Free Growth Diagnostic·$1,500 Value
All Case Studies
Established Ecommerce Brand
Meta Ads · Account Rebuild

Ecommerce Brand. Meta funnel structure rebuilt. 5.87× blended ROAS on $316K.

5.87×
Blended Meta ROAS · Jan–May 2026
~3× Prior Blended ROAS
5 Months Documented Window
First 30 Days Structure Rebuild
Jan 1 – May 31, 2026 · Meta Ads Manager · MESAscale
5.87× ROAS
$1.85M Platform-Attributed Revenue $316K Managed Spend
Attribution: Meta Ads Manager · 7-day click / 1-day view window
Meta Ads Manager dashboard showing 5.87× blended ROAS on $316K managed spend, Jan–May 2026. Brand identity redacted per NDA.

Meta Ads · Jan 1 – May 31, 2026 · 5.87× blended ROAS · $316K managed spendBrand identity withheld per client NDA. Campaign names redacted.

When we opened the account, the brief had been to test more creative. The account had no audience exclusions, outdated lookalike seeds, and prospecting and retargeting running into each other. Adding new ads into that structure wouldn't have changed the underlying problem.

This brand was doing roughly 3× ROAS before we took over. Every performance dip had been answered with more creative testing, and the account had never had a structural review. The pattern was clear: prospecting and retargeting overlapping with no exclusions, lookalikes seeded from visitors instead of purchasers, ads running for more than six months, and years of purchase history never connected to Meta. The opportunity wasn't creative. It was a cleaner acquisition system — stronger signals, tighter funnel boundaries, and a testing cadence that let the team make budget decisions from something other than one blended number.

What was actually happening in the account

01

Prospecting and retargeting were competing for the same people

No exclusions separated prospecting from retargeting. The same person could see a cold ad and a retargeting ad in the same week, with no way to separate which drove the sale. More spend would have amplified the overlap, not the results.

02

Lookalikes were modeled on visitors, not customers

Lookalikes were seeded from 30-day website visitors, not purchasers. The seed list captured everyone who landed on the site, including people who bounced immediately, not just people who completed a purchase.

03

Creative was being replaced only after performance declined

Some ads had been running for six-plus months with no rotation. No testing cadence, no threshold for promoting a winner, no exit criteria. New creative only entered the account after performance had already softened.

04

Years of purchase data were sitting unused

The brand's purchase history had never been connected to Meta. At their prior spend level, that was a significant gap. The account was learning from website activity while verified purchaser data sat unused.

Free Growth Diagnostic  ·  No Pitch  ·  Yours to Keep

If you recognized your account in those findings, a diagnostic session surfaces the same structural gaps and gives you a prioritized plan to address them.

Apply for Free →

The first 30 days

Built clearer funnel boundaries

Prospecting, consideration, and conversion were split into separate campaign groups with audience exclusions at each boundary. Each stage got a distinct role, budget, and baseline so we could see where efficiency was shifting and where more spend made sense.

Rebuilt prospecting around high-value purchasers

We rebuilt the audience using the brand's highest-value customers as the source signal for prospecting, keeping the same campaign objective but starting from a much cleaner signal.

Connected purchase data to Meta across all stages

The brand's purchase history was activated across the account: to seed prospecting signals, suppress existing customers where appropriate, and support re-engagement. Years of unused data became part of the acquisition system.

Introduced a weekly creative testing cadence

A repeatable weekly cycle: new creative enters testing, winners get more budget, underperformers are retired against defined criteria. We refresh creative before fatigue shows up, not after performance has already dropped.

Aligned measurement to each funnel stage

Prospecting and conversion were measured against their respective roles in the funnel. Budget decisions no longer depended on a single blended account number.

The account was running with no purchaser data connected and prospecting and retargeting overlapping. The signal feeding the algorithm was mixed. Separating the funnel stages and uploading verified purchase history gave the platform a more specific reference point. That's what we fixed first.

Funnel Architecture

Funnel architecture: three stages with audience exclusions at each boundary

Stage Audience Exclusions Applied Objective
TOF Prospecting
Advantage+ seeded from high-LTV purchasers Top-spending customer cohort via Customer Match, not all site visitors
Existing purchasers suppressed. Consideration audiences excluded from prospecting budget.
New-customer acquisition
MOF Consideration
14-day product-intent site visitors Engaged with product pages but did not initiate checkout
Purchasers excluded. 7-day cart abandoners reserved exclusively for BOF.
Re-engage active browsers
BOF Conversion
7-day cart abandoners Dynamic product ads with urgency messaging at this stage only
Recent purchasers excluded on a 180-day rolling window.
Convert high-intent visitors

Each stage carries its own ROAS target. Prospecting and conversion are reported separately; no single blended number drives budget decisions.

Five months later

What changed in the rebuild
Prospecting Separated from retargeting into its own campaign with a distinct ROAS target calibrated to acquisition cost, not blended efficiency
Customer signals Purchaser list rebuilt and segmented by recency. Purchasers excluded from prospecting and used to anchor lookalike audience quality
Creative testing Concept-level testing isolated within defined budget limits before scaling. Winning creative rotated with controlled introduction of new concepts
Measurement Prospecting and conversion performance evaluated separately against funnel stage, not collapsed into one blended account ROAS

The metrics are in the screenshot above. The more useful operational change was clarity. When the client wanted to increase spend, the decision no longer came down to one blended number. Prospecting and conversion could be evaluated separately, and the team could identify which stage was actually ready for more budget.

Phil Chan Head of Growth MESAscale
View on LinkedIn
About the strategist

Phil has spent more than a decade in paid media across agency, enterprise, and ecommerce. Most of that time was in senior account leadership, directing multi-channel portfolios at seven-figure monthly scale. That work required thinking past individual campaigns to how the full acquisition system operates.

Through MESAscale, he applies that same lens to established ecommerce brands on Google, Meta, and YouTube. The focus: account architecture. Where spend is leaking, where cross-channel coordination is breaking down, and which structural changes compound over time.

10+ Years in Paid Media Enterprise + Ecommerce Google · Meta · YouTube
Free Ecommerce Growth Diagnostic

Find out whether your Meta funnel is separating audiences or blending them.