A major health insurer
CareFirst BlueCross BlueShield is a not-for-profit health insurer serving more than 3 million members in Maryland, Washington, DC and Northern Virginia.
How 32 interviews in three weeks pointed CareFirst to the one fix its brokers needed most, validated with a broker survey before anything was built.

Context
The company and the stakes
CareFirst BlueCross BlueShield is a not-for-profit health insurer serving more than 3 million members in Maryland, Washington, DC and Northern Virginia.
Independent brokers sell CareFirst plans to employers, then help those groups through enrollment, service issues and renewals.
At CareFirst’s 20% close ratio, every 10% more pipeline from existing brokers means about 2,000 new members. Poor service was putting that pipeline at risk.
At a glance
From noise to one priority
Challenges
What made it hard
Research and strategy had a three-month window, so all 32 interviews had to land in the first three weeks.
I logged every interview in the same template the day it happened, so themes surfaced while interviews were still underway.
CareFirst arranged every interview. We had no say in which brokers or employees we heard from.
I ran an 18-broker survey to test the interview themes with a wider group before we committed to a direction.
I ran all 32 interviews myself, with no notetaker and no AI transcription tools at the time.
I took notes live in every session, then synthesized the raw data by hand, affinity diagramming in Miro.
01 · Listen
Weeks 1–3
I ran all 32 interviews myself, taking notes live, across both sides of the relationship (the brokers selling CareFirst plans and the internal teams who support them after the sale). Two ideas came up again and again. People love CareFirst’s network and brand, and doing business with CareFirst is really hard. Every broker said response times were far too long.
17
Broker interviews with general producers and wholesalers
15
Stakeholder interviews across sales, implementation, underwriting and support
1
On-site visioning workshop with CareFirst
“Is it CareWorst, CareLess, or CareFirst?”
“A lot of the time [our clients] think it’s us… but it’s that we can’t reach someone at CareFirst.”
“They’ll put 65 people on an email hoping that someone will chime in and resolve it.”
“I can sometimes be on the phone for 45 minutes trying to get something small resolved. That time is costing me money.”
Instead of demographic personas, I built archetypes around roles and emotions. They showed that a broker’s reputation depends on every link. Brokers aren’t happy until members are happy.
02 · Trace & connect
Backstage → blueprint
Brokers only see the frontstage. To find out why things went wrong, I mapped eight backstage processes, from contracting and credentialing to renewals that go to RFP. The cost showed up everywhere.
The service blueprint spans sales, implementation and retention, and it tracks how brokers and CareFirst employees feel at every step. Both lines sink in the same place, plan administration.
03 · Diagnose
Fail points in the blueprint
The blueprint made one thing obvious. When CareFirst stumbles, the broker takes the hit in front of the group they’re trying to keep.
When a broker can’t give a group the best service because of CareFirst’s limitations, their reputation is on the line, and they risk losing the client.
A support center in the broker portal for submitting issues and tracking their status, so brokers can keep groups informed instead of waiting in the dark.
Jump to the solution ↓Renewing is easiest for everyone. But if a group isn’t happy with renewal rates, the broker has to shop for an alternative carrier.
A future-state renewal with meetings five months out, data-backed proposals, and a post-renewal retro so the next one starts with what was learned.
See the future state ↓The time it takes to resolve broker issues is tarnishing CareFirst’s reputation.
A service request system that routes every ticket to the broker’s designated customer success manager, tagged by urgency. No more fishing through inboxes.
Jump to the solution ↓04 · Validate
18-broker survey
Interviews told us what hurt. Before recommending a product, I wanted to know what would actually change broker behavior, so I surveyed 18 brokers on what would make them write more business with a carrier.
“Another login on top of the 50 or so I already have is an extra burden.”
“It would not work if it replaced my account rep.”
Brokers, survey05 · Decide
Strategy & roadmap
members for every 10% increase in pipeline from existing brokers, at CareFirst’s 20% close ratio. Better broker service isn’t a nice-to-have. It’s growth.
Primary
Secondary
Software alone couldn’t fix a relationship nobody owned. We recommended a new team that gives every broker a single point of contact 100% focused on broker needs, so account consultants can focus on sales.
06 · Design
Future state
I designed the experience on both sides of the counter, with a support center in the broker portal, and a service request system that routes, prioritizes and reports on every request.
ScenarioA member can’t get care because of an enrollment issue, and the broker needs CareFirst’s help to fix it fast.
ScenarioA group is setting up with CareFirst for the first time.
ScenarioA broker is negotiating a renewal for a client of one year.
Brokers already juggle dozens of logins, so support lives inside the broker portal they already use.
↳ Rule 01 · No new loginEvery open ticket and its status (New, In progress, Overdue) is the first thing a broker sees after signing in.
↳ 16 of 18 brokers asked for status visibilityA keyword search surfaces support resources, so simple questions don’t wait days for a reply.
↳ Brokers wanted to troubleshoot minor issues themselvesFrequently asked questions grouped by category, with a jump menu for fast scanning.
↳ Self-service for the simple stuffName and email come from the broker’s profile, and a request type routes the ticket to the right team.
↳ Replaces “65 people on an email”A request number, response times by priority (same day for urgent, 24–48 hours for the rest), and updates in the portal and by email.
↳ Ends “hear nothing for days”Status, dates and a named assignee, so the broker always knows who has their request.
↳ Rule 02 · A human owns every ticketCustomer success managers see their open tickets, the high-priority requests, and their progress against service goals like resolution time.
↳ Replaces fishing through shared inboxesReps change a ticket’s status, take action or reassign it, and every update reaches the broker without a follow-up email.
↳ Rule 03 · Status without chasing07 · Impact
Handoff
My part ended with a validated strategy, ready to build. It included the research, 4 archetypes, current- and future-state service blueprints, a prioritized portal roadmap, an org recommendation, and two clickable prototypes.
08 · Reflect
Looking back
I took every note by hand and affinity-mapped all 32 interviews myself. Today I’d use AI transcription and clustering to get there in hours and spend the time saved testing prototypes with brokers.
I’d turn the backstage numbers (time to resolution, consultants’ hours on service) into baselines at kickoff, so impact could be proven, not just predicted.
I’d push for a say in who we interview, including brokers who had moved groups away from CareFirst.