Case Study: 229 Applications from Targeting for the Business Forum on AI and Technology in Chișinău
April 28, 2025
30 min
Topic
Author at Level UP Academy
Yana Isaenko
A large 2-day business event – a forum on AI and technology transfer, held on November 21-22 at Arena Chișinău. The event format was both online and offline.
The goal was to attract interested participants through targeting. The traffic needed to be directed to the registration form on the website.
The goal was to attract interested participants through targeting. The traffic needed to be directed to the registration form on the website.
Context
This type of event was being held for the first time.
There was still low awareness among people about technologies like AI, how to use it in work, etc.
There were no well-known speakers.
The website was multilingual, as the audience was multilingual. The reasons why this became a challenge are explained below.
There was still low awareness among people about technologies like AI, how to use it in work, etc.
There were no well-known speakers.
The website was multilingual, as the audience was multilingual. The reasons why this became a challenge are explained below.
Challenges
The audience in Moldova actively uses Facebook and Instagram (as the campaign results showed), and we had a limited budget, so adding an additional platform like Google was problematic.
We started preparing for the launch 2 months before the event, and the campaigns were launched 38 days before it began. We developed a strategy that we followed throughout the campaign with minimal changes.
Facebook Ads Manager has a "smart" system. Depending on the selected objective, it can adjust its algorithms to focus on the desired action, rather than simply displaying an ad.
Adjusting means showing ads to those who are most likely to take the targeted action. In other words, each goal in Ads Manager is optimized for a specific action, and this was the basis of our strategy.
Initially, we launched ads with the goal of reach, to inform people about the upcoming event. At this stage, we didn’t segment the audience; we launched it broadly to ensure as many people as possible saw our ad. In Moldova, the audience is small, so even with a low budget, it’s possible to quickly reach a large portion of it, which became evident to us within a few days.
Next, we launched a campaign with the objective of "traffic - page views" to test interest sets for the subsequent lead generation campaign. We did not use broad audiences here, as the event was quite specific, and there was a risk that we would not be able to train the algorithm to find the right target audience. With this approach, we were able to test which audiences showed the most interest in the event, engaged with it, and explored our landing page. Additionally, this helped optimize the budget, as the "traffic" objective generally yields cheaper results, and the algorithm learns faster. The audience reach is also broader compared to lead generation. The more complex the objective and the customer journey to the target action, the more expensive the traffic is on Facebook.
We cannot disclose the specific interests that worked in this case, but even with the optimization goal of "page views" where the key action was viewing the site pages, we still saw registrations for the event.
The client also asked us to test hypotheses for attracting participants from neighboring Romania and Armenia (where the speakers were from) for the online broadcast, but this did not yield significant results. We realized that, for such events, except in rare cases, it is easier and more understandable for people to attend in person.
We also had one key challenge in setting up the advertising - the event website was in three languages, as people in Moldova speak two languages – Russian and Romanian, and the event was planned to be held in English with simultaneous translation. It turned out that the website had a registration form with three different "thank you" pages. At first glance, it seemed simple enough: set up three separate campaigns with the appropriate language and monitor the results.
However, the problem was that people who visited the site through Russian-language ads switched the language on the site and registered randomly in their preferred language. It was very difficult to track conversions correctly and train the algorithm this way.
It was impossible to resolve this technical issue under our conditions at that time, so we had to find a workaround.
In the end, we decided to run all campaigns only in Russian (since everyone in Moldova understands it), and prepared the designs and texts. Before launching the ads, a pixel was set up on the website, and special conversions were configured for each "thank you" page separately, in other analytics systems as well. This allowed us to track conversions from all languages and understand where they came from. To monitor registration metrics in the ad manager, we created a separate breakdown for all conversions.
After identifying the leading audiences based on interests, a campaign with the goal of Leads was launched. Within a week, the campaign learned and delivered decent results in terms of registrations. However, we noticed that the frequency of impressions and the cost per click were growing quickly, so the campaign required optimization. It was decided to slightly expand the audience by adding a few similar interests to the initial groups. This turned out to be the right decision. A duplicate of the previous campaign was created, necessary changes were made, and we were able to significantly improve the current results.
We started preparing for the launch 2 months before the event, and the campaigns were launched 38 days before it began. We developed a strategy that we followed throughout the campaign with minimal changes.
Facebook Ads Manager has a "smart" system. Depending on the selected objective, it can adjust its algorithms to focus on the desired action, rather than simply displaying an ad.
Adjusting means showing ads to those who are most likely to take the targeted action. In other words, each goal in Ads Manager is optimized for a specific action, and this was the basis of our strategy.
Initially, we launched ads with the goal of reach, to inform people about the upcoming event. At this stage, we didn’t segment the audience; we launched it broadly to ensure as many people as possible saw our ad. In Moldova, the audience is small, so even with a low budget, it’s possible to quickly reach a large portion of it, which became evident to us within a few days.
Next, we launched a campaign with the objective of "traffic - page views" to test interest sets for the subsequent lead generation campaign. We did not use broad audiences here, as the event was quite specific, and there was a risk that we would not be able to train the algorithm to find the right target audience. With this approach, we were able to test which audiences showed the most interest in the event, engaged with it, and explored our landing page. Additionally, this helped optimize the budget, as the "traffic" objective generally yields cheaper results, and the algorithm learns faster. The audience reach is also broader compared to lead generation. The more complex the objective and the customer journey to the target action, the more expensive the traffic is on Facebook.
We cannot disclose the specific interests that worked in this case, but even with the optimization goal of "page views" where the key action was viewing the site pages, we still saw registrations for the event.
The client also asked us to test hypotheses for attracting participants from neighboring Romania and Armenia (where the speakers were from) for the online broadcast, but this did not yield significant results. We realized that, for such events, except in rare cases, it is easier and more understandable for people to attend in person.
We also had one key challenge in setting up the advertising - the event website was in three languages, as people in Moldova speak two languages – Russian and Romanian, and the event was planned to be held in English with simultaneous translation. It turned out that the website had a registration form with three different "thank you" pages. At first glance, it seemed simple enough: set up three separate campaigns with the appropriate language and monitor the results.
However, the problem was that people who visited the site through Russian-language ads switched the language on the site and registered randomly in their preferred language. It was very difficult to track conversions correctly and train the algorithm this way.
It was impossible to resolve this technical issue under our conditions at that time, so we had to find a workaround.
In the end, we decided to run all campaigns only in Russian (since everyone in Moldova understands it), and prepared the designs and texts. Before launching the ads, a pixel was set up on the website, and special conversions were configured for each "thank you" page separately, in other analytics systems as well. This allowed us to track conversions from all languages and understand where they came from. To monitor registration metrics in the ad manager, we created a separate breakdown for all conversions.
After identifying the leading audiences based on interests, a campaign with the goal of Leads was launched. Within a week, the campaign learned and delivered decent results in terms of registrations. However, we noticed that the frequency of impressions and the cost per click were growing quickly, so the campaign required optimization. It was decided to slightly expand the audience by adding a few similar interests to the initial groups. This turned out to be the right decision. A duplicate of the previous campaign was created, necessary changes were made, and we were able to significantly improve the current results.
Why was targeted advertising chosen?
This is how we tracked registrations from all pages in different languages.
In addition to the pixel and other analytics systems, completed registrations were sent to a special bot linked to the registration form on the website. All links were tagged with special UTM parameters, so we could largely track where the traffic was coming from.
In addition to the pixel and other analytics systems, completed registrations were sent to a special bot linked to the registration form on the website. All links were tagged with special UTM parameters, so we could largely track where the traffic was coming from.
During the advertising campaign, we received 229 registrations from targeting across all pages we tracked.
The average cost per registration for the lead generation campaigns: 2.84 euros
(Before optimization, the cost per registration was 3.56 euros, after optimization - 1.94 euros)
Cost per registration across all expenses: 5.48 euros
Total spent: 1254.31 euros
Click-to-registration conversion rate: 17.68% before optimization
26.67% after the campaign optimization
Click-to-registration conversion rate across all expenses: 5.5%
The average cost per registration for the lead generation campaigns: 2.84 euros
(Before optimization, the cost per registration was 3.56 euros, after optimization - 1.94 euros)
Cost per registration across all expenses: 5.48 euros
Total spent: 1254.31 euros
Click-to-registration conversion rate: 17.68% before optimization
26.67% after the campaign optimization
Click-to-registration conversion rate across all expenses: 5.5%
Results
A separate part of the event's advertising was the paid masterclasses, and they came with their own challenges. We tested two versions of the advertising for them – one leading to the ticket aggregator's website and another leading to the event's website, where the masterclasses had a separate page with tickets.
The hypothesis with the aggregator website didn't work out. Apparently, people lacked sufficient information about the event there, and no purchases were made from that site.
On the forum's website, there were 7 masterclasses with different topics, speakers, and participation options, with payments going directly to the payment system, and a dedicated button for it. Given the number of variables involved, it wasn't possible to set up clear analytics, so the decision was made to direct traffic again to the target page views.
All the masterclasses were split into separate ad groups with their own creatives and texts, and the budget was set at the group level, with all groups having the same budget. Through ads on the website, we managed to attract an interested audience. Unfortunately, we don't have exact sales data (NDA), but it is known that all the spots for the masterclasses were sold. However, we cannot track which exact source of traffic the buyers came from.
The hypothesis with the aggregator website didn't work out. Apparently, people lacked sufficient information about the event there, and no purchases were made from that site.
On the forum's website, there were 7 masterclasses with different topics, speakers, and participation options, with payments going directly to the payment system, and a dedicated button for it. Given the number of variables involved, it wasn't possible to set up clear analytics, so the decision was made to direct traffic again to the target page views.
All the masterclasses were split into separate ad groups with their own creatives and texts, and the budget was set at the group level, with all groups having the same budget. Through ads on the website, we managed to attract an interested audience. Unfortunately, we don't have exact sales data (NDA), but it is known that all the spots for the masterclasses were sold. However, we cannot track which exact source of traffic the buyers came from.
*Photo of participants at one of the masterclasses
Nevertheless, on the first day, the event was attended by 420 people offline (they were recorded at the entrance) and over 150 people online. We can't track exactly how many people came from social media ads, but since this was the key source of traffic for the event, we can assume that a significant portion of the attendees came from the ads.
*The registration of the participants at the forum
*Forum visitors in the hall
Regarding the creatives and texts: over the course of 1.5 months, we tested different versions of ad layouts and texts, both standard ones with a description of the forum's activities and more targeted ones aimed at specific audience segments and their interests.
In the end, the targeted ads brought the most activity and conversions. The most attractive creatives for the audience were those featuring images of the arena where the event was to take place, as well as images of the audience.
In the end, the targeted ads brought the most activity and conversions. The most attractive creatives for the audience were those featuring images of the arena where the event was to take place, as well as images of the audience.
Author
&
Share
Оценка
Want to know more? Follow us on social media
Subscribe
Company
+373 7 66 12 222
Call us:
Email:
Quick Links
