Advice Moments

For a self-directed investor, advice is still a separate decision and motion. The investor who would benefit most, navigating a rollover, a windfall, a new goal, or simply rising complexity, rarely gets a relevant, plain-language prompt while the need is live. That activation gap, at the first $50K step on the continuum above, is the opportunity: detect the moment and respond to it well, in-product, at scale, and within fiduciary guardrails.

Drag the balance and pick a life event. Watch eligibility cross the real public thresholds, and see the governed nudge fire, or deliberately suppress itself below the benefit threshold.

A representative view in the team's own language. Real advice funnels are far steeper than the bars suggest: reach is enormous, conversion to human advice (PA/PAS) is a fraction of a percent, while Digital Advisor self-serve converts far higher. Drag the reach to see it move.

A recommendation engine is probabilistic, so the spec defines what good looks like in numbers, how it fails safely, and where it must stop. In a fiduciary context that is not optional polish, it is the product. Release gates (illustrative): suitability precision ≥ 99.5%, significant incremental benefit vs. a holdout, zero coercive outputs, factuality blocks on fail. The eval harness behind those gates: ground truth built from expert-reviewed samples, LLM-as-judge scoring calibrated against human raters, and production monitoring for drift, with rollback as a first-class path.

The case for advice is not only growth, it is durability. Advised relationships churn far less than self-directed accounts, so every appropriate match also lifts retention. A churn-propensity model can flag at-risk relationships for proactive outreach.

Illustrative annual attrition by tier, directional. Retention modeling and churn scoring are part of my prior analytics work.