Two accounts I ran, with the numbers I was measured on.

Work samples for the Senior Media Buyer role. One client is named with permission; the other is redacted under NDA and its figures are reconstructed from memory and rounded. I have marked exactly which is which, because a case study that hides its own precision is not worth reading.

Montreal ViagensGoogle, Meta, lifecycle · R$90–130K/mo · Nov 2025 – Apr 2026
US home goods retailer (redacted)Amazon, Google, TikTok · $60–100K+/mo · 24 months

Montreal Viagens

Brazilian travel agency selling high-consideration packages. Lead generation was expensive and the leads that did arrive were being wasted after the click.

The problem

Cost per lead sat at around R$22 and the account had been scaled by raising budgets rather than by fixing structure. Brand and non-brand traffic shared campaigns, so cheap branded clicks were flattering the average and hiding what prospecting actually cost. Downstream, leads landed in a shared inbox with no scoring and no guaranteed first response, so a meaningful share of paid media was buying contacts nobody called back.

What I did

Rebuilt account structure by destination and intent, separating brand from non-brand so prospecting cost became visible and could be optimised on its own terms.
Reworked audience segmentation and retargeting windows around the real consideration cycle for a trip, which is weeks, not days.
Restructured the CRM so every lead was scored and routed to a named person with a response deadline, instead of sitting in a queue.
Built WhatsApp automation for first-touch response, which is where this market is actually won, plus an email sequence for leads not ready to book.
Ran weekly pacing reviews against the CPL target and cut what did not clear it, rather than waiting for the month to close.

Results

Metric	Before	After	Change
Cost per lead	R$22	under R$10	−55%
Monthly media budget	R$90K	R$130K	+44%
Lead response	shared inbox	scored & routed	automated

Budget grew because the unit economics justified it, not the other way round. That is the sequence I look for on any account: fix the structure, prove the cost per outcome, then buy more of it.

Named with the client's permission. Budget figures are approximate monthly ranges.

US home goods retailer

A US-based furniture and home goods seller, managed while I was Senior Client Services Manager at Quartile in New York. Client name and absolute revenue withheld under NDA.

The problem

The account was running on Sponsored Products alone, harvesting demand that already existed and competing hard on a narrow set of high-intent keywords. Efficiency looked acceptable on ACoS, but ad-attributed sales had plateaued because nothing upstream was creating new demand. Every incremental dollar was going into the same auctions and getting more expensive.

The play

Kept Sponsored Products as the capture layer, with weekly search-term and ASIN harvesting, disciplined negatives, and separate campaigns for competitor conquesting so its economics stayed readable.
Added Sponsored Brands to hold category and branded search, which is where a growing brand starts losing sales to competitors bidding on its own name.
Used Sponsored Display and DSP for retargeting product viewers and for cross-selling across the catalogue, which matters in home goods because the second purchase is a different room, not a repeat of the same item.
Ran Google and TikTok above the funnel to build awareness and drive traffic, then watched branded search volume and Amazon organic rank as the proof that the top of funnel was working.
Scaled on TACoS rather than ACoS. ACoS only tells you how the ad performed; TACoS tells you what paid media is doing to total revenue, and it is the number that survives contact with the client's P&L.

Results

Metric	Start	After ~8 months	Change
TACoS	baseline	20% lower	−20%
Monthly ad spend	~US$60K	US$100K+	+70%
Channels live	1	6	+5

Spending 70% more while total advertising cost of sale fell 20% means the incremental spend was buying incremental revenue, not cannibalising organic. That is the only version of scale worth having. The client stayed on the account for two years.

Client anonymised under NDA. Figures are reconstructed from memory and rounded to the nearest meaningful step. I can walk through the campaign structure and the reasoning behind each decision in detail on a call.