Blend audited SELCO Community Credit Union's website for generative search and built a schema and content roadmap to make its content readable and citable by AI engines.
SELCO Community Credit Union has spent years building a deep financial education library, so when it was time to extend that library into the AI era, SELCO reached out to long-time partner Blend Interactive for a generative engine optimization audit and a new structural schema to help AI engines understand, trust, and cite that content.
Preparing a content library for the future of AI search.
Ask an AI assistant where to find an auto loan in Oregon, how credit union membership works, or what to watch for when buying a used car, and the answer no longer arrives as a list of links. It arrives as a written summary, pulled together from whatever content the AI tools can find and understand. For a credit union, that raises a simple question: when a member or a prospective member asks one of those questions, is your website part of the answer?
SELCO had the raw material to be part of a lot of answers. The site holds hundreds of pages of rates, product details, branch information, and financial education — articles on everything from budgeting to buying a first home. The challenge was that AI search engines were only picking up a fraction of what SELCO knows.
Great content needs a way to be found.
AI search engines don't read a page the way a person does. They look for structure and context — clear headings, direct answers, and behind-the-scenes markup that tells a machine "this is a financial product," "this is a branch location," "this is the answer to a common question." SELCO's site was built for traditional search, so it gave those engines very little of that structure to work with.
A baseline scan with SEMrush put SELCO's AI visibility score at 35 against an industry average of 63. The site was earning around 329 mentions a month across AI platforms, with 234 pages cited — a real presence, but well short of what the content could support. Most tellingly, the site carried no schema markup: the structured data that gives AI engines the clearest possible signal about what a page is and what it says.
Most tellingly, the site carried no schema markup: the structured data that gives AI engines the clearest possible signal about what a page says — in other words, what makes content legible to AI search.
Understanding what we’re working with.
Blend ran a full generative engine optimization audit of selco.org, which meant looking at the site from the angle of an AI crawler: reviewing the underlying HTML and semantic structure, heading hierarchy, meta content, existing schema, trust signals, and the way related content connected across the site.
The audit surfaced concrete, fixable issues — for example, an editorial scan showed that 20 pages shared titles, 87 pages had a generic description, and dozens of pages included headings that didn’t fully communicate the section’s purpose. More to the point, because this site was built before AI search was at the forefront, there was a complete lack of AI-forward schema.
At the same time, it also confirmed what was already working: descriptive URLs, a clean internal linking structure, and a deep content library just waiting to be noticed.
Turning schema into something editors can manage.
The heart of the project was a schema plan mapped directly to SELCO's existing content model. Rather than bolt on a generic markup package, Blend defined how schema.org structured data should work across every template and block on the site — from site-wide organization schema that identified SELCO as a credit union, to article schema with authorship and citations. Location schema helps surface hours and addresses, while event, bio, and glossary schema pulled from fields editors already fill in. And that’s really the key — schema is designed to work with many of the same fields that an editor is already filling out.
For the content that matters most to financial shoppers — rates and products — Blend designed standalone schema blocks. An editor building a page for a checking account or an auto loan can drop in a Financial Product block and rates will be structured based on existing page content. This gives AI search tools a clean, structured description of the offer. Accordion blocks can be flagged to produce FAQ schema, and resource lists can produce step-by-step How-To schema. The point throughout was to build a structure that SELCO's team can maintain on its own, not a one-time technical layer that goes stale after launch.
This work was faster and cleaner because Blend built and has supported SELCO's website for years. The schema plan maps to existing fields because Blend knew exactly what those fields were, and recommendations account for how SELCO's editors publish. That context — knowing the content model, the templates, and the team — was the difference between a generic audit and a plan SELCO could act on.
A roadmap, not a one-time fix.
Generative search is moving quickly, and trying to do everything at once would have stalled the work. Blend delivered a prioritized roadmap organized by impact and effort. Quick, low-risk editorial changes like heading updates and an llms.txt file allowed for instant progress, while site-wide and template-level schema were implemented as a part of SELCO’s ongoing support plan. Longer-term items, like video transcripts and PDF conversion, were flagged for longer-term fixes. Each phase builds toward better AI visibility while staying inside SELCO's existing development cycle.
Structure only goes so far if the writing underneath it buries the answer. Alongside the technical audit, Blend reviewed key pages for editorial clarity — whether paragraphs lead with a direct answer, whether headings describe what's actually in a section, or whether claims are specific enough for a machine to quote with confidence.
To make those habits stick, Blend led SELCO's team through a Writing for GEO workshop. The workshop covered how AI engines read content, how to structure a page for machine parsing, and how schema and clear writing reinforce each other. It also served as a great reminder to everyone: writing for GEO and accessibility means writing for real people.
Project summary: what Blend delivered.
- A generative engine optimization audit of selco.org, covering structure, editorial, schema, trust signals, and content organization
- A complete schema plan mapped to SELCO's existing templates, blocks, and content capsules — including editor-managed schema blocks, such as Financial Product
- An editorial audit of key pages with clarity and heading recommendations for AI readability
- A Writing for GEO workshop to equip the editorial team to publish AI-ready content
- A prioritized implementation roadmap phased across the year, from quick editorial fixes to site-wide schema and topic clusters
