Anonymous case · Pastry shop · E-commerce and shops

Over 26,000 online orders for a historic Milanese pastry shop

A historic pastry shop in Milan, with an artisan workshop, several shops in the city and an online store that ships to Italy and abroad. Its Google and Meta campaigns brought in over 26,000 online orders and more than 9,300 contacts. On Google, every euro invested brought back 13 in sales.

26,000+
Online orders from campaigns
9,300+
Contacts collected
13×
Return on Google campaigns
+23.1%
Easter 2026 vs 2025
Milanese panettone in neon graphics surrounded by glowing envelopes, case study of a historic Milanese pastry shop with over 26,000 online orders
Updated on22 November 2026
Reading time7 min
Matteo Coloru
Written byMatteo Coloru
This pastry shop makes 81% of its yearly online revenue in the Christmas period. So every campaign is there to get the most out of a few months and to prepare the ones that follow. In 2026 we rebuilt the Easter campaigns, found the purchases the platforms weren't seeing and built the emails that bring customers back for a second order.

The context

The brand is a historic Milanese pastry shop. It makes leavened cakes, tarts and festive sweets in an artisan workshop, sells them in its own shops in the city and ships them through the online store. The flagship product is the panettone.

The pastry shop has worked with the agency I work for since 2021. In 2026 the relationship restarted with a new contract, which ties part of the fee to three goals: online revenue, the people who reach the site and the share of those who buy.

People who buy here are looking for a gift that makes a good impression, not the supermarket panettone. The returning customer is worth a lot: one in three buys again within two years, and over time spends 1.82 times the value of the first order. Customers from abroad spend 40 to 54% more per order than Italian ones.

It all comes down to a few weeks

The Christmas period is worth 81% of the year's online revenue, and Easter is the second peak. Outside those windows advertising pays back little. If you reach Christmas without an audience that is already warm and without tracking that works, every mistake is paid for with a whole season.

When we picked the work back up in 2026, there were three knots to untangle:

  • the purchase wasn't reaching Google's and Meta's measurement tools as it should, so the platforms saw fewer sales than the real ones
  • many people looked at the products, but from there to the cart the step dropped tenfold
  • customers who had already bought weren't getting a journey designed to bring them back.

Easter 2026: the rebuilt campaigns

The Easter campaigns started on 9 March, on Meta and Google. Every week we looked at the numbers together. Midway through the campaign the client asked to raise the budget by 25%, and I put all of the extra on the ads that sold best, instead of spreading it across all of them.

In the third week the share of visitors who bought on the site hit 5.33%, a record. At the end of March, with Easter shipments still open, the campaign's revenue had already passed that of the whole 2025 Easter period.

Easter 2026 closed with +23.1% on the previous year.

Measuring when the platforms don't see everything

In May I ran a full check of the tracking, with a test order from the site through to the platforms. The purchase wasn't reaching Google Analytics, Meta and Google Ads as it should, and the add-to-cart event was missing too. That's why the platforms' numbers were lower than the real ones.

Until the tracking is fixed, a campaign isn't judged by the return the platform reports. You look at the ratio between all the online store's revenue and all the advertising investment, read from the order management system. It's a rougher calculation, but it doesn't lie.

The bottleneck was the product page

The clearest data came from the product page. Between those who looked at a sweet and those who put it in the cart there was a tenfold drop. Stronger advertising would have brought more people to the same page, and lost them at the same point.

That's why we decided to work on the page. The priorities became customer reviews, chosen from those rated four stars and above, and answers to the questions customers ask before buying. The client also approved a free-shipping threshold lowered by 30%, to remove a brake on the first purchase.

The emails that bring the customer back for a second order

In the contact management system (the CRM) there was a huge asset, barely used. In March two automatic sequences went live, the welcome for people who sign up and the reminder for people who abandon their cart.

In May I prepared the sequence for people who haven't bought in a while. It is five emails, and the product they put in the foreground changes with the time of year, instead of always staying the same. Then I also wrote the emails for the customer's birthday.

Christmas 2026: the season is prepared early

For Christmas 2026 we started with people, not with sales ads. At the beginning of November a campaign went out to collect a list of customers interested in the late-November promotion. In less than two days 243 contacts arrived, at €0.76 of investment for each.

Alongside Meta and Google, we also tried a new channel, ads inside ChatGPT, with a campaign on panettone. It's a small, measured test, like every new channel.

What this case teaches

1. In a seasonal business, the season is prepared out of season

If 81% of revenue arrives in two months, the work of the weak months serves those two months: contact lists, an audience that already knows the brand, emails ready to go. In November it is too late to build all this.

2. Before raising the budget, look at where people drop off

Here the weak point was the product page, not the ad. More traffic would only have multiplied the loss.

3. When tracking is broken, real revenue is what counts

The ratio of total revenue to total investment, read from the orders, is the compass until the platforms see every purchase.

4. The real profit comes from the customer who comes back

One in three buys again within two years. The emails that bring them back for a second order work all year round, even when the advertising is paused.

Frequently asked questions about marketing for pastry shops

How do you sell panettone online from an artisan pastry shop?

You start before Christmas. You build a list of interested people, set up tracking that records every purchase and fix the product pages. Then the Google and Meta campaigns bring the right people, and the automatic emails win back those who abandon their cart.

What should you do if the platforms don't record every purchase?

You fix the tracking with a test order, from the site through to the platforms. In the meantime you judge the advertising by the ratio of total revenue to total investment, read from the order management system, and not by the return the platform reports.

Why work on the product page before raising the budget?

Because advertising brings people to the page, but it's the page that makes them buy. In this case, between those who looked at a sweet and those who put it in the cart there was a tenfold drop. Reviews, answers to questions and a lower free-shipping threshold remove the brakes on buying.

Case study · Want a similar path?

Start the conversation

Do you run a seasonal business and want to reach the next peak with campaigns and emails already ready? Write to me on WhatsApp. I answer personally, within the day.

Message me on WhatsApp
Matteo Coloru available on WhatsApp