Imagine ordering food.

You search. Pick a restaurant. Add an item. Pay. Track the delivery. Get a notification.

You see one app.

Behind it, different parts of the product may have different jobs.

FOOD DELIVERY
one customer experience
Usersaccounts · addresses
Restaurantsmenus · availability
Orderscart · order status
Paymentscharges · refunds
Deliveryassignment · tracking
Notificationspush · email · SMS

So what is a microservice?

01
A small, focused software service

responsible for a specific business capability — such as orders, payments or notifications.

The services communicate with each other through defined interfaces, often APIs or messaging systems.

One big application vs many smaller services

MONOLITH
ONE LARGE
APPLICATION
OrdersPaymentsDeliveryNotifications
MICROSERVICES
OrdersPaymentsDeliveryNotifications

Each service can have its own code, ownership and deployment.

Why make the change?As products and teams grow, smaller services can make ownership, scaling and deployment more independent.

But microservices don't remove complexity.

MONOLITHComplexity inside one large application
MICROSERVICESSmaller services + more communication between them

The complexity hasn't disappeared. Some of it has moved into the relationships between services.

Here's where the PM comes in.

FEATURE REQUEST

“Add a new payment option.”

ProductPaymentOrderNotification

A seemingly small feature can cross several services and teams.

WHO?Which teams own the services involved?
WHAT?Which APIs or interfaces change?
WHEN?What has to happen first? What can run in parallel?
IF?What happens if one service succeeds and another fails?

One failure can cross the system.

Customer paysPayment succeedsOrder creation fails
Customer chargedbut no confirmed order

Now the delivery questions become important: retry? refund? reconcile? notify the customer? monitor the failure?

APIs become contracts between services.

ORDER SERVICE“Please charge ₹450”
→ API →
PAYMENT SERVICE“Payment successful”
Useful PM question

“Is the API change compatible with the services that already depend on it?”

What you actually need to remember

01One app can be many services.
02Each service usually owns a specific capability.
03Services have to communicate reliably.
04More services also mean more dependencies and failure points.
05For a PM, architecture becomes a delivery map.

Don't ask only: “Is the service done?”

Service completeEnd-to-end experience works

That's the shift: understanding the architecture well enough to see ownership, dependencies, sequencing, risks and the conditions for the whole feature to work.

Next time someone says “it's just a microservice change”...

you'll know there may be a whole network behind those few words.

Back to Writing →