GCP – The AI lens: How Arpeely uses multimodality and BigQuery to revolutionize AdTech
Traditional programmatic advertising often misses the mark. Flat pricing, limited targeting, and a focus on immediate conversions over long-term customer value leave advertisers wanting more. At Arpeely, we’re changing the game in three ways by:
-
Putting our own money on the line for performance with a different business model.
-
Taking care of creatives and the funnel end-to-end.
-
Optimizing on lifetime value.
Arpeely is on a mission to transform the way advertisers connect with their audiences online, fueled by one simple belief: To truly understand the internet, you need to see it. Not just the text and code — but the images, the emotions, the nuances that make each web page unique. That’s why we’re doubling-down on our ad-tech platform on Google Cloud.
In this post, we’ll take a closer look at how Google Cloud is helping us to leverage the power of multimodality and AI in products like BigQuery and Gemini to make media buying smarter, more efficient, and laser-focused on long-term value.
Transforming how Arpeely ‘sees’ the internet with multimodal AI
Our AI algorithms analyze massive datasets to identify the users most likely to become loyal, high-value customers for our clients, but we need to understand the web on a deeper level to do that effectively. That’s where multimodality — the ability to process multiple types of data — comes in. We don’t just look at text; we use Google Cloud’s powerful multimodal AI, including Gemini Pro Vision and Gemini Flash 1.5 to analyze web page screenshots, extracting visual information that enriches our understanding in real time. This allows us to cluster billions of websites with remarkable speed and precision, uncovering connections and insights that traditional methods would miss.
- aside_block
- <ListValue: [StructValue([(‘title’, ‘$300 in free credit to try Google Cloud data analytics’), (‘body’, <wagtail.rich_text.RichText object at 0x3e89d94d9be0>), (‘btn_text’, ‘Start building for free’), (‘href’, ‘http://console.cloud.google.com/freetrial?redirectPath=/bigquery/’), (‘image’, None)])]>
Taming the data deluge with BigQuery and Pub/Sub
Building an AI-powered, multimodal ad platform requires handling a truly staggering amount of data. With BigQuery, we can constantly crunch numbers, analyze user behavior, and generate insights from over 25 petabytes of compressed data, fueling our real-time bidding engine.
But AI demands speed as well as scale. That’s why we rely on Pub/Sub, Google Cloud’s real-time messaging service, to keep the information flowing. Pub/Sub acts like our central nervous system, connecting our microservices and ensuring that our AI algorithms have the up-to-the-second data they need to make smart decisions.
Going beyond keywords: BigQuery vector search for unprecedented ad relevance
Traditional ad targeting relies on keywords, which are blunt instruments in the nuanced world of online behavior. Apeely takes a smarter approach by using the ML.GENERATE_EMBEDDING function within BigQuery ML to generate embeddings for the webpage in real time. By representing web pages and ad creatives as vectors in a multi-dimensional space and using BigQuery vector search, we can understand the semantic relationships between them in real time. This means we can deliver highly contextual ads that go beyond simple keyword matching, resulting in greater relevance, higher click-through rates, and better campaign performance for our clients, with a 15% uplift in revenue.
Making ads blend in and stand out with Visual Question Answering
Our commitment to understanding the visual web goes even further with Visual Question Answering (VQA). By training AI models to “see” and interpret images, we can extract detailed information about web pages, such as dominant colors, layouts, and even emotional tones. Our VQA models enable us to dynamically adjust the look and feel of our ads to match the context of each page, creating a more seamless and engaging experience for users, resulting in a 28% increase in user engagement.
Image created using Gemini
The Google Cloud advantage: Building the Future of AdTech
Building Arpeely on Google Cloud has been instrumental in bringing our AI-powered vision to life. The platform’s scalability, serverless offerings, and unified ecosystem give us the agility and efficiency we need to innovate at a rapid pace.
We’re incredibly excited about the future of ad-tech and the role AI will continue to play. With Google Cloud as our trusted partner, we’re confident in our ability to lead the way toward a more intelligent, effective, and value-driven advertising landscape.
Get started with multimodality use cases in BigQuery today.
Read More for the details.