质量就是增长：Stripe 设计负责人 Katie Dill 谈美感、巡店和带设计团队

Dill tells the story of her early days at Airbnb, where she was hired to lead the experience design organisation of about ten people, who had previously reported to one of the founders. She knew from interviews that collaboration between engineering, product management and design needed work, and engagement scores on the design team were low, so she says she "came in swinging."

About a month in she found a Thursday 8:30 AM meeting with half the design team and an HR partner. In a plain white room, five designers took turns quietly reading from papers about what she was doing wrong. She resisted the urge to defend herself and listened, and the common thread was clear: she had not earned their trust and was not bringing the team with her.

Dill calls it the biggest learning experience of her leadership career and is grateful they had the courage to say it. She shifted to listening to what individuals cared about and what motivated them, while still moving in the necessary direction. A couple of months later the team had the best engagement scores in the company, and she has taken the lesson to later jobs: come in listening, not swinging.

Asked about the famously co-located design room, Dill says she believes building together matters, so engineers, product managers and designers should sit side by side with shared goals. Designers also need a community of their own, so she brought them together at key moments each week, since people effectively wear two T-shirts: design and their cross-functional team.

Lenny notes that most PMs and founders understand the value of design in theory but rarely prioritise it over features because the return on investment is unclear. Dill says the question may never go away because the quality bar keeps rising. She sets out levels: the product works and delivers value, then it does so without errors, and beyond that it exceeds expectations in ways users did not seek, with the target level set by user expectations.

The temptation is to chase another feature users are shouting for, or to pick work you can measure over work you cannot. Companies that see quality as non-negotiable do not play that game. Dill compares it to going to the gym: one workout will not give you abs, but skipping it today makes skipping tomorrow easier, and the benefit accumulates.

She notes that with no competition quality matters less, as with the first car whose wheel was hard to turn, but today's cars are distinguished by stitching, leather and the sound of the door. Her slogan is that "quality is growth." Stripe's growth team is maniacally focused on better experiences, and onboarding improvements that made products easier to understand raised activations.

The biggest example she gives is checkout. Stripe's research on top e-commerce sites found that 99% have errors in checkout that hinder a seamless, quick purchase. By improving details small and large, Stripe saw a 10.5% increase in businesses' revenue from an older form of checkout to a newer one. Stripe serves millions of businesses, including Amazon, Hertz, Shopify, Spotify and X.

Dill says one of her pet peeves is talking as if business goals and design goals are separate. The first conversation should be what the team is trying to build. A designer may weigh the emotional experience while someone else says to make the button bigger, so multidisciplinary teams act as checks and balances, but utility, usability and desirability together make something great.

On what beauty is, she cites the book Beauty by Stefan Sagmeister and Jessica Walsh, which notes that use of the word beauty in Google-digitised books has fallen sharply since the 1800s, in line with the idea that functionality is king. Dill says the two are not opposites: beauty makes things easier to use, more approachable and more compelling, and wide audiences tend to agree on what they prefer.

Another example from the book compared tweets from people passing through Penn Station and Grand Central Station, with the former more negative and the latter more positive. For Stripe, as a B2B financial infrastructure company, beauty matters for several reasons she lists: it builds trust because painstaking detail signals care about unseen details, it improves outcomes, it begets more beauty, and it is a matter of pride that attracts people who want to work with them.

Dill says the gravitational pull is toward mediocrity, and that Stripe is a work in progress. For its website, design, engineering, product and product marketing work closely, and engineering and design report into the design organisation for that team. They bat ideas back and forth in rapid iterations, melding art and science, to move quickly at a high standard.

From conversations with design, product and engineering leaders elsewhere, she draws several themes. Quality is a group effort and cannot be left to one talented hire or to QA. Teams also need shared vision and alignment, like a general contractor and architect making sure the roof, deck and siding fit together.

She adds editing, with someone who can narrow and remove what does not fit; at Airbnb she says Brian Chesky is the editor, and the Economist has a chief editor. Next is courage: saying "almost, but no" after a team has put in great effort is one of a leader's hardest decisions. Finally, you need to see quality from the user's perspective, which is a journey rather than an isolated surface.

Dill says products can ship at their best and then regress over time, as other work ships and organisations focus on their own areas, like redoing one room of a house so the rest looks worse. To counter that, Stripe selected 15 critical user journeys, a number she calls somewhat arbitrary, each with engineering, product and design leaders responsible for its quality.

Those leaders regularly review the journey as a user would, from a web search to the website, docs and dashboard, and write friction logs of what works and what does not. They file bugs, contact the teams that own each part and score the journey. Dill says this does not replace user research but adds to it, and that feeling the pain firsthand gives a more visceral understanding.

The scores are then calibrated in a way she compares to performance-review calibration, so the company builds a shared quality bar. The effects include finding upstream issues, such as an SEO or product description that did not match what users needed later in the journey, and converting people in other functions who had not seen their work through this lens.

Lenny observes that people often distrust their own judgment compared with research or data. Dill agrees these are all forms of input, not substitutes for each other. Stripe currently does the walkthroughs quarterly, which gives time for material changes and for scores to evolve, though she hopes some part of the organisation does them weekly.

Dill says each team does the walkthrough together, at a minimum an engineer, a product manager and a designer, because each notices different things such as load time or an inconsistency with the design system. She and engineering leader David Singleton also walk random flows together; she cannot code, but he can. Afterwards teams bring their scores to a product quality review, or PQR, with a few senior leaders and relevant stakeholders.

The friction log uses a template where each moment is tagged as a nice touch, a needed fix or a severe bug. A summary score uses a rubric covering usability, utility, desirability and the "surprisingly great" level. Stripe chose colors over numbers because the scoring is qualitative judgment, and it wanted to avoid debates over whether something is a six or a seven.

On prioritising the findings, Dill says some companies set percentages for fixing versus growing, but Stripe relies on hiring people with strong judgment and pride, and on advocacy at the top, with multidisciplinary teams planning together. She says it is important to show how quality moves metrics, citing an invoice screen where an unclear button led to support contacts, and fixing it cut them.

For performance, she says to define what impact means so that multi-quarter or hard-to-measure quality work is recognised. Stripe's ladder documents mention quality, and its operating principles include meticulous craft, expected of everyone from designing spaces to creating an API to handling support calls.

Asked what she has learned about scaling teams, Dill points to a formula she learned at Airbnb: performance equals potential minus interference. Leaders raise potential by hiring and developing talent, and must reduce interferences, which she likens to a muscle atrophying under a cast. As companies grow, processes break, and she has deliberately run teams hot to learn where the processes no longer fit.

To keep people aware of what is happening, she believes a picture tells a thousand words and a prototype saves a thousand meetings. Designers add screenshots or prototypes to a shared Google Slides deck, which is sent to product managers, engineers and leaders. It helps designers spot overlapping work and shared patterns, and shows others how the pieces of a journey fit together.

She says the deck needs to stay low maintenance and not be a status check; it moved from biweekly to monthly and focuses on medium and large projects, and she would prefer Figma but wants every function to be able to flip through it. Showing work in progress can feel exposed, but finding duplicate or colliding projects early makes the work better and saves time.

Dill's example of interference comes from Lyft. The design team sat in a beautiful room behind a locked door, separate from engineering and product. It gave designers a safe space for exploration, but she says there was wasted work and misalignment because product managers and engineers decided things together while designers worked elsewhere and then met up to find the work did not fit the goals.

Using what she learned at Airbnb, she approached it by listening to engineering and product and getting to know their goals first. She changed the org chart so designers worked on driver, rider or safety teams and sat with their engineers and product managers. She kept a creative space for working sessions and for non-embedded folks, and brought design together at key moments, resulting in faster iteration and clearer work.

At Stripe, which she joined in a Zoom universe, there is a studio space and places where designers sit together, especially in non-embedded functions such as brand and marketing creative and the website team. She goes to the office about half the time and expects to build out the creative space further. Lenny shares a quote about being in the state that makes art inevitable.

Dill says companies at many sizes fear bold ideas because big changes are hard to measure, while incremental steps feel safer and fit a quarter. But users judge quality across products, surfaces and time, so incremental work on your own scope may never make the whole better. Her phrase is to reach for the stars and land on the moon.

She says that means vision work that is not a beautiful deck left on a shelf, but a comprehensive journey-based picture of the ideal. She cites Brian Chesky's 11-star experience idea: if you do not know what the ideal looks like, you cannot increment toward it. From that north star you can work back, ship pieces in order and learn from data while still making progress.

For hiring designers, Dill says it is easier to teach tools and process than taste and character, so look at a candidate's hit rate for judgment and taste. Beyond talent and craft, look for humility, which means paying attention to users and colleagues, and for hustle or chutzpah, because creating something new and saying it is not good enough are both scary.

In early-stage companies she suggests a doer full time, possibly paired with a more senior design advisor. On measuring taste, she says it depends on context: a dense power tool for users feels different from a sparse consumer app, which is why she does not cite a single book.

Dill's favourite recent project is Stripe Press, which prints books on what Stripe calls ideas of progress, most with no link to financial infrastructure. She describes an upcoming reprint of Poor Charlie's Almanac, assembled by Peter Kaufman from Charlie Munger's writing and talks, and a teaser site built by the website team that treats books like spines on a shelf.

She says Stripe's mission is to increase the GDP of the internet, and while financial infrastructure is the core, ideas of progress are another angle aligned with it. It also relates to the pursuit of creativity and excellence that is part of Stripe's identity, and a way people get to know the company.

Dill closes the main discussion by saying that at its core design is intention, thinking about who a thing is for, whether it is a doorknob, an org structure or a strategy. Anyone, whether designer, engineer or product person, can bring that intentionality, though design expertise adds creativity and taste. She praises the podcast for "usable learning per minute" and for Lenny's promise not to ship a bad episode.

In the lightning round Dill recommends How to Win Friends and Influence People, which she has read four times, a book on the Wright Brothers by David McCullough, and The Boy, the Mole, the Fox and the Horse. She also names Oppenheimer and Shrinking, the Toniebox as a product discovery, and the motto "Tomorrow is today," kept on a Post-It. Her favorite interview question is to ask what work a candidate is most proud of, as it reveals taste and judgment.

She credits her parents for raising her to see accomplishment as based on merit and hard work, recalling her father having her chop wood and mix cement as a child. That, she says, shaped both her path as a designer and her willingness to step into rooms where she is outnumbered. She illustrates it with a Lyft driver who told her his daughter should not take out the garbage.

Dill is learning to fly, and one lesson is trust: when her instructor moved his seat back during a landing, she felt he trusted her to take on the challenge. She now asks how she can show that trust to her team, even if she cannot always move her seat back. She can be found as Lil_Dill on Twitter, on LinkedIn, and Stripe is hiring at stripe.com/jobs.