Twelve years ago Eric Ries published The Lean Startup, a transformational book that forced companies – big and small – to take a hard look at how they were building products and services.In short, Eric suggested that everything we work on is based on risky assumptions and challenged us to de-risk those assumptions with lightweight experiments. He called those experiments Minimum Viable Product or MVP. The business world adopted this language nearly universally and the MVP craze was born. I fully admit that I got on the MVP train along with everybody else. Eric’s work and early support provided a solid foundation for my own progress and success and I’ll always be grateful to him for that. Somewhere along the way, though, the core Lean Startup concept – the MVP – got lost and its meaning bastardized. It’s time to retire it. Here’s why.
A very brief history of the MVP
Eric wasn’t the originator of the term MVP. It was Frank Robinson who originally quoted the term. In Robinson’s definition the MVP is an actual product designed to be “just the right size” for what your company and customer need right now. It should work, be secure and perform well. It just doesn’t have to serve every customer at scale at launch. You use the MVP, per Robinson, to learn where to expand development activities, where to scale back and where to add new features. It’s a learning tool. Sound familiar?
Eric, as written in The Lean Startup, defined MVP as, “the smallest thing you can make or do to test your hypothesis.” There’s no explicit need for the smallest thing to be an actual working product. In fact, The Lean Startup expressly advises against building out code until you have enough evidence to justify that level of investment. By Eric’s definition the MVP is an experiment. And, as my friend Jeff Patton likes to say, Eric sold the most books so he gets to define the term regardless of who originated it. 😉
1001 definitions of MVP
As Eric’s book became a bestseller, organizations around the world adopted the language of lean startup. Many people read the book. Many people didn’t. Yet, the vocabulary became commonplace. I’ve yet to find a single organization that I work with that hasn’t heard of MVP. However, if you ask people in those organizations what MVP means you’ll get at least as many definitions as the number of people that you ask.
“It’s Phase 1 of the product.”
“It’s the worst product we can get away with.”
“It’s as much work as we can fit in before the launch date.”
In almost all of these definitions though there is a product to be shipped to people. This isn’t a surprise. After all, the phrase is minimum viable product. Shouldn’t there be a product involved in the process? If you’ve not read the book, the answer is obviously yes. Except an MVP isn’t always a product.
Let’s stop using the phrase MVP
It’s time to stop saying MVP. It’s lost any consistent meaning – even inside organizations. Instead, let’s use the word experiment. It’s what MVP was always intended to be. It’s a test, a way to learn whether or not your riskiest assumptions are true and how you might mitigate them. You can do that quickly and cheaply in a variety of ways that don’t involve building any kind of product. These experiments should absolutely be minimal – designed to learn the most important thing you need to know right now in the fastest way possible. They don’t necessarily have to be viable and in the early stages of your product lifecycle they certainly don’t have to be products.
Experiments change expectations
When you tell a colleague or stakeholder you’re running an experiment you change what they expect your output to be. They don’t necessarily expect you to ship a product while at the same time they do expect your work to be temporary, designed to drive learning and inform future work. We don’t even commit to the work being viable. In fact, the experiment may be designed to test exactly that assumption. Is this idea even viable? By moving away from MVP we reduce confusion, set clear expectations about what we plan on doing and what the result of the work will be – learning, not customer-facing products (in most cases). So, what do you say? No more MVP’s, many more experiments?
A quick note of thanks for continuing to read this blog week after week. I’m grateful for your support, attention and feedback. I hope you have a fantastic holiday season and a happy new year. See you in 2024 with a new book!
Yeah I agree, I have been using it as well to let them know that it is experiment.
Thanks–these are great points about how “MVP” isn’t a useful term any more.
And yes, it’s an experiment. But there are lots of experiments that aren’t the-thing-formerly-known-as-MVP. So “experiment” doesn’t seem precise enough to me.
Here’s an attempt to pin down the traits of one of these experiments:
– it’s a *release* (of a feature or project that contributes to a product);
– it’s the *next iteration* in a series of such releases;
– it’s the *smallest* next iteration that is useful as an experiment/for de-risking
So maybe we could call it the Minimum Next Iteration?
You can’t sell an experiment. I’ll stick to “product”!
not sure if experiment and MVP are meant to relate to the same stage of a process. I’d say an experiment would preceed an MVP.
I agree. What is described here is something like POC.
In the end it is important to agree in organization what is what and play with it.
I’ve worked in organizations that attempt to adhere Eric Ries’s definition of MVP. They have often distinguished MVP from “the smallest product we can get away with” by introducing terms such as Minimally Marketable Product or Minimally Billable Product.
If Frank Robinson came up with the phrase, and his focus was on actual products, why not continue to use MVP in the way he intended it, whilst simultaneously experimenting throughout the product lifecycle as Eric Ries promotes, without reducing the former to the latter?
Eric’s contribution was brilliant, but you seem to be merely suggesting that he hijacked someone else’s concept and turned it into the central message of his book.
He shouldn’t be allowed to get away with it! Don’t encourage us to help him! 😉
The difference to me seems to be that visible implies that it will be valuable, the fact that it is the minimum amount of value we can justify in a light amount of effort or the best value we can get in the minimum amount of effort is what defines it as viable. The experiment to me doesn’t even infer that there will be any value to the user except that we won’t waste their time with something that wasn’t valuable being delivered due to lack of experimentation + feedback loop.
How about instead of discarding a good idea, just emphasise three true meaning and focus on experimental nature that bit better and not sweat the people outside of your own circle who aren’t getting that correct.