Performance advertising for a perfume brand: over 70 Meta ad creatives, campaigns by season and fragrance, a product catalogue built from store data, server-side tracking, and weekly analysis based on revenue, not clicks.
You cannot smell perfume through a screen. The ad has to sell a fragrance the customer does not know at a price of 70 to 80 euros, in an account where two other contractors and the owner herself are working at the same time. And every creative has to show a price that checkout will later confirm.
The brand's offer on every image, fragrance pairs that the cart actually calculates this way, and a style that evolved with the data: from poster, to pair with a price, to a real photo with the genuine label. Underneath it all: a catalogue based on actual sales, server-side tracking, and weekly analysis based on revenue.
On the left is the style from September 2026: the bottle in the world of its notes, against a dark background, with the offer underneath. On the right is October: a real scene, a person and lighting, the two bottles on the table, and the headline in the empty upper corner. The change came from one sentence from the client, “they are too digital”, and from the data, which said the same thing: her own photo holding a bottle was the second-best-selling static creative in the account's history.
The AI model receives cut-out images of the real bottles as a reference and creates the scene with them. But it gets the label wrong every time: AKRARI, ANBARI, DLAY. That is why every label is measured on the image and replaced pixel by pixel with the genuine one from the store before the creative goes live. Nothing goes out until we have read every label at three times magnification.
Five steps that repeat. The client is involved at two points: when they approve the creatives and when they decide the budget.
A weekly report by campaign and creative: spend, revenue, seven-day click return, and store orders for the same week. The account log is read before every conclusion.
Which fragrances and which pair: based on twelve months of sales from the store database, stock availability, and what the client wants to promote.
Scene based on the brief, real bottles, the genuine label, the client's headline word for word, and the checkout price. A contact sheet with all variations.
The client reviews the creatives and says “yes” or tells us what to change. Nothing is uploaded without approval.
The ads are created through the API with names that are then visible in every order. The first honest review comes after seven days, not the next day.
The best-selling static creatives in the account's history have one thing in common: a number on the image. Our “four fragrances” and “buy one, get the second” offers follow the same pattern, with one additional rule: the pair and price shown in the creative must be the pair and price the cart will display. The cheaper bottle is half price, promotional products are excluded, and this applies both in the ad and in the store.
Akbari Perfume sells niche perfumes online and in five stores in Sofia and Plovdiv. Meta advertising is the main route to new customers, and we manage it in two stages: campaigns for new fragrances and seasonal occasions from October 2025 to February 2026 (VOYAGE, Sultan, Black Friday, Christmas, Layla and Majnoon, Valentine's Day), followed by monthly management from September 2026 featuring the brand's offer, “Buy one, get the second at -50%”.
There are over 70 creatives, shown in the gallery below in the order they were published. Their language changed with the data. At first, they were product posters: one bottle in the world of its notes, with a short line of copy. Then came the pairs with the price of both fragrances, because the offer on the image was what drove orders. From October 2026, the style has been “a photo, not a poster”: an AI model creates the entire scene using the real bottles as a reference—a person, lighting and setting—then the label is replaced with the genuine one from the store, because the model gets it wrong every time. The customer's headline remains word for word on every creative.
Each creative names a fragrance pair that the cart will actually calculate this way: both bottles are in stock, neither is promotional, and the price shown in the image is the checkout price. It may sound minor, but it is the difference between an ad that sells and an ad that leads to a dispute in the cart.
Behind the creatives is infrastructure without which the numbers mean nothing. The Meta product catalogue was rebuilt from the store data, using the thirty best-selling fragrances based on actual orders over twelve months. Meta Pixel and Conversions API were rewritten so that every purchase reaches Meta from both the browser and the server, with deduplication. Every store order records the campaign and ad it came from.
We review the account every week based on revenue and seven-day click attribution, never by number of purchases, because the average order ranges from 35 to 90 euros depending on the fragrance. Before drawing any conclusion, we also read the account log: two other contractors and the owner herself write in it, and a pause or budget change often belongs to someone else. In September 2026, the -50% campaign returned 6.8 times its spend according to Meta data, compared with 3.9 for the account as a whole.
Aesthetic centre in Ruse: three Facebook and Instagram posts a week, each about a real treatment offered by the centre, with an image generated from a brief, a custom text template and publishing that does not depend on someone being there that day.
Luxury brand for women's tights — visual identity and a bilingual online store with card payments and delivery by Econt.
We only use analytics and advertising cookies if you accept them. The site works without them too. Details ·
