Pioneering Next-Gen AI to Scale Customer Acquisition: Q&A with Dzenis Softic

With 61% of marketers claiming AI is essential to their strategy, its adoption is no longer a luxury but a necessity for maintaining competitiveness.

AI-powered marketing tools offer a wealth of benefits, including up to a 30% reduction in customer acquisition costs and a revenue increase of up to 20%. This is why AI is well-established in ad tech, with its capabilities spanning data processing, pattern recognition, traffic forecasting, and ongoing analysis. The next frontier of innovation will be found in ways to weave AI into the backbone of every process involved in brands’ customer acquisition campaigns.

Q&A with Dzenis Softic

We spoke with Dzenis Softic, Chief Technology Officer & Head of AI Innovation at Perform^[cb], to discover how he and his team harness AI to drive innovation for marketers across the company’s proprietary suite of customer acquisition tools. 

Dzenis Softic
Chief Technology Officer & Head of AI Innovation

How are Perform^[cb]’s AI initiatives maximizing results for marketers?

The key to our success lies in the quality of traffic and service we provide. Our goal is to constantly elevate the value we offer to our clients, and AI plays a pivotal role in achieving this objective. AI uncovers intricate patterns within the vast sea of data our industry generates that humans might miss. By leveraging a mix of predictive and generative AI, we empower our clients to achieve their goals more efficiently than ever before. What once required extensive resources and time can now be accomplished swiftly with the aid of customized foundational models tailored to their specific business needs. This innovation translates to faster results and enhanced effectiveness, enabling precise targeting, higher conversion rates, and optimal utilization of marketing budgets. 

Thanks to the vast amount of data we’ve accumulated through years of working with diverse clients, traffic types, and verticals, we’ve identified key data points that correlate with quality. This has enabled us to develop and roll out a highly effective machine-learning model that optimizes ad traffic for maximum quality. We’re coining this new model PerformSense AI, and it has achieved up to 6x improvements in conversion rates. This success is possible because of our extensive, long-term market insights.

What forms of machine learning does Perform^[cb] use in its AI-powered media buying engine?

Perform^[cb] employs a combination of advanced and unsupervised machine learning techniques, including:

• Predictive Modeling: To forecast traffic quality and user engagement.
• Cluster Analysis: For effective audience segmentation and precise targeting.
• Large Language Models (LLM): To analyze textual data and enhance ad relevance.

What role does predictive analytics play in Perform^[cb]‘s AI-powered approach?

Providing predictive analytics to our Growth Managers is integral to our strategy, allowing us to predict traffic quality and volume with precision. This data fuels real-time optimizations and provides valuable insights to Growth Managers across client campaigns. By focusing on predictive insights that drive human + machine collaboration, rather than the traditional industry standard of historical reporting, marketers are able to achieve more efficient and effective marketing outcomes than older methods. Imagine being able to predict the outcomes of your actions – it becomes easy to identify the most valuable steps to take.

The Outcome Engine also uses AI to optimize ad placements through:

• Real-Time Data Analysis: Continuously analyzing campaign performance and user interaction data to make instant adjustments.
• Dynamic Budget Allocation: Automatically reallocating budget to the highest-performing channels and placements.
• Creative Testing and Selection: Running tests with multiple ad versions and choosing the top-performing ones.

How does Perform^[cb] ensure transparency and accountability in its AI-driven processes?

Explainable AI models, such as decision trees and certain types of regression models, provide insights into how the AI reached its conclusions, offering a transparent view of the decision-making process. This contrasts with unexplainable models, like some deep learning and black-box algorithms, which deliver results without clear, interpretable explanations of how those results were derived.

Explainable AI models not only deliver predictions but also offer explanations behind them, making it clear what parameters in the data caused the prediction. By using explainable models, Perform^[cb] fosters trust between the machine and our Growth Managers, and among clients by providing clear insights into campaign optimizations.

How is AI streamlining workflows and boosting productivity across the company?

Our approach has two facets. 

We automate repetitive tasks that were previously manual. My belief is that any data-driven decision can and should be automated. This not only eliminates the potential for errors but also significantly enhances scalability for marketers, enabling them to manage more complex data and delve into much more granular detail than ever before. 

Generative AI has become embedded throughout the company, with our tech team leading the way in leveraging it for rapid software development. This is especially true in our integration capabilities with clients and partner platforms, allowing us to ingest their quality metrics and gain deeper insights into the traffic, while at the same time, sharing our insights with partners so that they can optimize their ROI in the process. Another example is that we can now swiftly build API integrations with marketers and create custom data ingestions, expediting processes that were once time-consuming.

What advancements in AI technology do you believe will have the most profound impact on our industry in the next five years?

We are in the early stages of human-in-the-loop AI-boosted workflows, unlocking unparalleled efficiency and quality improvements. Imagine an AI that leverages all historical data and current data, performs detailed creative analysis, and then crafts a unique strategy for each client. Our team will be able to focus on using their judgment, experience and intuition to add their own ideas, test, and verify the strategy with the machine as a thought partner. Once finalized, the AI can execute it with speed, precision and at massive scale. This method, with AI at both the front and back end, with humans in the middle, ensures extremely high efficiency while preserving the creativity and intuition that only humans can offer.

How does Perform^[cb] stay ahead of the curve in the ever-evolving field of AI-driven marketing?

Our passionate team members, from top leadership to operational staff, infuse AI into everything we do. By staying at the forefront of AI advancements, we ensure our clients can utilize cutting-edge solutions that drive superior outcomes.

We’re committed to innovation through substantial investments in research and development, harnessing our proprietary data to train AI models and employ Large Language Models (LLMs) in an imaginative way that you wouldn’t expect. For example, while many chatbots currently rely on RAG (retrieval augmented generation) to extract data from company documents, essentially functioning as ChatGPT wrappers, we believe this approach will soon become outdated. Think about the early days of the iPhone, where flashlight apps were getting millions of installs but became obsolete once iOS integrated the feature natively – we anticipate a similar evolution in AI applications. 

True innovation lies in leveraging LLMs for non-chat applications, such as analyzing creatives and websites to understand what makes them successful or not, using LLMs for reasoning and creating strategies, or empowering LLMs with practical tools. Think of LLMs as brains that we can currently talk to, but with the addition of hands (tools), they can actively perform tasks. This is the future I’m excited about and working towards.

Ready to harness the power of AI to level up your customer acquisition? Reach out to our team of experts today.

aiartificial intelligencemarketing technologyMarTechoutcome-based marketingperformance marketingq&atechnology