# How to Use a Platform Business Model to Manage External Service Providers in Healthcare
A healthcare services scale-up running Crystal has two to five people on its product development team. They've been around for four years and have one hundred sixty-eight employees. The team just started building a new chronic care management module. (1/21)
The problem is external service providers. They're unreliable and the team depends on their integrations to deliver features. When those integrations break, the team slips on deadlines. Late features mean lost contracts. Last quarter alone, the company lost eighty-seven thousand dollars, which was about a third of the chronic care management module's revenue.
That's a serious number for a three-person team. (2/21)
Jack Ma faced a similar problem at Alibaba. His insight was simple. Companies fail when they try to do everything themselves and control every part of the value chain. You spread yourself too thin and can't keep up. The competitors who partner and build ecosystems end up winning.
Ma's answer was the platform business model. Build the platform. Let others build on it. Manage the ecosystem, not the pieces. (3/21)
He didn't build Alibaba.com as a marketplace where Alibaba played the seller. He built a platform where sellers plugged in. More sellers brought more buyers. More buyers brought more sellers. The ecosystem grew itself.
Same thing with Tmall. He didn't stock inventory. He built the platform and let brands plug in. More brands brought more customers, and the cycle fed itself. (4/21)
A healthcare scale-up dealing with unreliable external providers can apply the same thinking. The goal isn't to micromanage every provider. It's to build a system that makes providers reliable.
## The Core Principle
The best way to manage external service providers is to stop trying to control each one individually. Instead, build a platform with clear standards and interfaces. Let providers plug in on their own. The ecosystem starts managing itself. (5/21)
Ma didn't manage Alibaba's millions of sellers one by one. He built a platform with standards. Sellers plugged in. Those who performed well got more business. Those who didn't got churned out by the system. The platform did the heavy lifting.
For this healthcare team, the approach is the same. Stop chasing individual providers. Build infrastructure that makes reliability the path of least resistance.
## Four Steps to Apply the Platform Business Model (6/21)
### 1. Build an Integration Platform
Ma created an integration layer at Alibaba with clear standards and interfaces. Sellers could plug in on their own without custom hand-holding.
This team should do the same. The product owner should build a layer between the core product and the external providers. The core product should never depend on any single provider directly. That layer makes the whole system resilient. (7/21)
Last quarter, a product owner actually did this. It took about three weeks. The layer had three components.
First, an API gateway. One single entry point for all external integrations. The team can monitor everything in one place and see immediately what's working and what's not. (8/21)
Second, a standard data format. One schema that defines how data is structured. Every provider sends data the same way, so the team doesn't need custom parsers for each one. Adding a new provider becomes fast instead of painful.
Third, a provider interface specification. This is a contract that tells each provider exactly what they need to implement. Providers can do their own integration work. The team doesn't have to build custom bridges every time. (9/21)
After this platform was built, the team went from three weeks to three days to onboard a new provider. That flexibility saved the company thirty-one thousand dollars. Those were contracts they would have lost without it.
For a Crystal team of two to five, the integration platform should be a layer with at least three components. It should cut down onboarding time significantly. This belongs in the team's reflection workshop.
### 2. Create a Provider Scorecard (10/21)
Ma tracked provider performance at Alibaba. He could see who was good and who wasn't. Good providers got rewarded. Bad ones got replaced.
This team should build a provider scorecard and track three metrics for each external provider.
Reliability is the first one. Just uptime. What percentage of the time is the provider actually available. Give them a score from zero to one hundred. (11/21)
Response time is the second. How long does it take for a provider to respond to a request. Again, zero to one hundred.
Data quality is the third. What percentage of the data is correct. Zero to one hundred.
Combine all three into one total score and share it publicly. When providers can see their own scores, they compete on performance. That competition drives improvement.
Last quarter, one team actually did this. They tracked three providers. (12/21)
LabCorp scored a ninety-point-seven total. Quest Diagnostics came in at eighty-four. CVS Pharmacy led at ninety-four-point-seven.
When Quest saw their score publicly, they improved. That alone saved the company twenty-four thousand dollars in contracts that would have been lost.
For a Crystal team, the scorecard should be a dashboard tracking at least three metrics. Share it publicly. Use it as a communication tool.
### 3. Build a Self-Service Onboarding Process (13/21)
Ma built a self-service flow at Alibaba. Sellers could sign up without Ma's team doing the work for them each time.
This team should build a portal that lets external providers integrate themselves. No three-person team should be a bottleneck.
The portal needs four steps.
Registration. A form that collects the provider's name, type, and contact info. Nothing fancy.
API key generation. An automated process that generates a unique key so the provider can authenticate. (14/21)
Sandbox access. A test environment where the provider can try out their integration and verify it works before going live.
Certification. An automated test that confirms the provider meets the interface specification. Pass the test, get approved, go live. (15/21)
Last quarter, a team built one of these portals. Took two weeks. Two new providers used it and integrated themselves. The three-person team never had to touch the integrations. They delivered two features on time because of it. That saved nineteen thousand dollars.
For a Crystal team, the portal should have at least four steps and let providers integrate themselves. This is a delivery mechanism for the team.
### 4. Run a Feedback Loop Every Iteration (16/21)
Ma ran regular reviews at Alibaba. He looked at provider performance data and made decisions. Keep the good ones, replace the bad ones.
This team should run a fifteen-minute feedback loop at the end of every iteration. Three parts.
First, five minutes to review the scorecard trends. Are providers improving or declining?
Second, five minutes to decide. Keep providers on an upward trend. Replace the ones that are declining. Add new ones if there's a capability gap. (17/21)
Third, five minutes to update the integration platform if needed.
Last iteration, a team ran this loop and saw Quest Diagnostics had improved their response time score from seventy-nine to eighty-five. They kept Quest, saved themselves the time of finding a replacement, and shipped a feature on time. That saved thirteen thousand dollars.
For a Crystal team, this should happen every iteration. Review the trends, make a decision, move on. Make it part of the reflection workshop. (18/21)
## Managing the Ecosystem, Not the Pieces
Jack Ma didn't build Alibaba by micromanaging every seller. He built a platform with clear standards. He tracked performance and shared scores publicly. He let sellers onboard themselves. He reviewed data regularly and made decisions based on trends.
It's the same playbook a small healthcare team running Crystal can use. Build the integration platform. Create the scorecard. Open up self-service onboarding. Run the feedback loop. (19/21)
A team of two to five can put this in place within a few iterations. The eighty-seven thousand dollars the company lost last quarter doesn't have to happen again. The fix isn't more headcount. It's better infrastructure.
Start with the integration platform this iteration. Build the scorecard and onboarding process next. Run the feedback loop at the end. The external providers won't become magic, but the system will make reliability the default. (20/21)
agile