- Cross-Functional
- Posts
- Cross-Functional #180: The backdown
Cross-Functional #180: The backdown
Transformation theatre, DORA metrics, quantitative discovery, using LLMs to write tests and more.

The Main Thing: The backdown
The rollercoaster of PWA support in 2024 continues.
After a loud and resounding backlash against Apple's decision to remove PWA features from Safari they have backed down and confirmed that they will keep all existing features.
This is positive news for developers, companies and consumers but Safari still supports the least amount of features for web apps and given Apple's recent behaviour I suspect that they will not be adding anything new any time soon.
From one controversial corporate policy to another, what is your company's policy towards remote/hybrid working? |
This Weeks Updates
Enabling the Team
The Wrong Way to Use DORA Metrics by Nočnica Mellifera
Are you using DORA metrics as intended? Google says you might not be.
How Leaders Miscommunicate When Tensions Are High by Liz Fosslien
When tensions run high, miscommunication is far more likely to occur. Liz shares five common errors and what to do instead.
Product Direction
Transformation Theater by Lea Hickman and Marty Cagan
Lea shares the different types of fake transformation that she has encountered including title rebrands, faux discovery, fake agile, strategy theatre and more.
How to Gain an Edge as a B2B PM by Alexis And Adrienne
Make it easy for customers to give you direct feedback instead of only hearing it from sales and support so you can understand context and actions.
Continuous Discovery
Guiding Choices in UX: The Role and Ethics of Nudging by Dr Maria Panagiotidi
Nudges are incredibly powerful ways of encouraging user behaviour - but they can be used for bad as well as good.
Mastering Product Discovery - Quantitative Discovery by Adam Fishman And Behzod Sirjani
Everything you ever wanted to know about Quantitative Discovery with stories from Lyft, Patreon, Facebook, Slack and more.
Continuous Design
Beyond numbers: Proving the value of design(ers) by Ida Persson
If we invest in the proper research and testing, much of the impact of design is measurable with numbers but humans are ultimately motivated by emotions.
There Are No Original Ideas. But There Are Original Executions by Luke Wroblewski
"There is no such thing as an original idea". This may be true but there are infinite ways to execute and that's what makes the difference.
Continuous Delivery
The Value of Socially Driven Architecture by Richard Bown
Our product architecture is unduly influenced by HR. We need to actively influence our org structure to influence our architecture.
Automated Unit Test Improvement Using Large Language Models at Meta by Nadia Alshahwan et al
Facebook ran an LLM test-a-thon to improve the automated test coverage. 73% of recommendations were accepted to production increasing test coverage by 25%.

FREE COMMUNITY EVENTS
IN-PERSON 27 Mar: Dublin 🔔 Want a UXDX Community event in your city? or, alternatively, if your company wants to host an in-person event please reply and let us know. | ONLINE 05 Mar: Using Comedy and AI Insights for Team Success 12 Mar: Leading Design Innovation: From Teams to Market Impact 19 Mar: Mastering Design and Market Dynamics |
THE UXDX MAJOR EVENTS
UXDX USA | UXDX EMEA |
New Videos Released This Week
In this session, Susan will discuss CarMax's challenges and successes over the past few years in creating an improved online and offline experience for their customers. 👇👇👇
The job of the Week
Senior Product Manager
🏬 ThoughtWorks
📌 Fully Time, Milan, Italy
About ThoughtWorks
Is a worldwide technology consultancy known for blending strategy, design, and engineering to foster digital innovation. With over 30 years of experience, it has earned the trust of clients for creating unique solutions through autonomous teams.
The Results of Last Week’s Poll
The question was: Are you worried about the risk of manipulation when it comes to AI system outputs?

Almost 90% of people have concerns over how the AI's that we will increasingly rely on are being biased.
For me, the key thing here is to try to support multiple competing systems. I personally find it useful to compare outputs from Google's Gemini against ChatGPT. And Mistral is starting to make quick improvements in the open source space as well.