The System Behind the Symptom: From One Fix to a Five-Function Engagement

Challenge

It started with a credibility problem: a damaging internal narrative claiming Embarc's Member Experience function had quietly lowered acceptance standards to keep up with a sharp rise in applications. But a deeper problem was already there waiting to be found — Embarc's entire member lifecycle ran across eight-plus disconnected systems, with no single record showing where any applicant or member actually stood, quietly shaping the daily work of not just Member Experience, but Content, Partnerships, Programming, and PR as well.

Approach

The first fix came quickly. As the embedded operator already inside Embarc's systems, I reconciled 388 pre-2026 applications against 88 from 2026 and 325 all-time accepted companies, traced the apparent standards drop to a data artifact, and built a scoring rubric plus a custom GPT triage tool. That fix turned out to be the first visible symptom of something much bigger.

Following the same thread, I ran a full-scope operational audit: 95 application records, 117 company records, 261 individual member records, finding only 62% of accepted applicants matched to both a downstream company and founder record. I evaluated an incumbent vendor's proposal against this reality, found it addressed reporting rather than the underlying breakage, and designed an alternative — an “Onboarding Case” architecture replacing notification-based automation with state-based tracking, plus a live measurement framework now serving as the operating dashboard for ongoing work.

What began as one correction expanded into ongoing, embedded management spanning multiple functions, closer in scope to a fractional AI officer than a single project — because the same broken foundation touched nearly every team that depends on member and applicant data.

Results

• Member Experience moved from defending an unverified narrative to operating on evidence — and reclaimed 40–120 hours a year in the process, worth an estimated $1,800–$9,000.

• Replaced ad hoc judgment with a validated, evidence-based triage rubric — still in active use for every application since.

• Content can trust its source material again. Founder LinkedIn data had dropped from 94 of 95 present at application to just 23 of 123 in active records — now carried forward instead of re-collected.

• Partnerships can stand behind what it reports to funders and corporate partners, on a foundation that's actually been checked.

• Member Programming can trust who's actually eligible for a cohort invite, instead of cross-referencing multiple systems before every decision.

• PR stopped structurally missing member wins that were previously tracked across scattered, incomplete areas.

• The audit outperformed an outside vendor's proposal already in motion, finding the actual root cause where the vendor's plan addressed only reporting.

• Built a live measurement framework now used as an operating dashboard — tracking lead time, wait time, chase touches, and data carryover completeness — not a one-time diagnostic that got filed away.

• What began as a single fix grew into a broader, ongoing engagement — expanding because the same root cause kept surfacing across functions, not because a bigger scope was sold.

Why This Matters

This is really two moments in the same engagement: a fast fix that happened to be the first proof of what embedded access could catch, and the full-scope audit that followed once the same root cause kept surfacing function after function. Neither moment is as convincing without the other — the fast fix proves the instinct, the broader audit proves the pattern. That's the real case for treating AI operations as an ongoing, embedded relationship instead of a single project: some of the highest-value work looks like nothing until you're already close enough to the data to catch it.