Offer, creative, pages or retention: which drives growth?

Noah Haynes's teardown of Everyday Dose and Arie Scherson's teardown of Smooche show the same four pillars, and neither can tell you which one pays for the others. The pillars are a first order loaded with discounts and free gifts to start a subscription, hundreds of active ads, a page or advertorial for every angle, and retention behind it all. Your own data can tell you which one carries the growth, one pillar at a time.

The two videos, published on YouTube on 22 and 24 September 2026, both work from the outside: the offer on the page, the ads in the Meta Ad Library, the page each ad lands on, and, in the Everyday Dose video, what arrives by email and text after a sign-up. Both credit the growth to the whole playbook at once. From the outside, that is all anyone can do, and it is a fair way to learn the shape of a playbook. It is not a way to learn which part of it works.

Which part of the playbook is driving the growth?

You do not need all four pillars to learn from these videos, and you cannot read four at once. Each pillar has one number that tests it, and most of those numbers are already in your store, your subscription app and GA4.

The four pillars, and the number that tests each:

1. Offer. In the Everyday Dose teardown, the first order is built to break even or lose money, so first-order ROAS looks bad on purpose. The number is payback by cohort, loss included: First order at a loss? Judge it on payback, not ROAS. For the discount itself, see is your welcome popup discount incremental?, and for price, read margin per visitor.
2. Creative volume. An active-ad count shows testing, not results. Tag each angle and hold regions out: Hundreds of active ads: which angle caused the sales?. Two warnings come with it: a competitor's ad library shows ads, not results, and your Meta creative test is not an A/B test.
3. Angle-matched funnel. A matched page or an advertorial changes who arrives, and the ads that open paths vanish from last-click: The five-minute story ad that never gets the last click and ad-matched landing pages: test the page, not the ad.
4. Retention. The flows, the texts and the community keep the subscribers the offer paid for. Read survival to the second charge by first-order channel, and test a retention tactic by choosing at random who gets it: does your customer community cause retention?

The ad platform optimises to the first purchase event and credits itself, and every other tool in the stack does the same for its own pillar.

Find out from data you already have

Add up the claims. Put each ad platform's reported revenue next to your store's revenue for the same weeks in the ad platform over-reporting checker. If the claims add up to more than you sold, no single report can say which pillar did the work.

Read the paths. In GA4, open Advertising, then Attribution paths, set 12 months or more, pick your purchase key event and download the CSV. The paths show which channels open and which close, and days to key event shows how long your buyers take (how long do your multi-channel buyers take?).

Build subscriber cohorts by first-order channel. Survival to the second charge, channel by channel, is the retention pillar's number, and payback per channel is the offer's.

Change one pillar at a time, in some regions. Try the new offer in half your regions, or switch the ads off in a few, and read the rest as the control group for at least as long as your buyers take to convert. The holdout test planner sets the length. Change the offer, the ads and the pages in the same week and no read can separate them.

Where a read fits

Later, a causal attribution read like Causality Engine's does the path step for you at the channel level. From the same Attribution paths export, it shows what each channel caused next to what last-click gave it, with Direct split back to the channels that sent those buyers. It tells you which channels carry the growth. Which pillar inside a channel does it still takes a test. Every channel gets a data-health score from 0 to 100 and a next step. It takes 1 to 2 minutes and costs EUR 99 once per upload, excluding VAT, with a full refund within 30 days, no questions asked. The first finding is free: your browser works out how many days buyers who saw two or more channels take to convert, and the file never leaves your machine.

Sources, accessed 28 September 2026: Noah Haynes's teardown of Everyday Dose (YouTube, 22 September 2026); Arie Scherson's teardown of Smooche (YouTube, 24 September 2026).