Context
A multi-brand online education company: several school brands plus a corporate parent, serving teens, their parents, and adult learners returning for a diploma. Before I arrived: no CRM, ever. Pipedrive ran some email automations. Data existed in segments, scattered across siloed systems.
What Looked Broken
Depending on who you asked: the websites, the leads, the follow-up. But mostly nothing looked broken, because everyone could find a data fragment that supported their preferred story.
What I Noticed
That was the actual problem. Segmented data doesn't just limit visibility — it enables cherry-picking. When no system holds the whole funnel, "what's working" becomes a matter of which export you open, and every initiative can be declared a success by its owner.
The org also couldn't answer the question that mattered most: of the people who said "send me info," who enrolled? Enrollment was known. The journey to it wasn't. (The third appearance of this pattern in my career; see The Missing Middle.)
What the Evidence Showed
- Legacy web architecture: hundreds of pages as flat URLs, inconsistent hierarchy, overlapping content competing for attention, and .php/.html legacy redirects confusing both visitors and search engines — structural complexity accumulated over years, not anyone's failure of effort.
- No lifecycle: an inquiry landed wherever the nearest tool was, and close-lost prospects were never analyzed or re-engaged.
- No shared reporting: leadership had never had a dashboard.
The Actual Constraint
The company had grown a real business on top of invisible infrastructure. "CRM" wasn't a rejected idea — it was a foreign concept. The constraint wasn't technology selection; it was that visibility itself had never been an organizational capability, so building the system meant simultaneously building the literacy to want it.
My Hypothesis
Baby steps would beat a big bang. If we stood up a CRM, connected the webforms, and simply let data accumulate, then in six months the organization would be able to see something it had never seen, and seeing it would create the appetite for the next step.
What I Architected
- The company's first CRM (HubSpot), with integrations into the proprietary student information system.
- Web architecture across five brand websites: information architecture, message hierarchy, URL structure, redirect cleanup — clarity before visual polish, because families in a high-stakes decision need to know where they are and what to do next.
- Audience-specific content architecture: teen students, parents/guardians, and adult learners are three different anxieties requiring three different kinds of reassurance. Plain-language content for adult learners was deliberately structured so LLMs (ChatGPT, Gemini, Claude) could accurately index, summarize, and cite it alongside traditional search.
- The company's first executive dashboards.
- The leadership education to go with all of it: what a CRM is, what a lifecycle is, why "which export did you open" is not a reporting strategy.
What I Personally Built / Did
HubSpot setup and SIS integration work, the dashboards, the web IA and redirect strategy, the audience content strategy and key content, and a great deal of patient explaining in rooms where "CRM" needed defining before it could be defended.
What Happened
Webform conversion improved once we could see it (I no longer have access to the exact figures, so I'll say only that: it improved, and we could finally know it improved rather than assert it). After six months of collected data, the company could for the first time follow a person from "send me info" to enrolled-or-not. The cherry-picking era ended, not because anyone was scolded, but because a shared view existed.
What Became Possible
Real questions: which sources produce students, not just leads. Why we lose the ones we lose. Re-engagement of close-lost inquiries as a strategy rather than an accident. None of these were askable before, and all of them are the difference between marketing spend and marketing investment.
What I Learned
When an organization has never had visibility, the hard part isn't the software. It's that nobody has ever experienced the questions the software answers, so nobody misses them. You aren't implementing a tool; you're introducing a sense the organization has never used. Baby steps aren't a compromise. They're the method.
Evidence / Artifacts
The five brand websites shipped and are publicly visible; screenshots and dated archive captures from my tenure serve as the record (some post-departure changes have degraded elements of the live sites). System map recreated as an illustrative diagram. CRM internals are not shown.
website / Content & Authority
Plain-Language Content Built for Search and LLM Citation
Long-form content for adult learners, written in plain language to reduce cognitive load while structured densely enough for accurate indexing, summarization, and citation by language models.
Written for anxious adults and, deliberately, for the machines they now ask first.
Related Work
Pattern siblings: The Missing Middle and Documented ≠ True (the five-interpretations test was run here too). For the personal story behind one of these builds, the manifesto chapter I Still Built the Dashboards tells it in a different register.