Bill Scott is the Senior Director of UI Engineering at PayPal, and he is in the middle of transforming Paypal from a bloated, slow beast to an agile, lean, learning team. His talk last week at QCon in NYC was awesome, and since the video is not yet up, I wanted to share my raw notes (emphasis below are my favorite quotes). You can see his slides here: Lean Engineering: Applying lean startup principles at Paypal
Intro
– wants to talk about lean ux and lean engineering
– when he got started in software, people wanted to roll their own everything, you had to ship disks so a disconnected delivery experience, there was a culture of long shelf life for software, an inward design focus, and risk averse approaches to things
– when he got to Paypal in 2011, he found all of these things still in effect
– today, the Apple App Store slows down release cycles, like old days of disk releases where you need more pixel perfection and bug perfection, no dynamic updating
– when you in a delivery mindset, you aren’t in a learning mindset
– at Netflix, they did A/B testing and metrics (good for incremental), but started to get out of the building to do user testing (better for bigger leaps)
– at Netflix, they ported Webkit to the PS3 just to be able to test, release, build measure learn; they were able to release 4 different PS3 experiences on the same day to test
– qualitative is important, but people don’t know what they think when you talk to them, you need to try things on them
– volatility was not what he learned in school, but rather craftsmanship and engineering, but volatility is the truth of most real world software product creation
– they want to try things out on a small number of people but not fully engineer until it was proven
– in consumer web, if you launch a single experience, you have failed. you aren’t learning
– in 2011, a copy change took 6 weeks to get on site
– arrived and talked about moving from culture of long shelf life to rapid experimentation; tech choices were important but secondary to transforming mindset
– David Marcus joins Paypal as CEO with an entrepreneurial attitude
– April 2012 formed lean UX team to reinvent checkout
– brought in new talent
– get to shared understanding, deep collaboration, and continuous customer feedback
– lean engineering: shift to minimal viable everything: not just product but also process, team size, technology, tools
– his number one goal is to enable learning.
Key Principles
1. engineer for experimentation
– experiences must learn, our software is always tearing itself apart, different layers have different velocities
– move from a culture of delivery to culture of learning
– “hermes project” – they experiment with more of a Kanban model rather than pure scrum; form lean ux scrum team and dump scrum ceremony; went directly to sketching; took designers away from axure so that they didn’t disappear into their tools; built things fast; freaked out the designer — “what about my job?” because he had been trained that his job was documentation
– lean ux: enable a brain for agile; with agile, you can have great velocity and acceptance tests for crap
– agile without thought focused on delivery, not learning
– Bill is focused down at how do I build and iterate this product
2. refactor your way out of debt
– sign of how screwed up an enterprise is how many innovation teams they have; because they are fighting against anti-bodies; rather than everyone innovating
– tech debt — need to be able to continuously deploy and iterate
– minimal viable everything – What are startups using?
– zuck’s rule: how will the startup down the street do this tomorrow? let’s do it like that now
– had to change the front-end tech stack to enable more flexibility; really wanted a templating system to render pages and wanted same on client or server
– for product “hermes”, pulled in node.js, dustjs (linkedin also using), bootstrap; prototype side uses node.js, production uses java; same code could be used on prototype or brought over to production stack drag and drop; has even brought the ability to do dust on C++ stack
– this was a learning play, not a tech efficiency play though that is good
– with wrong tools, took 40 people and now can do it 3-5 people
– now testing if node can scale enough to drop java
– and this helps them recruit elite talent
– offer “good tech stack, learning, lean, value” and great devs want to come work for you
3. be “open source” top to bottom
– uses enterprise github, democratizes teams and coding (i.e. take an open source approach to things, even with internal code… he doesn’t mean literally open source everything to the public)
– inside your company, stop putting code behind hidden walls, let anyone do pull requests, etc
– use open-source tools religiously — something open is better, throw out your home-grown stuff and use the open stuff; this also makes it much easier for new people ramp up
– this ties back to how would a startup approach this?
– give back to open source too
– use github for continuous integration
They changed technology, process and people, but they started with the mission of enabling learning and iteration
Culture
Q: how to mitigate the fear from other groups feeling threatened?
A. don’t be territorial; don’t hold things to just your own people; bring em in, let them get involved; you can’t be building an empire or people will detect that; give credit and make others successful
– need top leadership focused on right things, not size of team under a manager or delivery, but rather value creation, then it spreads
– have to get down into why they are afraid
p.s. want to hear Bill for yourself? He will be at Lean Day West this September.