Nezar Kamel

1,861,422 SAR

On SAR 444,104 of spend · 4.19 ROAS · 7,564 purchases

What was happening

I took over a perfume store that depended on a single ad platform. Its audience, its prices and the algorithm all shift every season, so what works in Ramadan does not work in summer. The task was to keep the account profitable for six straight years, not to win one month.

What did not add up

In the first two years I noticed that an account left on what it is used to erodes slowly. So at the start of every season I began asking three questions: what changed, what has stopped working, and where has the audience moved.

What I changed

• I set a fixed naming system for every campaign covering type, market and date. After six years and more than thirty campaigns, that is what makes the account readable at all.
• I reviewed performance by season rather than by month, because buying perfume is itself seasonal.
• I kept spend above breakeven in each individual period, not just on the overall average.

What happened next

Revenue

SAR 1,861,422

Spend

SAR 444,104

ROAS

4.19

Purchases

7,564

Clicks

344,557

Duration

Six years and seven months, continuous

Final twelve months

5.66 ROAS · 1,092 purchases

The evidence

What this means for a similar account

If your account is seasonal, the answer is not a new ad idea every month. It is a reading system that spots when the audience, price, or algorithm has changed.

Any account can give you one excellent month. Very few give you six years above breakeven. The clearest sign this was not luck is that the final twelve months returned 5.66 against 4.19 for the whole period, which means the account improved with time instead of eroding.

61:1

Best month: SAR 107,160 on SAR 3,040 · SAR 1.94 cost per sale

What was happening

A digital product with a thin margin and a small ticket size. Any waste in targeting eats the margin whole, and the account cannot carry one combined campaign that mixes categories together.

What did not add up

I tried the combined campaign first and it gave me a comfortable average. When I split it apart, one campaign inside was losing and another was returning several times over, and the average had been hiding both.

What I changed

• I split campaigns by product category instead of running one combined campaign.
• I ran Shopping and Search on separate objectives, because each is judged on a different measure.
• I tracked cost per sale daily, the only indicator that means anything at this margin.

What happened next

Best month

SAR 107,160 on SAR 3,040

ROAS, best month

61:1

Cost per sale

SAR 1.94

Largest month by volume

SAR 216,053 on SAR 4,981

ROAS, largest month

43:1

The evidence

What this means for a similar account

If your store sells many categories, the blended account average may be comforting while one part of the budget is leaking. Good management makes each category readable on its own.

The highest returning campaign in this account is the one that got banned. It hit 95:1 and was then stopped because it was bidding on a registered trademark. What I took from it is that a number far outside everything around it needs checking before you build on it, because the cause may be a violation rather than skill.

695,456 SAR

On SAR 105,849 of spend · plus SAR 209,640 from organic search at zero spend

What was happening

I took over a store with no measurement at all: no analytics, no Search Console, no Tag Manager, no Merchant Center, not even an ads account. The Snapchat account was a personal profile rather than a brand account, so it was paying several times the correct rate.

What did not add up

The category names themselves were losing demand. People do not search for “indoor lighting section”, they search for “chandeliers” 8,100 times a month, “bulbs” 5,400 times and “spotlights” 4,400 times. The store sold all of it and named none of it that way.

What I changed

• I built the measurement stack from scratch: Analytics · Search Console · Tag Manager · Merchant Center · ads account.
• I renamed the categories in search language and built a tagging system across the whole catalogue, which took a month.
• I upgraded the Snapchat account to a brand account, which opened the catalogue and halved the cost per swipe-up.
• I wrote marketing copy for every product and every category.

What happened next

Paid revenue

SAR 695,456

Spend

SAR 105,849

ROAS

6.6:1

Organic search revenue

SAR 209,640 at zero spend

Category impression share

30% nationally

The evidence

What this means for a similar account

If your store has good products but sales do not reflect that, the issue may not be budget. It may be tracking, taxonomy, or how the product appears before the ad ever runs.

The biggest gain on this project did not come from advertising. SAR 209,640 came from organic search after one decision: naming categories with the words people actually search, rather than the words used inside the company. It cost nothing, and it kept returning after the ads were switched off.

7.6:1

SAR 306,430 on SAR 40,109 on Google · 782 orders

What was happening

An account that had been stable for a year, then January sales collapsed within days. Nothing had changed in the ads, the targeting, or what competitors were doing.

What did not add up

I looked at the campaign first and found nothing, so I moved to measurement. A tracking line had been deleted from the store, the conversion signal had stopped arriving, and the campaign was still optimising toward an objective it could no longer see.

What I changed

• I checked measurement before blaming the campaign, and the deleted tracking line surfaced.
• I reinstalled it, then fired the event myself and confirmed it arrived before relying on it.
• I paused the Snapchat campaign and moved its budget to Google, where the return was higher.
• I stopped a Twitter test at SAR 199 after 43,019 clicks produced no sale.

What happened next

Google revenue

SAR 306,430

Google spend

SAR 40,109

Google ROAS

7.6:1

Orders

782

Total across four platforms

SAR 410,204 on SAR 98,725 · 4.2:1

The evidence

What this means for a similar account

If sales suddenly fall, the riskiest move is to react only inside the ad platform. Sometimes the campaign is fine and the signal it learns from is broken.

When numbers collapse suddenly, check measurement before you blame the campaign. Most people I have seen in this position either raise the budget or rebuild the campaign, and both deepen the loss because the fault is not there.

336 sales

On SAR 9,694 in eight weeks · against 313 sales on SAR 31,550 in the full year before

What was happening

The account had spent SAR 31,550 over a full year and produced 313 sales. Then it came to me. Same store, same products, same market. The only variable was who was managing it.

What did not add up

One indicator explained the whole account: 85,901 clicks against 313 sales. When clicks are plentiful and sales are scarce, the problem is not how many people arrive. It is who arrives.

What I changed

• I stopped a campaign type Google had already discontinued support for.
• I removed bidding on keywords unrelated to the products in the store.
• I rebuilt the campaigns around the store’s actual categories.
• I lowered daily spend rather than raising it, because waste is not fixed with more of it.

What happened next

Previous management

313 sales on SAR 31,550 in a year

After the ad account moved to me

336 sales on SAR 9,694 in eight weeks

Revenue

SAR 205,370 on SAR 18,940

The telling indicator

85,901 clicks against 313 sales

The evidence

What this means for a similar account

If clicks are high and sales are low, raising the budget is not the first move. The first question is whether the right people are reaching the store at all.

Had the owner doubled the budget that year, he would have doubled the loss. The account was spending twice what I spent, using a discontinued campaign type, and bidding on keywords unrelated to its products. The budget was never the problem.

14.3:1

SAR 126,500 in memberships on SAR 8,850 · 110 sign-ups in 18 days

What was happening

Three years of campaigns on a Messages or WhatsApp objective. The reports produced conversation counts and never a revenue figure, so the ads’ effect on the club’s income stayed unknown that entire time.

What did not add up

The campaign objective decides what you will see in the dashboard a month later. As long as the objective is messages, not a single sale will appear no matter how much you actually sold.

What I changed

• I changed the objective to store visits with membership value passed through, and revenue appeared for the first time.
• I separated the men’s and women’s branch campaigns from day one.
• I collected everyone who engaged with the ads and served them again.
• I built a two-step funnel instead of a single campaign.

What happened next

Memberships

SAR 126,500

Spend

SAR 8,850

ROAS

14.3:1

Sign-ups

110 in 18 days

Documented period

Twenty monthly reports

The evidence

What this means for a similar account

If your business sells through WhatsApp or a branch visit, knowing how many messages arrived is not enough. You need to know which campaign moved revenue.

Three years on the wrong objective hid SAR 126,500 that was actually being earned. I did not add a riyal to the budget. I changed the campaign objective.

0.55 $

2,217 WhatsApp conversations on $1,216 via YouTube · best month $0.42

What was happening

Marble is not bought in a click. It is a project decision: the buyer needs to see the material, compare, and speak to someone before committing. So the only sensible outcome is a qualified conversation, not a purchase.

What did not add up

The product is visual by nature and video explains it better than any still image. I also noticed YouTube in this sector was far less crowded with ads than Instagram, and that the people using it were using it for awareness rather than lead generation.

What I changed

• I ran a YouTube video campaign on a conversion objective explicitly, not a views objective.
• I held the budget flat at $10 a day for five straight months.
• I ran a parallel Instagram campaign on the same objective over the same period, so I could compare the two side by side.

What happened next

YouTube conversations

2,217

YouTube spend

$1,216

Cost per conversation

$0.55

Instagram at comparable spend

469 conversations at $2.48 each

The gap

4.7× the conversations for the same spend

What this means for a similar account

If every competitor is in the same channel, that channel may be the most expensive one rather than the best one. The decision starts from comparing the same objective across channels.

The default channel for a sector is not always the right one, because everyone bids on the same audience there and the price rises. Instagram did not fail here either: it produced 982 new followers, an asset that compounds and that YouTube does not produce. The real mistake would be judging both on one measure.

Idea and script

Perfume sells on impression, not on specification, so an ad that lists notes does not work. The script was built on a single moment: a hand setting the bottle down in a quiet room, and the rest is light and silence.

Building the scenes

Scenes were built with paid generation tools, with composition, angle and lighting decided per shot before anything was generated. Twelve seconds came out of far more footage, and what explained the moment was cut.

Sound

No voice-over. Music only, because a human voice here narrates what the image already says.

Edit and finish

Shots cut to the music, with a single closing title. No text over the scene: the ad shows rather than explains.

Facts

Length

12 seconds

Platform

Meta · Instagram and Facebook

Campaign objective

Purchases

Published

7 August 2025

Outcome

Distributed as paid media against a purchase objective, not views. The full case and its declared data limits are below.

81,029 SAR

On roughly SAR 14,630 · 5.5 ROAS · 402 purchases at SAR 36.39 each

What was happening

Perfume sells on impression rather than specification. A cinematic shoot for a single fragrance costs thousands and takes days, and then it only gets tested once. If it does not work, you wait for another shoot.

What did not add up

The obstacle is not the creative idea. It is the cost of trying it. Anyone who does not own production waits on a third party for the material, and loses the fastest tool they have: swapping the ad based on what the data says this week rather than next month.

What I changed

• I produced a 12-second cinematic video with AI, built on impression rather than a feature list.
• I launched a Meta campaign on a purchase objective, not views and not engagement.
• I distributed the video as paid media instead of posting it and waiting.
• I repeated the experiment with a second piece a week later, to confirm the result was not a fluke.

What happened next

Revenue

SAR 81,029

ROAS

5.5:1

Purchases

402

Cost per purchase

SAR 36.39

Spend

Roughly SAR 14,630

Video views

335,395

Click-to-view rate

1.94%

What this means for a similar account

If your account waits on creative from one side and media buying from another, you lose testing speed. Here creative, buying, and reading results sit in one loop.

The video that sells is not the most beautiful one. It is the one I can swap on Tuesday if it did not work on Monday. The gain here is that the produce, launch, read, adjust cycle shrank from weeks to days, because all four links sit under one manager and none of them waits on another.

Idea and script

The body needed to introduce its sites to visitors who did not know them. The script made the visitor the narrator: he moves between sites reading reviews from people who went before him, so trust is built by the public rather than by the institution.

Building the scenes

The Qatari locations were built with paid generation tools in late-afternoon light, and the reviews were composited to read inside the frame rather than on top of it.

Sound

Arabic voice-over in a Gulf dialect, timed to the movement between locations.

Edit and finish

Thirty-nine seconds across several locations, with music carrying each transition instead of a hard cut.

Facts

Length

39 seconds

Platform

Instagram

Account

The body’s own verified account

Published

17 April 2026

Outcome

Published on the body’s own account. The value is that a verified government body approved fully generated production, not the view count.

Idea and script

A shemagh is bought out of belonging, not specification. The script put it where it actually lives: a majlis, men of different generations, so the product reads as part of a scene rather than an item on display.

Building the scenes

The majlis and its characters were built with paid generation tools, with close attention to cultural detail: the cut of the shemagh, how it is folded, how a man sits. This was the hardest part of the work, because one error there costs the ad its credibility with the audience it is for.

Sound

Arabic voice-over in a Kuwaiti dialect, because the market is Kuwaiti and the difference is audible.

Edit and finish

Cut on the exchange between the characters, at a pace that suits a majlis rather than a fast retail ad.

Facts

Length

24 seconds

Platform

Instagram

Market

Kuwait

Published

17 March 2026

Outcome

145,454 views, of which 3,618 were organic on Instagram. The gap between the two means distribution went well past organic reach, which is to say the piece was run as paid media rather than posted and waited on.

Idea and script

A smartwatch in a market where everyone lists the same specifications. So the script dropped them and played on where the watch sits instead of what it does: a near-future world where it is simply part of the scene.

Building the scenes

Showroom and street were built with paid generation tools on one purple palette held across every scene, so the piece stays recognisable from a single frame.

Sound

Arabic voice-over, written to the full fifty-six seconds rather than trimmed into them.

Edit and finish

The longest piece in the set. Holding attention past the fifteen-second mark is the real problem, and it is solved by changing location rather than by cutting faster.

Facts

Length

56 seconds

Platform

Instagram

Sector

Watches

Published

20 April 2026

Outcome

125,557 views, of which 9,028 organic. It is evidence that long-form works when there is budget behind it and it is built to be watched rather than scrolled past.

Idea and script

The brand is called Nagham, which means melody, so the script took the name literally: the perfume is presented as a piece of music, and the conductor is a lion. The idea is deliberately strange, because a small perfume store with no budget does not get watched if it makes a conventional ad.

Building the scenes

A classical opera house and a composite character built with paid generation tools. Holding that character consistent across scenes is the hard part, and it was done with fixed references for every shot.

Sound

No voice-over. The music is the subject of the ad, so anything spoken over it contradicts the idea.

Edit and finish

Thirty-six seconds built on the swell of the piece, cutting on the beat rather than on the sentence.

Facts

Length

36 seconds

Platform

Instagram

Ad spend

Zero

Published

9 March 2026

Outcome

44,828 views, all of them organic. It is the weakest number in this row and the strongest signal in it: everything else here had budget behind it, and this had none.