At a glance
Industry
Banking
Location
Netherlands
Challenge
Maintain brand visibility in a “zero-click” world by using agentic AI to optimize thousands of marketing web pages for AI search.
Success Highlights
- Thousands of marketing web pages now optimizable for AI search—achievable in days
- Five agentic AI building blocks, creating a complete solution for AI search optimization
- 276% higher optimization for AI search, compared with existing marketing web pages
The challenge
A chance to get ahead in AI search with agentic AI
Our client wanted to be ready for this shift, especially as it presented an opportunity to broaden the bank’s customer base. By optimizing its web pages for AI search as well as traditional search, it would not only maintain strong brand visibility but also get ahead of competitors still focused on SEO.
However, the bank’s website has thousands of marketing pages. Optimizing each one manually, even once, would take too much time, resource and budget. And GEO isn’t a “one-and-done” exercise. As LLMs evolve, the way they sort and prioritize information changes. Our client would have to continually optimize its web pages to stay visible in AI search results.
The bank had attended some of our Cognizant Mindshare sessions where we showcased the potential of agentic AI for intelligent automation. Their digital marketing team knew what needed to happen. The obstacle was scale.
Our approach
Search engine optimization gets brands to the top of the search results. What's needed now is something different: getting the brand into the AI-generated answer. That discipline is called generative engine optimization (GEO). Where SEO is about ranking, GEO is about being cited. GEO content should be clear, well structured, on-brand, tailored to the audience’s needs and easy to summarize. In highly regulated sectors like banking, it also needs to adhere to strict industry rules. These needs formed our brief for the project.
We designed an initial multi-agentic AI solution that ingests web pages, optimizes them for GEO, applies style and compliance rules, scores the results and outputs an updated version for human review. With a pilot solution delivered in just four weeks, our client could quickly and cost-effectively evaluate the potential of agentic AI for GEO before committing to broader rollout.
A solution based on five agentic building blocks
Our solution treats the entire content refresh cycle as an automated pipeline—ingesting pages, optimizing them, checking them against brand and compliance rules, scoring the output, and routing anything below threshold back through the cycle until it passes. The solution runs on five interconnected agent networks:
- Multi-language content ingestion: This agent network ingests web pages in Dutch and English and gets them ready for SEO/GEO optimization.
- Rules and guidelines: These agents regularly retrieve brand style and tone guides, legal and compliance rules, and changing guidelines on SEO/GEO best practices. They store it all in a knowledge base, along with a best-practice framework for SEO/GEO writing.
- Gap analysis: These agents compare ingested web pages with the rules and best practices in the knowledge base and flag up content that needs to be refreshed.
- Content refresh: This agent network refreshes the flagged-up content using the best-practice writing framework and the other rules and guidelines from the knowledge base.
- Content scoring: This network reviews the output of the content refresh agents and scores it for conformity with the rules, guidelines and best practices. Lower-scoring content is sent back through the cycle until it reaches an acceptable threshold.
- Orchestrator: Similar to a project manager, the orchestrator agent monitors and directs the activity of all the agent networks in the solution.
At the end of the process, refreshed content that has achieved a high enough score is passed to a human subject matter expert for review. As the content has already scored highly, this is just a light pass, rather than an in-depth editing job for the reviewer.
Making ROI measurable: The CLASSIC framework
Our client wanted to be able to measure the value of the agentic AI solution to understand its ROI potential. Cognizant addressed this directly by building the content scoring layer around CLASSIC—a proprietary framework purpose-built to measure AI search effectiveness. It produces a weighted score across GEO criteria including clarity, authority, relevance and conversational usefulness, with a maximum of 450.
Working with the bank, we established clear thresholds. Above 300, content is approved for release. Below 200, it goes back for rework. The framework isn't static—it can continue to track scores post-release, so the bank can measure GEO effectiveness and recalibrate thresholds over time. Our joint mantra: Test before you publish. Improve as you learn.
Business outcomes
The results were immediate and measurable. The initial agentic AI solution quickly showed it could refresh the content, tone, structure and formatting of marketing web pages to optimize their performance in AI search. Key outcomes from the pilot included:
- 276% higher optimization for AI search: On average, our client’s existing web pages scored just 85/450. This increased to 230/450 after a first cycle through the agentic AI solution, and 320/450 after the second—a 276% boost in AI search optimization.
- Ultra-fast content production and review cycles: The pilot demonstrated that thousands of web pages could be ingested, refined and ready for final human review in days rather than months—that speed is operationally transformative. This would make it an ideal rolling solution, allowing web pages to be rapidly refreshed whenever AI search parameters change, the brand’s style guide changes or new compliance rules come into force.
- Significant ROI in terms of brand visibility and lead generation: The risk with “zero-click” searches is that if the brand isn’t visible and positively portrayed in the AI search output, there’s no second chance to impress. The agentic solution ensures that every web page is given the best possible chance of showing up in AI summaries and recommendations, maintaining brand visibility for our client and maximizing the potential for leads and conversions.
