WNWEBSNEXUS

Case study — Telehealth

PillTabs

1,000 patientswas the breaking point

Telehealth prescriptions for men’s health — intake, clinical review, subscription billing and discreet recurring fulfilment in one flow. Here is what happened when the platform that proved the model had to carry it.

Sector: TelehealthEngagement: Platform re-architecture & scalingpilltabs.com ↗
0Patients before
0+Patients now
0Downtime migrating
0+Patients served
0xCapacity increase
0Downtime during migration
0Dropped renewals

The ceiling

Built to prove it,not carry it

PillTabs was never short on demand. The problem was a platform built to prove the model, not carry it at scale.

Around 1,000 active patients, the system started to fail — and every new patient made the platform slower for everyone else.

  • 01Database queries slowing to a crawl as records accumulated
  • 02Concurrent traffic spikes causing timeouts and crashes
  • 03Subscription renewals running as one batch that grew — and eventually failed
  • 04Clinician review queues lagging behind real time

That is not a demand problem. It is an architecture ceiling — and for a subscription health service, hitting it is not just churn. It is a missed refill.

What we built

Re-architected,not patched

We removed the ceiling rather than raising it.

01

Database layer

Rebuilt the schema and indexing strategy around the platform’s actual access patterns, with connection pooling and read replicas. Queries that once degraded linearly with patient count now run in near-constant time.

02

Caching & sessions

A caching layer for the hot paths — product catalogue, plan pricing, patient session state — cutting redundant database round-trips on every page load.

03

Async job processing

Subscription renewals, payment retries, notifications and fulfilment handoffs moved onto a proper queue with retry logic. Renewal cycles that once ran as one fragile batch now process in parallel, with failures isolated instead of cascading.

04

Horizontal scalability

Containerised behind a load balancer with auto-scaling, so traffic spikes add capacity instead of causing outages — instrumented end to end so degradation is caught before patients notice.

The result

15,000+patients

Fifteen times the patient volume the original platform could carry, migrated without a single dropped renewal.

DatabaseConstant-time queries
RenewalsIsolated retry, no cascades
ScalingAutomatic under load
MigrationZero downtime

What now

Still runningon a proof of concept?

If your platform proved the model but is starting to buckle under real usage, that is exactly the problem we solve. We build the systems your growth depends on.

Talk to us about your system