How O2 increased the effectiveness of influencer marketing by 216%

When strategy is based on data

O2 is one of the largest brands on the Czech market in the field of mobile and internet services. It has been building marketing communication over the long term that goes far beyond the usual product claims and promos for mobile phones or modems – it focuses on relevance, authenticity and building relationships with customers. And influencer marketing is a great channel for this, which O2 began to focus intensively on in 2024. And it's not just that – investments in advocacy have increased by 300% in Europe over the last 4 years.

"Thanks to Carl, we know that branded CPM varies widely across creators. For some, it's a few hundred, for others, it's thousands."

Marketa Zvolska, Social Media Strategist O2

As a strongly data-driven brand, O2 wanted to manage its influencer campaigns based on real data, not impressions, from day one. The goal was not just to implement individual campaign flights and deliver short-term value, but to create a system that would work in the long term – across topics, formats and communication goals. That is why it started working with Carl for Social Business from the beginning – with a tool that allows you to manage and measure influencer marketing with maximum accuracy. Together, we built a clear data structure that allows you to monitor influencer performance in real time across campaigns, periods and networks, effectively select creators (or, conversely, not extend some collaborations), allocate budget appropriately and clearly show what value influencer marketing brings compared to other communication channels.

A well-thought-out foundation from Carl that delivers results

As a large telecommunications brand, O2 expects influencer marketing to primarily provide an uplift in upper-funnel metrics - i.e., increased awareness of the brand, products, and key topics, especially among younger target groups.

However, traditional methods such as brand tracking provide results with a delay and, in the case of influencer content, with limited depth. This makes them unsuitable for day-to-day management of channel effectiveness. O2 was therefore looking for a way to manage this dynamic channel faster, more accurately and based on data in near real time . Thanks to Carl, measurement is based on clearly defined metrics that correspond to specific marketing goals.

In the case of O2, branded impressions are one of the most important metrics. " The most important metric for us is branded impressions. That is, only those impressions in which a message about O2 or our product was heard. Because it's great that an influencer has 300,000 views on a video, but if viewers stopped watching in the 20th second and we were only talked about in the 22nd, it's useless to us," says Markéta Zvolská, Social Media Specialist at O2. 

To comprehensively understand the performance of creators, O2 tracks:

• Branded Impressions – While regular impressions show how many times the content was viewed, branded impressions track how many times viewers actually encountered the brand or key campaign message – that is, the moment when the brand truly comes alive in the content .
• Branded CPM measures the cost per thousand views where the brand was actually seen. As Marketa mentions: “ Thanks to Carl, we know that branded CPM varies A LOT across creators. For some, it’s a few hundred, for others, thousands.” 
• Completed branded CPM goes even further - it shows how much a thousand completely completed branded messages cost - regardless of where the brand or key USP was.
• Brand attention time measures how long an audience pays attention to a brand – how much time they actually spend with it. Because the longer a brand or topic is exposed to viewers, the more likely it is that the user will remember it.

From these metrics, it is then possible to clearly monitor:

• long-term performance and price of individual influencers across campaigns ,
• types of outputs and their real impact – differences between individual social networks and in detail between individual formats,
• qualitative brand-oriented metrics and their development and comparison across content topics and campaigns (e.g., did the message about the new iPhone or 5G internet work better for the audience's attention (Attention time)?).

These metrics have become a key perspective for O2 in its evaluation and strategic decision-making. They allow us to find out exactly what works, why it works, and how to scale it .

More than 400 influencer outputs analyzed in detail

Since the beginning of the collaboration between the Carl for Social team and O2, we have managed to measure:

• 23campaigns,
• 106influencers and creators,
• 64unique product messages (campaign messages),
• 430outputs across platforms.

This already quite large amount of data allows us not only to compare individual campaigns, but also to evaluate influencers among themselves, monitor the development of their performance over time, and thus optimize content and influence strategy. As Markéta Zvolská mentions: “Delivering branded impressions to the target audience often has nothing to do with the number of followers. An influencer with 250,000 or 70,000 followers can have the same performance in conveying a message .”

Real-time measurement and data collection thanks to the mobile application

The key is that the data that O2 works with is not collected manually, as has been the case until now - there is no endless and tiring collection of screenshots and analytics from creators and transcribing them into tables. Carl works through its own mobile application Carl for Social , thanks to which the data is accurate, detailed and, above all, safe.

The application ecosystem connects three layers of data analysis:

1. Hard metrics – a direct line to data thanks to a mobile app
Creators involved in campaigns download the Carl for Social mobile app , in which they connect their social networks – Instagram, TikTok, YouTube or Facebook . Influencers who participate in campaigns voluntarily connect their accounts with the Carl for Social tool, which, thanks to the official APIs from the platforms, allows brands to analyze the performance of only public campaigns and their metrics.

2. Content analysis using AI
Each post is then analyzed second by second using advanced AI models. These recognize whether a brand or product appears in the content, what topics the creator mentions, in what context or with what sentiment. In total, more than 40 content parameters are analyzed every second. AI thus accurately detects what is actually happening in the content. The analysis only takes place on publicly available outputs marked as part of the collaboration.

3. Clear dashboard for brands
The collected "hard" data and content analysis results are automatically linked to advanced metrics and the analysis is displayed in the brand dashboard . Brands, such as O2, can see the performance of individual creators in relation to budgets and whether influencers were able to effectively convey the campaign's message (USPs), for example "O2 has the fastest 5G internet" or "the most reliable network" .
The result is clear feedback on which campaigns, topics or creators are actually meeting communication goals - and vice versa.

**\

Result: 216% more efficient budget allocation to creators (as part of O2's internal data comparison)**

The first influencer campaign is often more of a test than an immediate success. In the case of O2, however, a detailed look at the data quickly revealed where the weakness was. After adjusting the strategy and selection of creators, O2 managed to increase the overall effectiveness of the campaigns by 216% within a few months (this is a comparison of the first and most successful campaign within the framework of the cooperation). Thus, it spends the budget primarily on those creators who have low branded CPM and good quality of message delivery.

Improvement is seen across all metrics that matter to the brand:

• 3-fold increase in brand impressions coupled with a significant price drop
• the cost per thousand completed impressions (completed CPM) decreased by almost 20%,
• and the average time spent with content and audience attention on a post increased by 350% .

​

• 300%increase brand impressions associated with a significant price drop
• almost 20%lower cost per thousand completed impressions (completed CPM)
• +350%longer average time spent with content and audience attention to the post

Thanks to the data, the best of them are then evaluated. And they don't have to be big names. "A network of smaller and medium-sized creators has proven to be more effective than a few big faces with hundreds of thousands to millions of followers ," says Markéta Zvolská. 

And what's the best? Every shift in an individual metric means that the same budget has a much higher ROI for O2 today and, above all, a real impact on brand perception . "In the case of O2 and with regard to the product, influencers are not a sales channel, but serve to build the brand. We monitor this in data from Kantar and I am very happy to see how awareness and consideration are increasing among gen Z and gen alpha. Because young people are not so affected by ATL campaigns and our social media activities have a direct impact," adds Zvolská.

Data as an active tool

The collaboration with Carl for Social Business has brought O2's influencer marketing to a data-driven level. As a data-driven brand, O2 has been building its influence marketing strategies on data from the very beginning. But it was thanks to unified real-time reporting and new in-depth metrics that it became a precisely managed, long-term sustainable and constantly evolving system.

Influencer marketing has transformed for O2 from a communication channel into a strategic tool that really delivers results . Because when you have the right data in your hands, it's no longer about guesswork or luck. It's about knowing what works and why .

Carl for Social Business is a Czech influencer marketing management and evaluation tool that allows brands to analyze campaign performance across social networks and work more effectively with data in real time.