The Answer: What Does "I Test in Prod" Really Mean?
"I test in prod" is the developer's way of saying: my code meets real users before it meets a test suite. "Prod" is short for production, the live environment that real customers depend on, and the joke is the confession that changes go straight there. No staging environment, no dress rehearsal, just confidence and a deploy button.
Like the best developer memes, the phrase has two completely different readings. Read one way, it is a confession of recklessness. Read the other way, it is a serious engineering philosophy, because a growing school of thought argues that every team tests in production whether they admit it or not. The shirt-wearing crowd enjoys that nobody can tell which reading applies to them.
The Origin: From a Beer Ad to a Badge of Honor
The meme that started it
The long form of the joke is "I don't always test my code. But when I do, I test in production." It is a parody of the Most Interesting Man in the World campaign, and it spread through sysadmin and DevOps culture in the early 2010s as an image macro. It showed up in conference slides, on laptop stickers, and in every retro where someone had just shipped something untested. The short form, I test in prod, is what survived.
The plot twist: the joke became a philosophy
Then something unusual happened: serious engineers started saying it sincerely. The observability movement, led by companies like Honeycomb, made the case in public: production is the only environment that actually matters, so you should be watching, validating, and yes, testing there on purpose. Increment ran a whole essay titled "I test in prod". The meme and the methodology now share a wardrobe.
Why Testing in Production Is Partly Unavoidable
Staging is a polite fiction
Staging environments are approximations. The data is synthetic or stale, the traffic is fake, the third-party integrations are stubbed, and the scale is a fraction of the real thing. Plenty of bugs only exist at production scale: race conditions under real concurrency, cache behavior with real access patterns, that one customer with 40,000 items in their cart. You cannot rehearse reality.
The tooling that made it respectable
What separates the methodology from the meme is control. Feature flags let you ship code dark and turn it on for 1% of users. Canary releases roll changes out to a small slice of traffic and watch the error rate. Blue-green deployments keep an instant rollback one switch away. Observability tooling tells you what actually happened. Test in prod responsibly, as the saying goes.
The Two Ways to Wear the Phrase
The i test in PROD shirt works precisely because of the double reading. Worn by an SRE, it is a manifesto. Worn by the intern, it is a warning. Worn to standup the morning after an incident, it is a confession that requires no further documentation. It is consistently one of our best sellers for exactly that reason: every engineering team has at least one person it describes perfectly.
The Meme Family
"I test in prod" sits in the middle of a whole lifecycle of deployment humor. Upstream is "it works on my machine", the excuse for why testing never happened locally. Downstream is breaking prod, the natural consequence. And governing the whole timeline is the rule that you should never deploy on Friday, because prod testing on a Friday afternoon is how weekends die.
FAQ
What does testing in prod mean?
Testing in prod means validating changes in the live production environment, with real users, real data, and real traffic, instead of only in a staging environment. Done with feature flags, canary releases, and monitoring, it is a standard practice at scale. Done without any of that, it is the meme.
Is testing in production bad practice?
Uncontrolled, yes. Controlled, no. Mature teams test in production deliberately using feature flags, canary deploys, and observability tooling, because staging can never fully replicate production. The joke targets the uncontrolled version, where production is the first place the code has ever run.
Where did the phrase "I don't always test my code, but when I do, I test in production" come from?
It is a parody of the Most Interesting Man in the World ad campaign, adapted by developers in the early 2010s. The image macro spread through DevOps culture and conference slides until the short form, I test in prod, became sticker and t-shirt canon.
Is the I Test in Prod shirt a confession or a brag?
Both, and that is the point. Site reliability engineers wear it as a statement about observability. Everyone else wears it as a confession. Nobody can tell which one you are, which makes it the perfect engineering shirt.
The Bottom Line
"I test in prod" endures because it sits exactly on the line between malpractice and best practice, and the industry keeps moving that line. Ten years ago it was purely a joke. Today, with feature flags and canary deploys, the most sophisticated teams in the world test in production deliberately. The meme was right all along. It just needed better tooling.
More from the world of deployment humor: browse coding shirts or the full Developer Culture blog.