Skip to content
Kuraib Ali
Selected work

UAE skincare brand (confidential)

Reasonable CPM. Reasonable CTR. Return on ad spend stuck below 2× for eleven months.

Sector
E-commerce / beauty · UAE
Timeframe
42 days
Engagement
Tracking audit, event correction, CAPI — no campaign changes

The situation

A UAE e-commerce brand had run Meta Ads for eleven months with a campaign structure that looked fine from the surface.

Return on ad spend had never passed 2.1×, and the brand had accepted that as the ceiling — competition, margins. In the first thirty minutes of the audit the cause was on the checkout confirmation page: the pixel fired AddToCart instead of Purchase. For eleven months the algorithm had been told that adding to cart was a sale, and had optimised, very effectively, for people who add to cart and never check out.

The move

  1. 01

    I corrected the event in GTM — AddToCart removed from the confirmation page, Purchase added with value and currency — and verified it in Events Manager test events before it went live.

  2. 02

    I implemented server-side CAPI through a GTM server container, taking signal quality from 7.2 to 9.0, and rebuilt the purchase retargeting audience from verified purchase events instead of a pool full of non-buyers.

  3. 03

    I changed nothing else for 28 days. The point was to isolate the signal fix as the only variable, so the result could be attributed to it and not to a structure change made at the same time.

What changed

States of the system, not metrics — these are things that either were or were not true, so they can be stated without claiming a result.

Before and after states for the UAE skincare brand (confidential) engagement.
BeforeAfter
Purchase signalAddToCart fired on the confirmation pagePurchase event with value and currency
TrackingBrowser-side only, ~35% signal lost to iOS and blockersServer-side CAPI, 9.0/10 signal quality
Retargeting audienceBuilt on AddToCart — non-buyersBuilt on verified purchases
Optimisation targetPeople likely to add to cartPeople likely to buy

Scroll sideways to compare →

What shipped

  • Purchase event corrected in GTM, with value and currency
  • Verification in Meta Events Manager before deployment
  • Server-side CAPI via GTM server container
  • Purchase retargeting audience rebuilt from real purchases
  • Learning-phase discipline: no budget or structure changes for 28 days

The scorecard

What this engagement was held to, and what the platform reported. Each figure names the source it was pulled from.

Return on ad spend
1.8× → 5.8× (+222%)
Meta Ads Manager purchase-event reporting, 42 days from the fix
Signal quality
7.2 → 9.0 / 10
Meta Events Manager event match quality score
Revenue per campaign
3.2×
Platform-reported purchase value, same budget
Cost per acquisition
−34%
Spend ÷ platform-reported purchases

✓ Every figure is sourced from platform reporting — Meta Ads Manager, Meta Events Manager, Google Ads or the client’s CRM — never modelled or estimated. The full write-up, as published, follows below.

Full write-upMeta Ads E-Commerce ROAS UAE: 1.8x → 5.8x After Pixel Fix | Case Study
Quick Answer

A stalled ROAS often traces to one misconfigured tracking event rather than the campaign itself, this case shows an 11-month-old Meta Ads account where a single wrongly-tagged event was capping ROAS at 2.1x, and correcting it alone moved ROAS to 5.8x in 42 days.

01. The Situation

Reasonable CPM. Reasonable CTR. ROAS stuck below 2x.

A UAE-based e-commerce brand had been running Meta Ads for 11 months. Campaign structure looked reasonable from the surface: prospecting, retargeting, decent creatives. But ROAS had never exceeded 2.1x. The brand had attributed this to competitive pressure and product margin constraints, and had accepted it as the ceiling.

The issue was identified in the first 30 minutes of the audit. The Meta pixel on the product purchase page was firing an AddToCart event instead of a Purchase event. For 11 months, the algorithm had been told that adding a product to cart was a purchase, and it had been optimising, very effectively, for cart additions. People who were likely to add items to cart but never check out.

02. What Was Found

One wrong event. Eleven months of misdirected optimisation.

  1. 01

    Purchase event on checkout confirmation page firing as AddToCart, incorrect event parameter in GTM trigger configuration

  2. 02

    Signal quality 7.2/10 in Meta Events Manager, acceptable rating masking the wrong event type being optimised

  3. 03

    Custom audience for purchase retargeting built on AddToCart events, meaning the retargeting pool was full of non-buyers

  4. 04

    Server-side CAPI absent: all tracking browser-side only, with an estimated 35% signal loss from iOS and ad blockers

03. What Was Done

Fix the event. Rebuild the audiences. Implement CAPI.

The purchase event was corrected in GTM: AddToCart trigger removed from the checkout confirmation page, Purchase event added with correct parameters including value and currency. Verified in Meta Events Manager test events before going live.

Server-side CAPI implemented via GTM Server Container. Signal quality brought from 7.2 to 9.0. Purchase retargeting audience rebuilt from scratch using verified purchase events. The algorithm was given 28 days to exit the learning phase on clean signal before any budget changes were made.

No campaign structure changes were made initially. The objective was to isolate the signal fix as the variable, not to change structure and signal simultaneously and attribute results ambiguously.

Days 1–3
Event correction
Purchase event fixed in GTM, verified in Events Manager, CAPI implemented
Days 4–28
Learning phase
Algorithm recalibrating on clean purchase signal, no budget changes made
Days 29–42
ROAS acceleration
1.8x → 5.8x ROAS confirmed, no structural changes, isolated signal fix
04. The Result

1.8x to 5.8x. One broken event tag for 11 months.

ROAS moved from 1.8x to 5.8x over 42 days from the date the purchase event was corrected. The algorithm exited the learning phase after 18 days and began delivering to audiences with genuine purchase propensity for the first time. Revenue per campaign increased 3.2x with no budget change.

5.8xFinal ROAS
+222%ROAS increase
9.0/10Signal quality
42 daysTime to result
✓ ROAS figures sourced from Meta Ads Manager purchase event reporting. Purchase event corrected and verified via Meta Events Manager test events before live deployment.
← Real Estate CPL DubaiHealthcare Bookings →KSA Investor Leads →Google Ads Clinic →All Case Studies →
What This Demonstrates

Wrong event type can misdirect an algorithm for months, and the data will not tell you

A ROAS ceiling of 1.8x for 11 months was not a creative problem or a budget problem. It was a measurement problem. Meta Events Manager showed signal quality of 7.2/10, sufficient to mask the underlying issue. The platform was reporting performance accurately against the event it had been given. The event itself was wrong. This case demonstrates why every engagement starts with a tracking audit before any campaign decision is made.

See Meta & Google Ads Service →
Frequently Asked

How can one tracking error affect ROAS for almost a year?

The Purchase event was firing as AddToCart on the checkout confirmation page, a parameter error that had gone undetected for 11 months. Meta’s algorithm was optimising toward add-to-cart behaviour, not actual purchases, so campaigns kept reaching browsers rather than buyers regardless of budget or creative changes.

Was anything changed in the ad creative or targeting?

No creative or audience changes were made before the event was corrected. The fix sequence was: correct the Purchase event and its parameters in GTM, verify it in Meta Events Manager, migrate to CAPI, then let the algorithm relearn on accurate data.

How long after the fix did ROAS improve?

ROAS moved from 1.8x to 5.8x over 42 days from the date the Purchase event was corrected. The algorithm exited its learning phase after 18 days and began delivering to audiences with genuine purchase propensity from that point forward.

By Kuraib Ali, Search & Growth Consultant, Dubai · Meta Blueprint · Google Ads Certified  ·  Last updated: August 3, 2026

Next project

Dubai aesthetics clinic (confidential)

Good cost per lead. Poor bookings. The clinic thought it needed more leads.

Working on something with a similar shape?

admin@kuraib.site