Share
Embed
Strategy workshop from LDX3 Director+
This is a 90 minute workshop aimed at technology leaders who want to develop their engineering strategy using Richard Rumelt's strategy kernel
Alice Bartlett
June 18, 2025
Transcript
101 slides
-
Alice Bartlett, Head of Engineering @ Rightmove June 2026 For
Director+ De fi ning your engineering strategy -
I’m an head of engineering at Rightmove I work with
Rightmove’s Product and Engineering leadership to set and execute the technical strategy. I’m Alice Bartlett -
Icebreaker Overview: what is a strategy? Building your own strategy:
- Diagnosis - Principles - Actions Agenda -
Activity: Icebreaker - intros and experience Share with the people
near you: 1. Who are you? 2. When it comes to strategy setting are you: • A tourist - new to it, here to explore • A regular visitor - some experience, still figuring things out • A tour guide - done it a few times, could probably show others around? 5 minutes -
Activity: Icebreaker - intros and experience Share with the people
near you: 1. Who are you? 2. When it comes to strategy setting are you: • A tourist - new to it, here to explore • A regular visitor - some experience, still figuring things out • A tour guide - done it a few times, could probably show others around? 5 minutes time’s up! … minute remaining minutes remaining 1 2 3 4 5 - -
Icebreaker Overview: what is a strategy? Building your own strategy:
- Diagnosis - Principles - Actions Agenda -
“Good strategy is coherent action backed up by an argument.
An effective mixture of thought and action, with a basic underlying structure that I call the kernel” Richard Rumelt, Good Strategy / Bad Strategy -
Diagnosis A description of the problem to solve. This is
the bedrock of your strategy and should encapsulate the challenge precisely. -
Guiding Principles Policies that shape every decision. These are the
active constraints that you have to work within. They should guide your approach to how to solve the diagnosis. -
Coherent Actions Actions you will actually take to solve the
diagnosed problem. These should be concrete, have owners, timeframes and people committed to them. -
Diagnosis A description of the problem to solve. This is
the bedrock of your strategy and should encapsulate the challenge precisely. Guiding Principles Policies that shape every decision. These are the active constraints that you have to work within. They should guide your approach to how to solve the diagnosis. Coherent Actions Actions you will actually take to solve the diagnosed problem. These should be concrete, have owners, timeframes and people committed to them. Rumelt’s Strategy Kernel -
These are (some) universal forces Technical debt System reliability Legacy
architecture Scalability constraints Delivery throughput Organisational structure Cognitive load Security and compliance surface […etc…] -
The work of the tech strategy is to find out
which universal forces are going to mess up your business strategy -
AcmeSaSS Example • B2B SaaS - project management tool for
creative agencies • 8 years old, 2,000 paying customers • Mostly small and mid-sized agencies, a few larger ones • Engineering team of 40 • Monolith with a React front-end, deploys twice a week • Good uptime but occasional slow page loads on larger accounts • Two competitors: one winning on enterprise features and compliance, one winning on speed of new feature delivery -
Strategy A: Move upmarket to enterprise agencies and in-house creative
teams • security and compliance surface • system reliability • scalability constraints on large accounts Engineering response: -
Strategy B: Win the small/medium business market through feature velocity
• technical debt • delivery throughput • Conway's Law Engineering response: -
AcmeSaSS Example Same company, but different company level strategy means
a different engineering strategy is required. -
This is so obvious but… Your technology strategy should be
enabling the business to achieve its goals -
Icebreaker Overview: what is a strategy? Building your own strategy:
- Diagnosis - Principles - Actions Agenda -
• 200 engineers • Subscriptions model • B2B and B2C
subscribers • 1 apps team, 12 web teams • Web app with native wrappers for Android and iOS • Features often appear on web first, then on the app • App is a 15 years old codebase with lots of tech debt and legacy • Only the apps team can work on the app • App is where the highest B2C subscriber engagement is - with B2B engagement also climbing. AcmeNews Factsheet -
Everyone agrees that the mobile apps of AcmeNews are where
the business’s future growth comes from, they are where their most engaged customers are, but the engineering team can’t ship work in them quickly enough to keep up with customer expectations. Our starting point: -
Rough problem statement: Delivering features to the app is really
slow. AcmeNews’ strategy is to do more with the app, but the primary constraint currently is a bottleneck getting features into the app -
The diagnosis identi fi es the core challenge facing the
organisation. -
Rough problem statement: Delivering features to the app is really
slow. AcmeNews’ strategy is to do more with the app, but the primary constraint currently is a bottleneck getting features into the app -
Why is delivery in the mobile apps slow? Because most
changes require support from the central Apps team rather than being made independently by the teams who own the features. Why #1 -
Why do changes require support from the central Apps team?
Because contributing teams don't have enough understanding of the app to make changes safely on their own. Why #2 -
Why don't contributing teams understand the app well enough? Because
the app is highly complex — the architecture, the patterns, the interdependencies — and that complexity has grown over time without being addressed. Why #3 -
Why has the complexity grown without being addressed? Because simplifying
the app has never been prioritised alongside feature delivery. Why #4 -
Why has simplification never been prioritised? Because nothing (incidents, external
pressure, strategic requirement) has forced the cost of the complexity to be high enough to act on. Why #5 -
WHAT IS GOING ON HERE? Delivery is slow on the
app because changes require deep support from the central apps team. The mobile app is very complex, and hard for contributing teams to understand. Simplifying it has never been prioritised. -
What are the universal forces that we might want to
focus on? • Conway’s Law • Cognitive load • Delivery throughput • Technical debt and legacy architecture -
Is this a good diagnosis yet? Delivery is slow on
the app because changes require deep support from the central apps team. The mobile app is very complex, and hard for contributing teams to understand. Simplifying it has never been prioritised. -
Properties of a good diagnosis: 1. It identifies the critical
issue 2. It explains how we got here 3. It answers “why” and “why now?” 4. It creates direction 5. It avoids wishful thinking -
1. It identi fi es the critical issue A good
diagnosis should clearly point to the single most important constraint, blocker or threat. -
1. It identi fi es the critical issue Delivery is
slow on the app because changes require deep support from the central Apps team. The mobile app is very complex, and hard for contributing teams to understand. Simplifying it has never been prioritised. Delivery is slow on the app because changes require deep support from the central Apps team. ++This dependency exists because the app is structured in a way that prevents autonomous contributions. Simplifying it has never been prioritised. -
2. It explains how we got here A good diagnosis
should show how the problem came to be, and why -
2. It explains how we got here Delivery is slow
on the app because changes require deep support from the central Apps team. This dependency exists because the app is structured in a way that prevents autonomous contributions. Simplifying it has never been prioritised. ++…because the maintaining team had enough internal knowledge to work efficiently despite it. -
Your diagnosis should link to the business impact of the
problem, and it should show why it is important to fix the problem now by showing what has changed 3. It answers “why” and “why now?” -
3. It answers “why” and “why now?” ++ This structure
no longer fits our growing organisation with multiple contributing teams. ++ Failure to address this has resulted in delivery bottlenecks and coordination overheads, limiting our ability to ship features rapidly, respond to user needs, and compete on product experience. Delivery is slow on the app because changes require deep support from the central Apps team. This dependency exists because the app is structured in a way that prevents autonomous contributions. Simplifying it has never been prioritised because the maintaining team had enough internal knowledge to work efficiently despite it. -
4. It creates direction A diagnosis that creates direction makes
some solutions obvious and rules others out. -
4. It creates direction Delivery is slow on the app
because changes require deep support from the central Apps team. This dependency exists because the app is structured in a way that prevents autonomous contributions. Simplifying it has never been prioritised because the maintaining team had enough internal knowledge to work ef fi ciently despite it…. IN SCOPE: • Restructure the app so multiple teams can contribute • Invest in platform work that enables autonomous contribution • Modularise the codebase along team boundaries • Change the ownership model OUT OF SCOPE: • Hire more people into the Apps team • Rewrite the app from scratch • Process improvements without architectural change • Speed up the Apps team -
5. It avoids wishful thinking A good diagnosis should be
grounded in reality, not optimism. -
5. It avoids wishful thinking “We need to modernise our
mobile app architecture to unlock innovation and accelerate our path to becoming a mobile-first organisation.” Doesn’t name what’s actually wrong “Modernise” could mean anything “Unlock innovation” is a vague Not falsifiable Could be the diagnosis for any company -
5. It avoids wishful thinking Delivery is slow on the
app because changes require deep support from the central Apps team. This dependency exists because the app is structured in a way that prevents autonomous contributions. Simplifying it has never been prioritised because the maintaining team had enough internal knowledge to work efficiently despite it. This structure no longer fits our growing organisation with multiple contributing teams. Failure to address this has resulted in delivery bottlenecks and coordination overheads, limiting our ability to ship features rapidly, respond to user needs, and compete on product experience. -
Your diagnosis should be speci fi c enough to guide
actions and be proven wrong -
A fuzzy unfalsi fi able diagnosis will lead to a
bad strategy “We need to modernise our mobile app architecture to unlock innovation and accelerate our path to becoming a mobile-first organisation.” -
Instead “Delivery is slow on the app because changes require
deep support from the central Apps team. This dependency exists because the app is structured in a way that prevents autonomous contributions. Simplifying it has never been prioritised because the maintaining team had enough internal knowledge to work efficiently despite it. This structure no longer fits our growing organisation with multiple contributing teams. Failure to address this has resulted in delivery bottlenecks and coordination overheads, limiting our ability to ship features rapidly, respond to user needs, and compete on product experience.” -
Most of this is falsi fi able “Delivery is slow
on the app because changes require deep support from the central Apps team. This dependency exists because the app is structured in a way that prevents autonomous contributions. Simplifying it has never been prioritised because the maintaining team had enough internal knowledge to work efficiently despite it. This structure no longer fits our growing organisation with multiple contributing teams. Failure to address this has resulted in delivery bottlenecks and coordination overheads, limiting our ability to ship features rapidly, respond to user needs, and compete on product experience.” -
Activity: Sketch and share your diagnosis 15 minutes Task •
(8 mins solo) Sketch a diagnosis for your problem • (7 minutes 2-3 people) Share your diagnosis sketch and what you still need to improve about it. -
Activity: Sketch and share your diagnosis 15 minutes Task •
(8 mins solo) Sketch a diagnosis for your problem • (7 minutes 2-3 people) Share your diagnosis sketch and what you still need to improve about it. time’s up! … minute remaining minutes remaining 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 - -
Icebreaker Overview: what is a strategy? Building your own strategy:
- Diagnosis - Principles - Actions Agenda -
Guiding principles de fi ne an overall approach chosen to
cope with or overcome the obstacles identi fi ed in the diagnosis. -
Tips for good guiding principles 1. Your principles should stem
from your diagnosis 2. They should clarify what not to do 3. They should be concrete enough to help someone make a decision 4. They should be consistent with one another -
Guiding principles 1. Every part of the codebase has a
named owning team 2. Contributing teams should be able to merge to their part without central review 3. The Apps team exists to enable contributors, not to be a gatekeeper Diagnosis Contributing teams can’t make app changes independently because the codebase has no clear module ownership — every change requires the central Apps team to get involved. 1. Your principles should stem from your diagnosis -
Help people say no to attractive but misaligned options 2.
Guiding principles should clarify what not to do -
We will improve the app architecture while continuing to support
contributing teams with new feature delivery No Yes We will not take on new feature work that increases central team dependency, even when the feature is valuable. 2. Guiding principles should clarify what not to do -
3. Guiding principles should be concrete enough to help someone
make a decision If your principles are too abstract people won’t know how to apply them -
No Yes Focus on improving our CI/CD pipeline. No Yes
Prioritise improvements to our CI/CD pipeline, especially where workflows require frequent manual intervention, or block independent team delivery. 3. Guiding principles should be concrete enough to help someone make a decision -
4. Guiding principles should be consistent with one another Your
principles should all be moving in the same direction and it should be possible to do one without undoing another at the same time. -
4. Guiding principles should be consistent with one another No
Yes 1. Every app release must maintain 100% stability and zero regressions. 2. Prioritise rapid iteration and experimentation in the app. No Yes 1. Every app release must maintain 100% stability and zero regressions. 2. Invest in robust automated testing and release validation pipelines. -
Tips for good guiding principles (recap) 1. Your principles should
stem from your diagnosis 2. They should clarify what not to do 3. They should be concrete enough to help someone make a decision 4. They should be consistent with one another -
Solo activity: Write some guiding principles for your diagnosis 7
minutes Examples 1. Every part of the codebase has a named owning team 2.Contributing teams should be able to merge to their part without central review 3.The Apps team exists to enable contributors, not to be a gatekeeper -
Solo activity: Write some guiding principles for your diagnosis 7
minutes Examples 1. Every part of the codebase has a named owning team 2.Contributing teams should be able to merge to their part without central review 3.The Apps team exists to enable contributors, not to be a gatekeeper time’s up! … minute remaining minutes remaining 1 2 3 4 5 6 7 - -
Icebreaker Overview: what is a strategy? Building your own strategy:
- Diagnosis - Principles - Actions Agenda -
Your coherent actions are a set of speci fi c
things you are going to do to carry out your strategy -
The key word here is “coherent” They should reinforce each
other, not be an unrelated list They should centre people/money on the same focal point They should be derived from the guiding policy. If actions “make sense on their own” they are just good ideas - not strategically coherent. -
Common failure modes A list of initiatives that each address
a different problem, and calling that a strategy. Coherent actions address one diagnosis, through one set of principles. -
Diagnosis Shipping features to the App is slow because a
single team owns the whole codebase [etc] -
Diagnosis Shipping features to the App is slow because a
single team owns the whole codebase [etc] Guiding Principle #1 Decentralise App ownership by assigning domain aligned teams clear responsibility for specific modules to reduce delivery bottlenecks -
Diagnosis Shipping features to the App is slow because a
single team owns the whole codebase [etc] Guiding Principle #1 Coherent Actions Decentralise App ownership by assigning domain aligned teams clear responsibility for specific modules to reduce delivery bottlenecks • Map the monolith to domains and assign named owning teams • Establish a platform team to abstract shared infrastructure, so domain teams can contribute without deep App knowledge • Establish an enablement team to build the capability of domain teams to own and ship independently -
Coherent Actions Good because: ✓They’re aligned to the guiding principle
✓They’re specific and actionable ✓They work together ✓They address root causes, not symptoms ✓They show strategic investment ✓Are achievable in a meaningful timeframe • Map the monolith to domains and assign named owning teams • Establish a platform team to abstract shared infrastructure, so domain teams can contribute without deep App knowledge • Establish an enablement team to build the capability of domain teams to own and ship independently -
Proximate Objectives This is the nearest thing you can actually
accomplish that meaningfully advances your strategy. Helps your teams focus on something concrete and achievable, rather than struggling to see how they can contribute to a wider strategic objective. Needs to be at the edge of your current capabilities. -
A Proximate Objective could be… The homepage team ships a
feature end-to-end without touching the central team's backlog. It’s proximate because it’s the first meaningful signal that the strategy is progressing. At the current point in time, this proximate goal is simply not possible. Once you have achieved your proximate objective - set the next one! -
Pairs activity: start sketching out your coherent actions 7 minutes
Guiding Principle 1 Example actions: • Map the monolith to domains and assign named owning teams • Establish a platform team to abstract shared infrastructure, so domain teams can contribute without deep App knowledge • Establish an enablement team to build the capability of domain teams to own and ship independently -
Pairs activity: start sketching out your coherent actions 7 minutes
Guiding Principle 1 time’s up! … minute remaining minutes remaining 1 2 3 4 5 6 7 Example actions: • Map the monolith to domains and assign named owning teams • Establish a platform team to abstract shared infrastructure, so domain teams can contribute without deep App knowledge • Establish an enablement team to build the capability of domain teams to own and ship independently - -
Diagnosis A description of the problem to solve. This is
the bedrock of your strategy and should encapsulate the challenge precisely. Guiding Principles Policies that shape every decision. These are the active constraints that you have to work within. They should guide your approach to how to solve the diagnosis. Coherent Actions Actions you will actually take to solve the diagnosed problem. These should be concrete, have owners, timeframes and people committed to them. Rumelt’s Strategy Kernel -
Table Activity: Share your kernel 15 minutes • For the
whole table, take it in turns to run through your kernel sketch. • Do not give feedback, just listen and think about what you can learn from how the rest of your table has approached their challenge. -
Table Activity: Share your kernel 15 minutes • For the
whole table, take it in turns to run through your kernel sketch. • Do not give feedback - your job is to think about what you can learn from them, not help them improve. time’s up! … minute remaining minutes remaining 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 - -
LLMs for strategists You cannot outsource the thinking. You cannot
outsource the listening. You cannot outsource the relationship building. But you can use LLMs as thinking partner. And you can use them to challenge your diagnosis, principles and actions. -
Writing a strategy is about taking a clear eyed look
at your problems and cutting through the noise enough to fix them -
“A strategy that makes you comfortable is probably not a
real strategy.” Roger Martin — Playing to win -
Further reading on strategy • Good Strategy Bad Strategy: The
Difference and Why it Matters. Richard Rumelt • Peter M. Senge, The Fifth Discipline: The Art & Practice of The Learning Organization. • Playing to Win: Roger Martin