IA models with ROI and hidden costs

About the event

​Susan will be showing a few use cases to explain how a structured Information Architecture (IA) can help a start-up gain control of scattered documentation, reduce wasted effort, and strengthen operational credibility, resulting in saved staff time, company costs, and client trust.

​About the speaker

Susan Kraft Yorke is a technical writing consultant with more than 30 years in the tech industry. She spent most of my career in Palo Alto, California, and now works from home in the Hudson Valley near New York City. My technical background started as a geophysicist and in the fine arts. Her beginnings influence how she approaches complex systems and communication.

​She specializes in information architecture and developer documentation for startups, growth-stage companies, and global enterprises. Her work focuses on diagnosing documentation breakdowns, architecting scalable publication systems, and leading large-scale documentation migrations. She helps organizations reduce search time, eliminate content deadwood, and restore trust in their documentation through structure, governance, and measurable outcomes.

​She brings both in-house and consulting experience, grounding IA decisions in real systems, real constraints, and real operational impact. Her first IA project was in 2019 at Intuit. It's been an eye-opening experience ever since.

​Susan's LinkedIn: https://www.linkedin.com/in/susankraft-yorke/

Susan's website: https://www.susankraftconsulting.com/

---

Event recording

Vimeo: https://vimeo.com/worldiaday/ia-models-with-roi-and-hidden-costs

---

Key Takeaways^_*

• Organizations lose approximately 20% of employee productivity daily due to poor information architecture and documentation retrieval failures

• The shift from controlled documentation systems to decentralized wikis created unregulated digital publishing without proper governance

• A mid-sized service firm can realize approximately $1.3 million in benefits over three years through updated IA infrastructure

• AI governance infrastructure can leverage existing IA structural frameworks, but requires understanding that value lies in retrieval and knowledge infrastructure, not just physical and transactional systems

• The ROI equation for IA improvements is: (Total Benefits - Total Investment) / Total Investment × 100%

• Total investment includes recurring costs (compliance blind spots, 20% daily search time, duplication, author interruptions) and one-time costs (file sorting, structure design, migration, testing)

• Information architects must speak the language of finance and business to overcome resistance to change, addressing concepts like valuation gaps and productivity leakage

• Companies often remain stuck at operational maturity levels despite size and age, indicating poor information management practices

^{_{*Zoom AI generated}}

---

Q&A^_*

• How do you calculate total benefits in the ROI equation, since it feels very broad?

  • Financial benefits include the cost of 20% daily productivity loss, which requires analysis to measure actual time wasted. Benefits also include reduced support costs (fewer calls about how things work), compliance improvements (hospitals sometimes can't find required documents during reviews), elimination of documentation duplication, reduced rewriting of existing content, and decreased author interruptions. For example, at Intuit, engineers writing tools for marketing and sales were constantly interrupted by calls asking where documentation was located, as it was scattered across GitWiki, Confluence, Google Docs, and three other platforms. Sometimes benefits include eliminating unused software licenses like Confluence seats. While not as precise as grocery bills or rent, these costs can be reasonably estimated.

• What's the best way to convince companies beyond showing calculations, especially when they see large numbers and say "not now, we'll do it later"?

  • The mistake was not talking to them about financial issues. Interview stakeholders like you would any subject matter expert to understand what's blocking them—some are afraid to know how bad things really are, others genuinely believe they don't have time, but they're losing time by not sharpening the axe. Business people will help more than engineering managers. If you can reach someone who considers business aspects, especially the valuation gap between finance and engineering output, they might not be aware of current trends since they rely on past knowledge. AI can help here—use it to craft better questions and presentations. The containment recipe approach (available on susancraftconsulting.com under the AI channel) shows how to create tight AI questions that go beyond treating it like Google, always asking it to "self-analyze and try again" for better responses.

• What cases have you observed where using LLMs actually makes a difference?

  • It made a difference for me in writing markdown documentation by knowing my style guide through long-term memory. For 700-800 files migrated from multiple wikis and Confluence to GitHub for a think tank group, AI helped ensure style guide consistency. You must give it long-term memory instructions—mine is 5 pages for technical writing and research. For every request, use a containment recipe: who they are, the goal, the task, the context, and output format. Keep it contained with guardrails.

^{_{*Zoom AI generated}}