Skip to content
ScaleFieldLab
Performance · Example engagement

Signal Rebuild

Server-side tracking and a TikTok and Meta creative-testing rebuild for a UAE fashion retailer ahead of White Friday.

Example engagement. An anonymised scenario showing how we approach this kind of problem. The numbers are illustrative targets, not a published client result.

Target outcomes (illustrative)

Blended ROAS
3.4xfrom 2.1x
Cost per acquisition
-32%
Meta event match quality
9.1/10from 5.8

The 30-second version

  1. The problem

    Meta, TikTok, GA4 and Shopify each reported a different sales number, and ROAS had slid for three quarters. Nobody trusted the dashboards enough to move budget.

  2. What we built

    Server-side tracking with the Meta and TikTok Conversions APIs, a daily reconciliation against Shopify orders, and a weekly Arabic and English creative-testing rhythm.

  3. What changed

    Platforms optimise on real orders before White Friday. In this illustration ROAS climbs from 2.1x to 3.4x and budget follows winning ads within days.

The problem

A UAE fashion e-commerce retailer selling across the Emirates was spending heavily on Meta and TikTok, yet the platforms, GA4 and the Shopify backend all reported different sales numbers. ROAS had drifted down for three quarters, and nobody trusted the dashboards enough to move budget with confidence. This is an example engagement that illustrates how we work; the client is anonymised and the figures are illustrative.

Fashion editorial portrait representing the retailer's seasonal collection

The goal

Rebuild the measurement foundation so ad platforms optimise on real orders, then restructure TikTok and Meta around creative testing in time for White Friday, the biggest sales window of the year. The target was a blended ROAS above 3x at a stable or lower cost per acquisition, measured against backend revenue rather than platform-reported numbers.

What we built

Tracking came first. Only once the signal was clean did we touch budgets or creative.

  • Server-side tracking: server-side Google Tag Manager, Meta Conversions API and TikTok Events API with deduplication, plus Google Consent Mode v2 and PDPL-aligned consent records.
  • Backend reconciliation: a daily check comparing platform-reported purchases with Shopify orders, so the gap stayed visible.
  • Account restructure: fewer, broader campaigns on Meta and TikTok, giving the algorithms enough conversion volume to learn.
  • Creative matrix: weekly tests of hooks, UGC-style try-ons and Arabic and English variants, with fatigued ads retired on a fixed rule.
  • Seasonal plan: a White Friday budget ramp that protected the learning phase instead of resetting it.
Abstract city lights suggesting live data flowing between platforms
Abstract lines of light representing conversion signals

What changed

In a scenario like this, clean signal and disciplined testing compound quickly. Over twelve weeks, blended ROAS in this illustration rose from 2.1x to 3.4x, cost per acquisition fell 32% and Meta event match quality climbed from 5.8 to 9.1 out of 10. More importantly, the team could finally see which campaigns made money, so budget moved to winners within days rather than at month end.

Before and after

The numbers we would hold ourselves to

Illustrative baselines and targets for an engagement like this, measured against the client’s own backend data.

  • Blended ROAS

    +62%
    Before
    2.1x
    Target
    3.4x
  • Cost per acquisition (AED)

    −32%
    Before
    142
    Target
    97
  • Purchases the platforms can see

    +54%
    Before
    61%
    Target
    94%

Channel mix

Where the effort went

Share of paid social budget, White Friday week

  • Meta62%
  • TikTok38%

12 weeks

How it would unfold

  1. 1Weeks 1–2Tracking audit, server-side GTM and Conversions APIs live
  2. 2Weeks 3–4Account restructure and creative testing framework
  3. 3Weeks 5–8Weekly creative sprints; budget shifted to winning angles
  4. 4White FridayScaled spend on proven creative with daily pacing

Lessons

What we would watch for

  1. 01

    Fix the signal before touching budgets. Every week spent optimising on bad data has to be unlearned later.

  2. 02

    Arabic hooks deserve their own test cell. Translating the English winner is rarely the Arabic winner.

  3. 03

    Start the seasonal ramp two weeks earlier than feels necessary, so the learning phase is over before peak demand.

Recognise the problem?

What we’d do for you

Every fashion business has its own numbers, but the first moves rarely change. Here is how it would start.

  1. Step 1A 30-minute callYou bring the number that matters. We tell you plainly whether we can move it, and if we are the wrong team, who is.
  2. Step 2Growth & Tracking AuditFixed fee, AED 7,500, ten working days. We check tracking, accounts and funnel, then hand you a ranked plan you keep either way.
  3. Step 3First experiment in 14 daysClean data first, then one test with a success metric and a deadline. Weekly numbers from then on.

Related work

All work
Growth SystemIn progressLive nowResults published when they’re real

Luxury Retail · Ongoing · UAE, GCC

The Watch Scanner

The growth system for a Dubai pre-loved luxury watch dealer, from first WhatsApp message to settled sale.

  • Growth Marketing
  • Performance Marketing
  • Digital Marketing
  • SEO & GEO
  • Marketing Technology
  • AI Automation
Growth SystemIn progressLive nowResults published when they’re real

Fashion / E-commerce · Ongoing · UAE, International

Caleums

The growth system for a Dubai fine jewellery brand, from positioning and measurement to search, paid social and bespoke orders.

  • Growth Marketing
  • Performance Marketing
  • Digital Marketing
  • SEO & GEO
  • Marketing Technology
  • Web Design & Development
Next case study · AI AutomationInstant Reply