Luixifer.

Selected work

The brief, the plan,
and the numbers.

Screenshot ROAS is easy to manufacture. These cases are written the way we report: blended metrics, real constraints, and what we'd do differently.

Case 01 — Dropshipping · Handcrafted goods

"Atelier"

A dropshipping brand in handcrafted home & gift goods — name withheld by agreement. The mandate: take a working store and find out how far three channels could push it in four months.

Niche
Handcrafted home & gift goods — dropshipping DTC
Channels
Meta, Google, TikTok
Scope
Full-funnel paid media, creative direction
Engagement
4 months
MetaGoogleTikTokName withheld

The brief

Handmade products carry a story worth paying for — and dropshipping margins that punish careless media buying. The store worked; what it lacked was a system that could scale spend across more than one channel without the economics falling apart.

What we did

Meta carried discovery, with creative built around the maker's story — the process, the materials, the small imperfections that prove a thing is handmade. TikTok ran demonstration and unboxing angles. Google caught the demand the other two generated: brand, Shopping and remarketing. Creative refreshed weekly, and every landing page had to look handcrafted, never dropshipped.

Over four months the account went from single-channel spend to three channels running in parallel against one blended target, with recurring revenue layered in along the way.

What changed

€1.53Mcaptured sales in the final four weeks of the engagement — reconciled in the payment processor, not platform dashboards
4 mofrom kickoff to that run rate
3channels scaled in parallel: Meta, Google, TikTok
In hindsight

At peak volume, a meaningful share of gross sales was lost to failed and blocked payments — a cost nobody budgets for at kickoff. Payment recovery — smart retries, local payment methods, fraud rules — now goes into every scaling plan from week one, not after the first spike.

Case 02 — Coffee & subscriptions

Harbor & Vine

A specialty roaster with a loved product, a solid store — and ad spend stuck at $9k/month because every attempt to scale broke profitability.

Niche
Specialty coffee, subscription DTC
Channels
Meta, Google Search & Shopping
Scope
Paid media, creative, landing pages
Engagement
14 months, ongoing
MetaGoogle SearchShoppingLanding pages

The brief

Every scaling attempt looked the same: raise budgets, watch ROAS collapse, retreat. The account was optimizing for one-off bag sales — decent AOV, no repeat math — while the subscription, the product with real LTV, got no traffic.

What we did

We rebuilt the funnel subscription-first. New landing flow — how do you brew, how much, how often — new creative angles about never running out, and Meta optimized on subscription starts instead of purchases. Google took the brand and Shopping demand the Meta spend generated.

Because a subscriber is worth ~4× a one-off buyer over a year, the account could suddenly afford CPAs that used to look scary. That's the whole trick: we didn't lower the cost, we raised what a customer is worth.

What changed

$9k→$60kmonthly ad spend, scaled over 10 months
3.4blended MER at peak spend — reconciled to revenue, not platform ROAS
62%of new customers now start on subscription

"We stopped arguing about ROAS screenshots and started planning the whole business around one blended number."

Daniel R. — Founder, Harbor & Vine
In hindsight

We spent the first three weeks being polite about the old funnel before the data made the argument for us. Next time the subscription-first test goes live in week one.

Case 03 — Supplements

Ridgeline

A daily-greens brand whose acquisition machine quietly died after iOS privacy updates — and whose dashboards were too broken to show why.

Niche
Supplements — daily greens
Channels
Meta, TikTok, Google
Scope
Tracking rebuild, paid media, UGC creative
Engagement
9 months, ongoing
MetaTikTokGoogleCAPI · GA4UGC

The brief

Spend had been cut from $45k to $15k a month out of fear: platform ROAS said everything was unprofitable, finance said revenue hadn't fallen nearly that much, and nobody could say which number was lying.

What we did

Tracking first, opinions second. Server-side conversion APIs on every channel, GA4 rebuilt for e-commerce, and a post-purchase survey to catch what pixels miss. Three weeks in, the real picture: Meta was underreporting by roughly a third, and the "losing" campaigns were carrying the brand.

With numbers we could defend, we rebuilt the account on a blended-MER target, put UGC creators on a weekly testing cadence, and added TikTok once Meta stabilized — same angles, native execution.

What changed

−41%blended CPA versus the pre-audit baseline
3×ad spend, restored with finance's sign-off
11winning creatives found in the first 90 days of testing

"They told us to keep spend cut for three more weeks while they fixed tracking. That honesty is why they still run the account."

Priya S. — Head of Growth, Ridgeline
In hindsight

TikTok could have entered a month earlier — we were conservative while the new tracking bedded in, and the first TikTok winners showed the audience had been waiting.

Case 04 — Home & kitchen

Alder & Oak

A cookware brand with beautiful product photography, zero ad history, and a founder who wanted proof before committing real budget.

Niche
Home & kitchen — cookware
Channels
TikTok Shop, TikTok Ads, Meta
Scope
Launch strategy, paid media, creator program
Engagement
90-day launch sprint → retainer
TikTok ShopSpark AdsMetaCreator program

The brief

No pixel history, no email list, a $6k test budget — and a premium price point that needed demonstration, not display ads. Cold-start economics where most agencies burn the budget "building learnings."

What we did

We launched where demonstration is native: TikTok Shop, with a creator affiliate program doing the selling. Twenty micro-creators seeded with product, commission-only, our briefs. The three whose videos moved units got paid amplification through Spark Ads.

Meta entered in month two purely as retargeting and lookalikes built from TikTok buyers — cheap, warm, compounding. The hero offer was a bundle we designed around AOV: the flagship pan plus two accessories, priced to make shipping economics work.

What changed

$214krevenue in the first 90 days, from a standing start
$38blended CAC by day 90 against a $127 AOV
2creator videos past 1M organic views — owned rights, reused as ads

"I handed them a test budget expecting a report back. I got a sales channel back."

Marcus T. — Founder, Alder & Oak
In hindsight

We under-planned inventory for the first viral spike — the flagship pan sold out for nine days and we paused ads at peak demand. Depth-of-stock now sits in every launch plan we write.

Case 05 — Fashion

Mara the Label

An independent fashion label running on drops — small quantities, hard dates, and no second chance if a launch misses.

Niche
Fashion — independent label, drop model
Channels
Meta, Pinterest, email capture
Scope
Launch ads, UGC pipeline, waitlist strategy
Engagement
3 drop cycles
MetaPinterestWaitlist funnelUGC

The brief

Drops don't forgive slow optimization. With 16 SKUs in limited runs, the ads have to arrive at launch day with demand already built — there's no month of "learning phase" when the inventory sells out or doesn't.

What we did

Each drop became a three-phase campaign. Two weeks out: teaser creative and Pinterest — where fashion intent lives early — driving a waitlist, not a product page. Launch day: waitlist email plus Meta ads to warm audiences only, when scarcity is real and conversion is cheapest. Aftermath: UGC from buyers becomes the next drop's proof.

Between drops we ran no acquisition spend — deliberately. The label banks its budget for the two weeks it matters and pays nothing to exist in the feed the rest of the month.

What changed

3 / 3drops sold out — the slowest in nine days
5,800waitlist signups before drop three, at $0.42 each
48%of drop-three revenue from repeat buyers

"Launch day used to be the scariest day of the quarter. Now it's a checklist we run."

Mara V. — Founder, Mara the Label
In hindsight

Pinterest earned budget a drop later than it should have — it quietly produced the cheapest waitlist signups of the whole cycle. It opens every campaign now.

A note on the numbers

How these case studies are written

Every figure above is a blended metric reconciled against the client's store revenue — not a platform-reported ROAS cherry-picked from a good week. Time ranges and baselines are stated, clients approve every number before publication, and the "in hindsight" notes are real: we think an agency that can't name its own mistakes can't fix yours either.

Client names are shown with permission. Where a client prefers confidentiality, we'll describe the engagement without naming them — ask us on a call.

Your brand next

The next case study
could be yours.

Tell us your niche, your numbers and what's stuck. That's where every one of these started.

Get an audit →