Improving completion in consumer health enrollment
Sierron provides AI and modernization advisory in support of the Maryland Health Benefit Exchange, focused on where artificial intelligence helps consumers complete enrollment rather than simply answer questions about it.
Program context
The Maryland Health Benefit Exchange operates Maryland Health Connection, the State’s health insurance marketplace. The exchange determines eligibility, presents plan options, and enrolls residents in qualified health plans and Medicaid, coordinating with the Maryland Department of Health on eligibility pathways that cross program boundaries.
Enrollment is seasonal and concentrated. Open enrollment generates volume the remainder of the year does not, compressing the period during which consumer facing systems must perform and narrowing the window for correcting problems that surface under load.
Challenge
Consumer enrollment presents a measurement problem that is commonly misdiagnosed. Programs track interaction volume, session duration, and satisfaction, each describing system activity rather than consumer success. The measure aligning the exchange’s interest with the consumer’s is completion, meaning the proportion of individuals who begin enrollment and finish it.
Conversational assistance can perform well against interaction measures while completion remains flat. An assistant that accurately answers questions about a difficult process does not make the process less difficult. Where an individual abandons enrollment because a document requirement was described in terms not matching what they possessed, assistance offered earlier in the sequence does not address the cause.
Enrollment additionally requires coordination across program boundaries, since eligibility may resolve to a qualified health plan, to Medicaid, or to a determination requiring additional documentation. Each transition point introduces an opportunity for abandonment.
Sierron’s role
Sierron leads the artificial intelligence approach for the consumer enrollment experience, defining the measurement model, selecting use cases, and setting the boundary between advisory and determinative functions.
Approach
Sierron instruments the enrollment sequence at step level before capability is selected. Aggregate abandonment figures establish that a problem exists; step level measurement establishes where. In consumer transaction flows, abandonment characteristically concentrates at a small number of specific points rather than distributing evenly across the sequence.
Those points are frequently known to frontline staff before analytics reveal them. Sierron begins with call center and navigator staff, on the basis that personnel handling exceptions can usually identify the difficult steps directly.
Once concentration points are established, candidate applications become concrete. In enrollment environments these commonly include eligibility pre screening providing a plain language indication before the consumer invests substantial effort, translation of statutory and plan terminology into descriptions of what the consumer must actually provide, validation of uploaded documentation at submission rather than at subsequent rejection, and status transparency sufficient to reduce inbound contact volume.
Three constraints hold in this environment. Artificial intelligence supports the consumer and the caseworker; it does not make eligibility determinations. Every consequential path retains a human route, since some circumstances are genuinely exceptional and some consumers require a person. Automated assistance is disclosed, with escalation to a human available without persistence.
Multilingual access warrants specific attention. Translation represents one of the stronger applications of artificial intelligence in consumer facing public service, provided translated content affecting eligibility or benefits passes human review before publication.
Outcomes to date
The engagement is ongoing. Work delivered to date includes:
- Adoption of completion rate as the primary measure for evaluating consumer facing capability
- A step level instrumentation approach for identifying abandonment concentration
- Use case selection grounded in identified abandonment causes rather than available technology
- Explicit boundaries maintaining human authority over eligibility determination