AI data specialist StageZero’s report “AI Adoption in Europe 2022: How high performers generate value” provides clear insights into how businesses across European industries are adopting AI. 

The findings focus on how companies in Europe develop AI solutions, how they obtain training data, what high-performing companies are doing differently to generate value, and how they overcome challenges along the way. The survey interrogates data science and machine learning leads, engineers, and decision-makers across multiple industries and functions. 

The report shines an optimistic light on AI implementation with the majority of companies enjoying strong value returns from their AI projects. Companies are increasing their investments in AI and launching more projects, signaling a maturing industry.

StageZero also reports what the high-performing companies do differently to stay ahead: most significantly, they collaborate more frequently with partners to source and annotate training data.  Rarely is that handled in-house. Their principal focus is on cost-savings and internal process efficiency. And they are pumping resources into MLOps, fine-tuning their models once deployed.

The full report is available for download here.

Challenges in AI adoption

StageZero identifies three main challenges that face companies that want to innovate within AI: data scarcity, data quality, and regulatory compliance. A lack of data, especially quality data, stands out as the most significant issue in developing successful AI models.

The report explains why companies would develop more AI solutions if data-sourcing challenges were solved. Companies realize that data quality and quantity are key to innovation… and innovation means enhanced competitivity. But data is an issue.

While data appears to be abundant, companies need enriched training data sets with high-quality annotations in order to deliver the desired results – not as easy as it might sound.  Even the high performers are still struggling to resolve this, but working with partners seems to be a suitable solution.

Finally, regulatory compliance is considered a top challenge to AI adoption across different markets. This is expected to become tougher over time as AI becomes increasingly regulated across the board.

Who leads AI implementation?

The report identifies companies that are ahead of competitors, how they behave, and how they generate value from AI. 

The report defines high performers based on how the companies themselves assess their AI efforts. The respondents were asked to rank their AI and machine learning efforts compared to their competitors. Almost half (43%) see themselves as ahead of the competition, whereas 32% position themselves as roughly the same as their competitors, and 24% think they lag (Chart 16). StageZero considers 43% as high performers. Company size was unrelated to performance.  

AI development stage

The companies most likely to succeed in AI development are already well underway with their machine learning operations. High performers prioritize AI and regard it as critical. None of the companies that are ahead of competitors are still in the early AI development stages (Chart 17), indicating that high performers are more likely to be scaled up – or they’re already at full speed and have MLOps running as standard.


High performers have a more centralized AI and ML approach, or work in a hybrid setup. 50% of leaders are working in a hybrid setup, 31% in a centralized setup, and 6% in a mostly centralized setup (Chart 19). StageZero reports that 44% of companies behind the competition are decentralized or mostly decentralized. This indicates that AI project success is tied to a dedicated strategic focus on AI, including centralized decision-making and budgeting.

Companies ahead of the competition work in a centralized or hybrid AI/ML setup.

Compared to other companies, management buy-in appears to be less of a challenge to high performers as around half (44%) report it as somewhat of an issue, with 38% saying it is not at all an issue and only 13% claiming it is an issue. The results of those ranking their AI efforts lower than their competitors differ, as of these companies, 33% report management buy-in as a significant challenge to gaining value from AI (see Chart 20).

The numbers suggest that companies with management buy-in have a higher success rate in AI development and that workers would appreciate increased support from those in leadership roles.   A whopping 33% report management buy-in as a significant challenge, implying ample room for improvement there.

Data sourcing

Most high performers (75%) collect existing customer data and other internal data. Many (69%) create their own data sets. Over half (56%) partner with data providers who collect real-world data (Chart 21).

Synthetic data is not yet widespread and only 6% of leaders collaborate with partners to get synthetic data. 25% create their own synthetic datasets. This signals that using real-world data is the default, and synthetic data is mainly used when scenarios are identified where real-world data is insufficient or unsuitable (e.g. due to regulatory constraints).

Leading companies are more willing to turn to external data providers for help. When asked if they would work with third parties to solve data issues, the majority of high performers (81%) reported either “yes, definitely” or “somewhat/probably” (Chart 22). This corresponds with the results in Chart 21. In comparison, the rest of the respondents are willing to work with third-party data providers, but to a smaller extent (67%).

The report concludes that high performers are more willing to collaborate and use third parties to help handle data issues. It indicates that more flexible companies that work with partners have a better success rate in AI projects.


While leaders see greater average value from implementations (Chart 23), others aren’t far behind, so the report concludes that companies gain benefits from AI regardless of their positioning among competitors.

When implementing AI, high performers cite cost savings (21) and internal process efficiencies/automation (21) most often. However, improved insights from analytics (including improved decision-making) are cited across the board.

Model monitoring and fine-tuning

StageZero reports a correlation between greater competitive strength and tracking the performance of deployed machine learning models and then fine-tuning them by adjusting them to improve performance. High performers are more concerned about model performance and fine-tuning their models at a higher frequency.

46% monitor some of their ML models, while 35% monitor all deployed models (Chart 24). Meanwhile, all leaders monitor their models, with 56% monitoring some models, and the rest monitoring all deployed models.

Most high-performing companies fine-tune their AI models continuously. 32% fine-tune at least quarterly and 38% fine-tune more than monthly (Chart 25).


StageZero’s report examines AI adoption throughout Europe and identifies what high-performing companies do differently from their competition. The report pinpoints key attributes of high performers and solutions to common challenges in AI adoption.

The report indicates AI implementations provide benefits for companies and their customer’s overall business functions and types of solutions. 

When it comes to data sourcing and annotation, high performers are more likely to collaborate with third parties. On an organizational plan, high performers are more centralized in decision-making and budgeting, and more frequently perceive management buy-in.

Regarding implementations, they focus more on cost savings and internal process efficiencies than the others. When it comes to deployment and monitoring, they focus more heavily on MLOps and fine-tune their models more frequently.

About The Author

Scroll to Top
Share via
Copy link
Powered by Social Snap