Squad all the things ! - MixIT 2024

Squad all the things!! Retour d'xp no bullshit sur la mise en oeuvre d'équipes pluridisciplinaires.

En gagnant en taille, nos équipes d’ingénierie logicielle rencontrent souvent des difficultés pour se distribuer efficacement le travail entre elles. Grâce ou à cause de Spotify, les organisations à base d’équipes pluridisciplinaires se sont majoritairement démocratisées pour adresser ce problème. Il n’en reste pas moins que basculer vers ce genre de modèle n’est pas si simple et ne s’opère rarement en une nuit. Cela tire en effet bon nombre de sujets épineux autour de l’engineering management, du leadership sur les stack technologiques, la juste répartition des périmètres fonctionnels ou encore du juste équilibre entre priorités fonctionnelles et techniques.

Je vous propose dans ce talk de partager mon retour d’expériences sur les organisations des équipes engineering dont j’ai eu la responsabilité ou que j’ai côtoyé (Batch, Leboncoin, Videdressing…) . De façon pragmatique et sans langue de bois, nous évoquerons ce qui a fonctionné mais aussi des points sur lesquels j’ai échoué pour vous !

Transcript

1. SQUAD ALL THE THINGS! HERVÉ LOURDIN, CTO AT BATCH 25 AVRIL 2024, MIXIT 1 — MixIT 2024 - Hervé Lourdin - Batch
2. SQUAD ALL THE THINGS! HERVÉ LOURDIN, CTO AT BATCH 25 AVRIL 2024, MIXIT 2 — MixIT 2024 - Hervé Lourdin - Batch
3. WHO AM I? > @HerveLourdin > Sun Microsystems > OCTO Technology, partner > Videdressing, CTO & co-CEO > Leboncoin, eng. director, Tribe Transaction > Today: CTO 3 — MixIT 2024 - Hervé Lourdin - Batch
4. WHO'S ? 4 — MixIT 2024 - Hervé Lourdin - Batch
5. THIS TALK'S PROMISE: "NO BS RETURN ON EXPERIENCE ABOUT X-FUNCTIONAL TEAMS" 5 — MixIT 2024 - Hervé Lourdin - Batch
6. 6 — MixIT 2024 - Hervé Lourdin - Batch
7. ANYONE HEARD ABOUT... > Conway's law > Spotify model > Team topologies 7 — MixIT 2024 - Hervé Lourdin - Batch
8. WHY CROSS-FUNCTIONAL TEAMS ? 8 — MixIT 2024 - Hervé Lourdin - Batch
9. SYMPTOM: SKILLSET DRIVEN ORGANISATION SHOW EVENTUALLY BOTTLENECKS IN WORKLOAD DISTRIBUTION. 9 — MixIT 2024 - Hervé Lourdin - Batch
10. MY CONTEXT AT THE TIME: > 4 countries > ~80 people (PO + UX + engineers + QA) > 1st agile experience for most of them 10 — MixIT 2024 - Hervé Lourdin - Batch
11. 2010, NOT THAT MUCH RESOURCES 11 — MixIT 2024 - Hervé Lourdin - Batch
12. FEATURE TEAM 1 > long-lived—the team stays together so that they can ‘jell’ for higher performance > cross-functional and cross-component > ideally, co-located > work on a complete customer-centric feature, across all components and disciplines > composed of generalizing specialists > in Scrum, typically 7 ± 2 people 1 extrait du "feature team primer" de C. Larman & B. Vode de 2010. 12 — MixIT 2024 - Hervé Lourdin - Batch
13. 10 YEARS AGO, THE SPOTIFY MODEL 13 — MixIT 2024 - Hervé Lourdin - Batch
14. FEATURE TEAM = SQUAD = CREW = <YOUR FANCY CONCEPT HERE> 14 — MixIT 2024 - Hervé Lourdin - Batch
15. WHEN DOES IT WORTH THE JUMP? FROM MY EXPERIENCE: 25 15 — MixIT 2024 - Hervé Lourdin - Batch
16. WHAT IS THE MAX SIZE? > No idea, > I've seen 600 people w. 55 squads, 6 tribes. It was ok. > Big organisations raise new questions, new limits 16 — MixIT 2024 - Hervé Lourdin - Batch
17. SHOULD EVERY ENGINEERING ORGANISATION ADOPT THIS MODEL? NO 17 — MixIT 2024 - Hervé Lourdin - Batch
18. SOME LARGE ENGINEERING ORGANISATION DO NOT USE SUCH MODELS. THEY USUALLY HAVE A STRONG CONTROL OF THEIR DEVELOPMENT FLOW. 18 — MixIT 2024 - Hervé Lourdin - Batch
19. IF I DECIDE TO DO THIS SQUAD THING: HOW DO I SPLIT MY ORG. INTO TEAMS? 19 — MixIT 2024 - Hervé Lourdin - Batch
20. THE ORGANISATION SPLIT EQUATION2 2 first created when I was at leboncoin 20 — MixIT 2024 - Hervé Lourdin - Batch
21. 21 — MixIT 2024 - Hervé Lourdin - Batch
22. ⚠ DEFINE DAY ONE, A CLEAR SCOPE OF RESPONSABILITY PER SQUAD > I failed for you > It hurts > Even if the technical component split is not there, it will rise progressively (thanks Conway ) 22 — MixIT 2024 - Hervé Lourdin - Batch
23. GROW AND SPLIT > Make sure the split is clearly done at least on a functional or a technical axis > If both, congrats it's you lucky day > reading tips: Dynamic Reteaming (Heidi Helfand) 23 — MixIT 2024 - Hervé Lourdin - Batch
24. REBALANCING THE ORGANISATION > an org is only good for a given context at a given date > get ready, your org will probably not last more than 18 to 24 months > Rebalancing, you will not fear. 24 — MixIT 2024 - Hervé Lourdin - Batch
25. KILLING A FEATURE TEAM? > when 80% of what we wanted to be done is done > the team will always find work to do > keep the team, change the scope. 25 — MixIT 2024 - Hervé Lourdin - Batch
26. CHANGE MANAGEMENT > involve everyone day one > even if "you know the endgame" > spend twice the time if required to share the problem you are trying to solve 26 — MixIT 2024 - Hervé Lourdin - Batch
27. THESE ORGANISATIONS HAVE CONSEQUENCES 27 — MixIT 2024 - Hervé Lourdin - Batch
28. CONSEQUENCES ON THE STACK AND ON TECHNICAL DEBT? 28 — MixIT 2024 - Hervé Lourdin - Batch
29. X-FUNCTIONAL TEAMS ARE FEATURE DRIVEN BY DESIGN > A tech driven organisational counterweight is necessary > Guilds or communities of practices will help to > keep control of technical debt > create technical synergies 29 — MixIT 2024 - Hervé Lourdin - Batch
30. CONSEQUENCES ON PEOPLE AND MANAGEMENT 30 — MixIT 2024 - Hervé Lourdin - Batch
31. ENGINEERING MANAGEMENT 31 — MixIT 2024 - Hervé Lourdin - Batch
32. ENGINEERING MANAGEMENT > Tech leaders are no managers > Disclaimer: I am not a chapter manager believer. (5 managers to solve a problem. Really?) > Conclusion: Go for a X-functional EM 32 — MixIT 2024 - Hervé Lourdin - Batch
33. PORTRAIT OF A X-FUNCTIONAL EM > Tech: need to understand, not being an expert > People: taking care > Delivery: lead and follow development w. the team > Product: product empathy 33 — MixIT 2024 - Hervé Lourdin - Batch
34. CAREER PATH STRONGLY RECOMMANDED 34 — MixIT 2024 - Hervé Lourdin - Batch
35. 35 — MixIT 2024 - Hervé Lourdin - Batch
36. STAFF ENGINEER > x-squads key actors > frame x-squads projects > contribute to define tech strategy > "act as a multiplier" on teams 36 — MixIT 2024 - Hervé Lourdin - Batch
37. SENIOR EM 37 — MixIT 2024 - Hervé Lourdin - Batch
38. SENIOR EM & THE CREW > Manager of managers > Downside: longer chain of command > Say no until too too many direct reports! > You can delegate strategic topics to her/him > will raise the question of product counterpart... 38 — MixIT 2024 - Hervé Lourdin - Batch
39. ORGANISATION AT SCALE 39 — MixIT 2024 - Hervé Lourdin - Batch
40. PRODUCT SQUAD & PLATFORM SQUAD (PLEASE READ TEAM TOPOLOGIES) 40 — MixIT 2024 - Hervé Lourdin - Batch
41. CREATING PLATFORM TEAMS RAISE GOOD QUESTIONS > which degree of liberty require product squads > what should be defined as an expertise to commoditize > info: you can split a platform team into several smaller ones 41 — MixIT 2024 - Hervé Lourdin - Batch
42. CONSEQUENCES ON ROADMAPPING 42 — MixIT 2024 - Hervé Lourdin - Batch
43. > Product Squad roadmaps are pretty easy > ... except for x-squads projects > such project require product & tech framing: > shared product discovery > tech exploration of the project to identify architecture choices and make rough estimate of the complexity 43 — MixIT 2024 - Hervé Lourdin - Batch
44. THE END 44 — MixIT 2024 - Hervé Lourdin - Batch
45. > each context is different > there is no unique perfect organisation > each scaling stage is different > do not rollout everything day one > wait for the problem to emerge > build the solution with the team 45 — MixIT 2024 - Hervé Lourdin - Batch
46. CREATE AND FEED A CULTURE OF LEARNING ORGANISATION 46 — MixIT 2024 - Hervé Lourdin - Batch
47. Q & A MERCI ! 47 — MixIT 2024 - Hervé Lourdin - Batch
48. SOURCES : > Spotify model by Henrick Kniberg : https:// storage.googleapis.com/production-eng/1/2020/05/spotify- engineering-culture-part1.jpeg > Grow and split pattern of Heidi Helfand : https:// www.oreilly.com/api/v2/epubs/9781492061281/files/assets/ drt2_0601.png > Hold the door. well... you know... : https://media.tenor.com/ 9hMxUx7qlUsAAAAM/hodor-game-of-thrones.gif 48 — MixIT 2024 - Hervé Lourdin - Batch