New — 2026 Edition

Stop guessing which AI actions drove revenue. Start proving it.

A practical framework for building the measurement infrastructure that lets live and social commerce brands know — with confidence — which AI decisions actually moved the needle.

Tested against Whatnot's show format architecture and TikTok Shop's affiliate attribution model. Built for operators who've outgrown gut-feel AI.

The Problem

70% of your traffic is invisible to your AI. Here's why that's expensive.

Anonymous traffic that never converts. AI compounding an attribution problem you didn't know you had. DTC brands running 5–8 disconnected tools — each claiming credit for the same sale.

The incrementality problem is different from the attribution problem. Attribution tells you what happened. Incrementality tells you what would have happened without the action. That's the question every AI decision actually needs to answer.

What's Inside

12 chapters. One mission: tie every AI decision to a dollar outcome.

From measurement infrastructure to operating cadence — every chapter designed to ship, not just read.

01

The Measurement Stack Every Social Commerce Brand Needs (But Doesn't Have)

Summary: Before you can optimize AI decisions, you need a measurement layer that can answer "what would have happened without this action?" This chapter maps the four components of a complete measurement infrastructure and identifies where most brands have gaps.
02
Show Format Architecture: How Live Commerce Changes the Incrementality Question
03
The Attribution Gap: Why 70% of Your Traffic Is Invisible to Your AI
04

Holdout Testing at Scale: Designing Experiments That Produce Actionable Signal

Summary: Most holdout tests fail not because the experiment was wrong, but because the design assumed too much. This chapter walks through experimental architecture for live commerce, TikTok Shop affiliate, and creator content — with concrete parameters for each.
05
Creative-to-Revenue Signal: Closing the Loop Between Content and Margin
06
Platform Signal vs. Brand Signal: Knowing What Belongs to You

07
Pricing Intelligence in Real-Time: When AI Should Move Price (and When It Shouldn't)
08

Creator Attribution Beyond UTM: The Method That Actually Works

Summary: UTM-based creator attribution systematically under-credits long-cycle and organic creators. This chapter covers incrementality-based creator measurement: how to set up holdout tests, separate creator effect from platform effect, and feed that data back into creator investment decisions.
09
Automation Drift: Why Your AI Rules Need Quarterly Audits
10
The Operating Cadence: How Top-Performing Brands Run AI Reviews
11
Building the Feedback Loop: Dashboard to Decision to Revenue
12
The Incrementality Case Study: What $8B in Live Commerce GMV Teaches Us About Signal

Sample Chapter

Chapter 4 — Holdout Testing at Scale

A preview of what's inside the Playbook.

Running a holdout test on a TikTok Shop affiliate campaign without a proper experimental design is like running a clinical trial without a control group. You get a number — but you don't know what it means.

The three most common holdout test failures in social commerce:

Control group that's too small to matter. Most brands run 5–10% holdout because they don't want to "lose" revenue. But at that sample size, the variance is so wide the result is uninterpretable. You need 20–30% holdout to detect the effect sizes that matter in social commerce, where baseline conversion rates are often 1–3%.
Test running too short to capture the full customer journey. A 48-hour TikTok Shop campaign test captures impulse purchases but misses customers who discovered you in the campaign and came back three weeks later to buy. Social commerce has a long consideration tail that platform attribution can't see.
No separation between platform effect and content effect. When a creator post runs as a promoted ad, the platform algorithm and the creator content are both in the test — but you're only measuring one. Proper experimental design isolates the creator effect from the platform amplification effect.

The fix: design your experiments around the question you actually need answered, not the question your analytics platform is set up to ask.

Who It's For

Built for operators who've already tried the obvious.

⬡

DTC Brands

Running AI across your stack and need the measurement layer that proves what's actually working.

◈

Sellers & Creators

Building on TikTok Shop, Whatnot, or live commerce and making attribution decisions with incomplete data.

◉

Social Platforms

Running affiliate or creator programs and need a methodology to show partners the incrementality case.

◆

Agencies

Advising clients on AI ROI and need the framework and metrics to make attribution claims defensible.

◇

Investors

Evaluating AI investments in social commerce and need the vocabulary and metrics to assess signal quality.

Early Access — First Cohort

The Playbook. $299–$499.

One investment. Systematic clarity on which AI decisions are actually moving revenue.

Preview a full chapter before you buy →

Playbook Only

$299

12-chapter operational guide. Lifetime access.

Full 12-chapter Playbook (PDF + web)
Chapter templates and test frameworks
Operating cadence worksheets
All future chapter updates

Best Value

Playbook + Model + Q&A

$499

First 20 buyers get a 30-min Founder Q&A.

Everything in Playbook Only
AI Incrementality Model (spreadsheet)
3 bonus chapters (extended case studies)
30-min Founder Q&A call (first 20 buyers)

The full 12-chapter Playbook is shipping in batches to early operators. We're invoicing the first cohort directly so we can match each buyer to the right chapters and offer a 30-minute walkthrough. Reserve your access below — we'll be in touch within 24 hours.

Reserve Early Access — we'll invoice you directly.

First Name *

Last Name *

Work Email *

Company *

Role / Title *

What are you trying to measure or fix? *

Budget / Scope *

Just me / individual operator Team license (2–5 seats) Organization rollout (6+) Need to discuss

Anything else?

✓

You're in the first cohort.

We'll reach out within 24 hours with invoice details and full Playbook access. Watch your inbox — this comes from the founding team directly.

FAQ

Questions we get asked.

Is this relevant if I'm not on Whatnot or TikTok Shop?

The frameworks apply to any social commerce setup — Instagram Shopping, YouTube Shopping, live commerce on any platform. Whatnot/TikTok Shop examples are the highest-signal environments and easiest to use for illustration.

How is this different from the free AI Revenue Signal Playbook at /playbook?

The free Playbook is a broad introduction to the five most common AI revenue leaks. The AI Incrementality Playbook is the operational implementation guide — deeper on measurement infrastructure, holdout test design, attribution methodology, and operating cadence. They're complementary; the free guide earns the conversation, the flagship closes it.

Who wrote this?

Peter Griscom (President & COO, It Works! | CEO, Tradefluence | Director, Social Commerce Partners), Conrad (operating intelligence), and Forrest (attribution methodology). The playbook draws from operational experience across direct selling, DTC, live commerce, and social platforms at scale.

When will checkout be live?

We're building on Stripe Connect integration now. Enter your email on this page and we'll notify you the moment the Playbook is available — no spam, just one email when it's ready.

Does the $499 tier have limited slots?

Yes. The 30-minute Founder Q&A is offered to the first 20 buyers of the Playbook + Model bundle. Once slots are filled, the bundle stays available without the Q&A — or with a Q&A at standard consulting rates.

Is there a team or enterprise version?

Not yet. Email hello@untitled-1779731017832-46ao.polsia.app and we'll be in touch.

Written By

Authors

Peter Griscom

President & COO, It Works! · CEO, Tradefluence · Director, Social Commerce Partners

Conrad

Operating Intelligence

Forrest

Attribution Methodology