Skip to content
Saint HighLoad++ · September 22, 2022

Channel. Product. Platform.

How the approach to building Tinkoff's mobile bank evolved

/ Channel. Product. Platform · Saint HighLoad++ 2022

Slide contents

  1. 1. Channel. Product. Platform.

    How the approach to building Tinkoff's mobile bank evolved

  2. 2. Alexander Polomodov

    CTO of the mobile bank, online acquisition and more

    I curate our System Design and Troubleshooting interviews

  3. 3. For anyone who has asked these questions

    Who this talk is for

    Why scale development at all?

    How do you do it evolutionarily, without dropping business work?

    Which target topology should your teams aim for?

    How do you balance platform and product development?

  4. 4. Sixteen years in six milestones

    The mobile bank as channel, product and platform

    2006 — the company is founded, 2008 — internet bank

    2011 — the mobile bank appears

    2015 — the mobile bank is the top product

    2019 — a platform, 2022 — what comes next

  5. 5. 01. The mobile bank as a channel

    2006–2011: from postal mail to the first app

  6. 6. A branchless bank on the Capital One model

    How it all started (2006)

    Credit cards only at the start

    Direct sales by postal mail

    Customer contact through call centres

  7. 7. The first version of the app was built by contractors

    The mobile bank appears as an additional channel

    Basic services were picked for the launch

    The first version was built to order by contractors

    Requirements were written screen by screen

    Planning ran as app versions with scope fixed up front

  8. 8. 02. The mobile bank as a product

    2015: development moved in-house

  9. 9. One team of twenty for three customers

    The mobile bank as a product

    Customers: retail bank, business bank, insurance

    One shared IT team of ~20 with shared prioritisation

    Large releases roughly once a quarter

    A monolithic app and the Mother API

  10. 10. This worked — and here is why

    While there are few products and the team is small

    There were a few main products taking the resources

    The team was small — around 50 people by 2019

    Priorities could still be managed by hand

    The focus was on extending functionality

  11. 11. 03. The mobile bank as a platform

    2019: verticals, teams and architecture

  12. 12. A monoline bank grew into a set of verticals

    The group's growth and the IT landscape

    2006–2019: from monoline bank to SuperApp

    Verticals: retail, SME, investments, insurance

    …the mobile operator and non-financial services

    And platforms: data, ML, origination, infrastructure

  13. 13. Several apps now, and still one core team

    Different apps and a shared core team

    SuperApp and the business bank app

    The investments and mobile operator apps

    One mobile core team serves them all

    And that team becomes the platform team

  14. 14. Six business teams and six platform teams

    The new team structure

    Banking products, Payments, Insurance products

    SME products, Invest products, NonFinancial Services

    Platform teams: Release, Platform, Design

    …Excellence, Performance, Reliability

  15. 15. Change the teams and you change the architecture

    Three states of the same diagram

    Organizations design systems that copy their communication

    Before: one shared team over a layered architecture

    After: autonomous teams with their own domains

    Target: their own modules and a modular architecture

  16. 16. Releases are tied to dates, not to scope

    Release trains and the engineering culture

    Teams have to be able to work autonomously

    Delivery managers roll out the Kanban process

    Automated testing is mandatory for new features

    Fitness functions run through Danger in CI/CD

  17. 17. Large releases every four weeks

    Interim results

    Plus smaller releases in between

    Processes documented, owned by the team and the DM

    Platform teams rolled out modularisation

    Set up mobile SRE and optimised performance

  18. 18. 04. So what next?

    2022: end-to-end delivery and governance

  19. 19. Three triggers that it is time to change the teams

    The business wants better end-to-end value delivery

    The software got too big for one team

    Delivery slowed down as a sustained trend

    Business services rest on scattered lower-level ones

    Each trigger has its own set of symptoms

  20. 20. The end-to-end process runs through a delivery manager

    Teams spanning SuperApp, APIs and Backends

    End-to-end metrics: TTM, lead time, cycle time

    Work on both downstream and upstream

    From «fetch your own data» APIs to a mobile BFF

    And towards read models built on an event stream

  21. 21. The rules check themselves automatically

    Platform technical governance

    What a team is and what attributes it must have

    The development, deployment and operations processes

    The rules show how healthy the teams are

    Metrics: productivity, quality, flow

  22. 22. 05. Wrap-up

    The questions I promised to answer

  23. 23. From a startup to a SuperApp team of ~400 engineers

    The questions I promised to answer

    At the start there was no in-house mobile development at all

    The app became an advantage — development moved in-house

    Products multiplied — a platform became necessary

    Cross-cutting teams demanded technical governance

    Take this experience and map it onto your own situation

    Why scale, how to do it evolutionarily, which topology, and how to balance platform against product

  24. 24. Materials for further study

    From the talk's closing slide

    The written version of the talk — bit.ly/mobileBankEvolution

    All recommended materials are listed at the end of the article

  25. 25. Thank you!

    polomodov.tech

    All slides and links are in the Telegram channel

    Alexander Polomodov, Technical Director & Fellow, T-Technologies

    @book_cube