, in a real budget cycle.

Free PDF · 14 pages Open →
Interactive Calculator

Cost Per Decision

Estimate the financial value of compressing the labor and capital required to make every marketing decision. The metric agentic AI actually moves. Outputs annual capacity unlocked, equivalent FTE capacity, and share of marketing payroll.

Live calculator Open →
Interactive Calculator · New

Decision Latency

Three fields. The dollar cost of every week your marketing decisions sit in committee while Google's AI bids on stale signals. The companion to Cost Per Decision — labor and capital on one side, time on the other.

Live calculator Open →

The problem it names.

Most enterprise marketing organizations are running a 2018 measurement stack against a 2026 ad system. Google Ads, YouTube, Performance Max, and Demand Gen are no longer channels you allocate spend across — they're one connected AI system that optimizes against the signals you feed it. The teams running it well have rebuilt their signal architecture on purpose. The teams stuck have not. The 2026 Signal Architecture Framework names what that rebuild actually consists of.

Who it's for.

Heads of growth, paid media leaders, and CMOs at companies spending $5M+/year inside Google. CFOs and finance partners reviewing a marketing technology investment. Agency strategy leads building a 2026 retainer model. Anyone who has been asked the question "are we using AI correctly in marketing?" and wants an answer that survives a board meeting.

What's inside.

  • The Signal Flow Map. A one-page diagram showing how YouTube, Search, and DV360 signals feed Demand Gen and Performance Max, with Meridian MMM and GA4 as the measurement substrate. Identifies the seven specific signal handoffs where most accounts leak.
  • The Meridian Readiness Checklist. Twelve checks to run before commissioning a Marketing Mix Modeling engagement — the data, the team, the cadence, the privacy posture. Designed to prevent the failure mode where a six-figure MMM project lands as a dashboard nobody trusts.
  • The Cost Per Decision Worksheet. The paper version of the live calculator, formatted to be walked into a planning meeting and filled in collaboratively.

When to use it.

Before the next annual planning cycle. Before the next agency RFP. Before the next "we should be doing more with AI" conversation reaches a budget decision. The framework is built to convert the abstract claim "AI changes marketing" into a number a CFO can put in a deck — which is the only version of the argument that survives contact with the operating plan.

The problem it names.

Cost Per Lead lives inside the marketing department's P&L. Cost Per Decision lives inside the entire operating expense line of the company. That difference is why CPL is a metric agencies optimize against, and CPD is a metric CFOs allocate against. The calculator converts the abstract argument (marketing decisions cost real money; agentic AI compresses that cost by orders of magnitude) into a single number — the annual capacity unlocked when the labor and latency of decisions is reduced by agentic systems.

Who it's for.

Marketing leaders preparing a board update on AI. Operators trying to size the ROI of an agentic-AI initiative before committing engineering hours. Finance partners stress-testing a marketing technology business case. Anyone who has heard the phrase "we should be using more AI" and wants to know what it's worth in dollars before they invest in finding out.

What it outputs.

  • Current Cost Per Decision — the fully-loaded labor cost of producing one marketing decision today.
  • Agentic Cost Per Decision — the modeled cost once agentic systems compress cycle time and human-touch.
  • Annual capacity unlocked — the headline number to take into a budget meeting. Not cash in a drawer; the financial value of returned human time, either as margin or as growth velocity.
  • Equivalent FTE capacity and share of marketing payroll — two readings of the same number, for the two audiences (HR and finance) who think about it differently.

Why this metric.

Because every other marketing metric is downstream of it. The full argument walks through why activity-based KPIs (CPL, CPA, CPM) have been a structural compromise since the discipline was instrumented, and why agentic AI is the first technology that makes Cost Per Decision computable on something faster than a quarterly cadence. The calculator is the executable version of that argument.

The problem it names.

Cost Per Decision is the labor and capital side of marketing decision economics. Decision Latency is the time side. Every week a marketing decision sits in committee is a week your bid strategy is allocating against stale signals, your creative is fatiguing against an audience that already moved, and your competitors are compounding learnings you didn't have the cycles to capture. Latency is a tax on every other metric. The calculator puts a dollar figure on it.

Who it's for.

Operators who have ever watched a decision die in approvals and known, in their bones, that it cost something — but didn't have a number to put in front of leadership. CMOs explaining why a faster operating cadence is a P&L issue, not a cultural preference. Agency leads making the case for in-flight optimization budgets. (Wondering whether your cadence is even unusual? The decision-latency benchmarks put a directional number on it by sector.)

The three fields.

  • Weekly media spend. What's flowing through the channels while decisions wait.
  • Average decision delay. How long the median optimization decision sits between "we should change this" and "it's live."
  • Expected efficiency delta. The percentage point improvement an in-time decision would have captured.

The output.

One number: the dollar value being left on the table by your current decision cadence, annualized. It is, deliberately, an uncomfortable number — and it is the companion measurement to the Cost Per Decision calculator, because labor and latency are the two levers agentic AI moves simultaneously.

More frameworks in development.

The Payer Signal Map (health insurance), the Retail Decision Stack, and the Agentic Memory Bank are next. Subscribers get them first.

Subscribe to Stay Sharp →