Joey & Pooh

Joey & Pooh began this engagement at roughly 3X ROAS. The approved case record shows where the account ended four months later and which systems changed along the way: full-funnel campaign structure, creative iteration, attribution, email, WhatsApp, and ongoing Shopify updates for SEO and frontend performance. It does not support a neat story about one ad or one hidden audience, so I will not manufacture one.

Four months later, the reported result was 15X ROAS, monthly revenue was 217% higher, email's share of revenue had moved from 5% to 22%, the measured WhatsApp flow returned 24X, and CAC was 53% lower. Those figures are specific to Joey & Pooh and the measurement window behind the approved case study. They are evidence of the operating system used on that account, not a forecast for another brand.

What the approved record tells us

• The baseline was approximately 3X ROAS.
• The campaign was rebuilt around TOFU, MOFU, and BOFU roles.
• Attribution was overhauled so acquisition and retention could be read together.
• Email and WhatsApp became material revenue channels rather than disconnected add-ons.

Scope and timeline

The full-funnel rebuild

Campaign and measurement foundation

We rebuilt the marketing architecture from the ground up:

• A funnel structure across platforms (TOFU, MOFU, BOFU) with clear objectives per stage.
• Audience segmentation into actionable clusters by intent, recency, and product affinity, not persona theater.
• Enough creative throughput to test hooks and offers, volume for learning rather than vanity.
• Platform-specific execution for Meta, Google, and TikTok.

Shopify, SEO, and site-optimisation work

Marketing traffic was only useful if the store could preserve intent, so the site work ran alongside the media rather than after it. On the Shopify build we tightened the frontend and fixed the technical layer that decides how the store is read by search engines, social platforms, and shoppers:

• Structured data so products, breadcrumbs, and the organisation are described as machine-readable entities rather than left implicit in the markup.
• SEO metadata: page titles, descriptions, and canonical URLs set deliberately per template, so collections and products present correctly and do not compete with each other for the same query.
• Open Graph and Twitter share cards, so a link to a collection or product posts as a proper preview with the right image, title, and description instead of a bare URL. On a brand that grows through paid social and sharing, a link that previews correctly is part of the ad.

This was ongoing operational work, not a one-off redesign, and no unsupported page-speed or ranking number is attached to it. The point was to stop the store leaking the intent the campaigns paid for.

Creative velocity and buying structure

We increased the test cadence and tightened the catalog-to-creative mapping, so each major collection had a clear hook, a proof point, and a landing match. Platform automations supported rotation and pruning, always reviewed against conversion quality, not dashboard CTR alone. We do not publish tactic-level lift percentages here, because those numbers are not comparable across accounts without category, margin, and audience context.

Retention and messaging continuity

• Email: welcome, browse, cart, and post-purchase sequences aligned with the same offers surfaced in the ads.
• WhatsApp: transactional and promotional flows where opt-in and policy allowed, tied to purchase and browse behavior.
• Service layer: where used, chat tools scoped to deflect repetitive FAQs, not to invent product claims.

Results

Reported ROAS moved from roughly 3X to 15X over the four-month engagement window.

The figures below are the public, approved proof set. The underlying platform exports, margin inputs, and attribution settings are not published on this page, so the claims should be read within that limitation.

Verified shareable proof points

These are the figures we can discuss publicly because the case study is verified as shareable. They still belong to this brand's catalog, margin, and season, so they should not be read as a forecast for another account.

Why this worked (operator perspective)

The repeatable version looks like this:

• Split prospecting and remarketing with distinct creative objectives and naming.
• Tie budget to margin and stock, not only to last week's ROAS.
• Refresh creative when performance evidence shows fatigue rather than on a universal calendar.
• Connect email and WhatsApp to the same offer and product story as the paid media.
• Review weekly on contribution and payback, not a single platform screenshot.

Principles that generalize (without fake multiples)

• Omnichannel means one economic model, not the same creative pasted everywhere.
• AI is useful for iteration and research, not a substitute for offer clarity and site experience. It is the same boundary we hold in AI product photography.
• Community and UGC work when rights, quality, and testing are managed, not from a universal "3X" rule.

The funnel-by-stage discipline here is the same idea as treating personalization as four aligned layers.

What repeats, and what does not

The 15X does not transfer. What transfers is the discipline: make acquisition and retention legible in one measurement model, give each funnel stage a distinct job, and use the resulting evidence to decide what to fund next. Whether that produces an improvement, and how large it is, depends on the offer, margin, market, and execution.

For a disciplined read on your catalog, margin, and funnel, talk to our team, or see how we connect performance and creative in our services. For the same operator approach in a different category, read the SHANZAY case study.