❝Story22:00
Spending a Week on a Menu Animation at Square
Building Square's restaurant point-of-sale, Kevin and a designer spent a literal week fine-tuning the milliseconds of a menu-group tap animation — because it made the difference between a fast bartender's muscle memory and a first-time worker's learnability. They brought real servers and bartenders in with iPad prototypes and watched for hesitation or flinches when the animation felt too slow. He pushed the launch date three times, arguing the deliverable mattered more than an artificial GA date.
- A menu-group pop-in animation determined how easily restaurants adopted the POS
- They tested with real servers and bartenders, watching for hesitation caused by slow animation
- The product had to serve both muscle-memory experts and first-time users equally
- Kevin pushed the launch three times because the deliverable mattered more than the deadline
- A PM can't wave off details as 'not my responsibility' — the final deliverable is fully on you
“a designer and myself spent like literally an entire week just fine-tuning how many milliseconds it would take to pop in and out so that…”
“it's easy I think for a PM to say that's not my responsibility I Define the requirements”
#design#square#craft#user-testing
❝Story38:30
Calling Shopify's Shop App Flywheel — and Having It Loop Back
A concrete decision-log example: when Shopify launched the Shop app for package tracking, Kevin drew a flywheel diagram showing how they were quietly hijacking consumer buying behavior to build an Amazon competitor, and tweeted it. Roughly 60 Shopify employees followed him, and years later people who worked on it confirmed he'd nailed the thesis — though their internal reasoning differed slightly from his while reaching the same outcome. It shows the log's payoff: writing down rationale, then having reality feed back to sharpen your model.
- Kevin diagrammed and tweeted his thesis on Shopify's Shop app as a consumer-buying flywheel
- ~60 Shopify employees followed him afterward
- Insiders later confirmed the thesis was right, though their reasons differed from his
- Broadcasting a decision-log entry can make reality loop back and correct your model
“there must have been like 60 Shopify employees that followed me I was like what the hell is this guy talking about”
#decision-making#shopify#product-strategy
❝Story47:00
The 'Unsell' Email: Front-Load Every Reason Not to Join
Born from failure — hires leaving within six months, surprised by the reality — Kevin's unsell email arrives at offer stage with no more than eight bullet points listing all the gnarly things a candidate will discover, calibrated to fears they hinted at but never said. If they read it all and are still equally excited, hire them; if it gives them pause, better to part ways now than lose them at month six. It cost him 30% of candidates at offer stage and enraged recruiters, but he kept sending it.
- Sent at offer stage, max eight bullets of the honest downsides
- Calibrated to unspoken fears you pick up during the process (e.g. work-life balance for a parent)
- If they're still excited after reading it, hire; if not, part ways now rather than at 6 months
- It lost 30% of candidates at offer stage and upset recruiters, but he refused to move it earlier or drop it
- The email is a conversation opener, not a yes/no — you then bend over backwards for strong hires
“when I first instituted this I lost 30% of candidates at offer stage oh wow which drove my recruiting Partners insane”
“if you can tell them that upfront and they can read that whole email and still be equally excited to join”
#hiring#management#recruiting
❝Story1:02:30
'Did I Draw That?' — Kevin's Daughter and the AI-as-Crayon Moment
When Midjourney's early beta dropped on a Saturday, Kevin let his seven-year-old daughter type a prompt — 'unicorn prancing in a field' — which produced a hideous, demon-flanked, two-rear-ended unicorn. He braced for her disappointment, but she was in awe and asked, 'did I draw that?' The word 'draw' stuck with him: to her, an image-generation model is no different from a crayon. He uses it as his touchstone for how unknowably far AI will reshape what a good product looks like.
- His daughter used Midjourney's early beta and asked 'did I draw that?'
- To her, an image-generation model is the same as a crayon — no conceptual difference
- A child raised on LLMs will define good products in ways we can't yet comprehend
- As a parent, Kevin gets a 'cheat code' to watch how kids adopt these tools in real time
“did I draw that like yeah I I guess you did and the thing that I got hung up on was that she used the…”
“the concept of these image generation models is the same as a crayon to her like there is no difference in her mind”
#ai#parenting#product-future
❝Story1:06:00
Finding Swagger: Getting Laid Off With a Baby on the Way
Kevin's hardest story: laid off in round four of rolling cuts from his first title-PM job while his wife was nine months pregnant with their first child. Beyond the financial fear, his identity crumbled — 'I thought I was a product manager, this is evidence I am not.' Through reflection he reached a reframe he still returns to: there's a difference between you not being good at something and a company not needing that thing right now, or you being good but not in the way that company operates. He wrote it up in a post titled 'Finding Swagger.'
- Laid off round four while his wife was nine months pregnant
- The deeper blow was to identity: 'I thought I was a product manager; this is evidence I am not'
- Reframe: being bad at something vs. a company not needing that thing, or not needing it your way
- The same person who can't make it work at Company A often thrives at Company B — often just environment fit
- He wrote the post 'Finding Swagger' mostly as a reminder to himself
“in the moment all I could think was I thought I was a product manager”
“there is a difference between you not being good at something and a business or company not needing that thing at a particular moment”
#failure#career#identity#layoffs