Fail Fast versus Fail Fest - The Brazilian BA

This article is also available in Portuguese.

There is a huge difference between fail fast and fail fest.

The first is one of the smartest ideas to emerge from the Agile movement. The second is a dangerous distortion of that principle.

Unfortunately, some organizations are confusing one with the other.

Failing fast never meant producing just anything, putting it into the customer’s hands, and waiting for them to discover the problems for us. That is not agility. It is simply a festival of failures.

The True Meaning of Fail Fast

Traditional software and digital product development follows a relatively simple logic: a great deal of time, money, and effort is invested in building something that seems like a good idea in order to deliver a “finished” product. With this approach, only after months (sometimes years) of work does the product finally reach the customer.

Only then is it possible to answer the most important question of all:

Does this idea really create value?

If the answer is no, the entire investment has been nothing more than a tremendous waste.

The idea of shortening feedback cycles emerged precisely to reduce this risk.

Since value can only be measured when a product reaches its beneficiary, it makes far more sense to discover quickly whether a hypothesis is correct before investing heavily in it.

That is exactly the spirit behind fail fast.

Fail early to learn early.

Not because failure is desirable, but because learning quickly is far less costly than spending months pursuing the wrong hypothesis.

MVP Does Not Mean a Poor Product

One of the best-known techniques for putting this principle into practice is the MVP (Minimum Viable Product).

Many people interpret this concept as “the worst product that still works” or “a rough, unfinished version of the product.”

In reality, that was never the intention.

An MVP is the simplest and most economical experiment capable of validating an important hypothesis.

Its purpose is not to save on quality, but to reduce risk and waste.

It seeks to answer a specific question in the cheapest, fastest, and safest way possible.

Once learning has taken place, new decisions are made, and new investments are undertaken with far more information than was available at the beginning.

AI Accelerated Everything

Artificial Intelligence has taken this logic to an unprecedented speed.

Building software has become incredibly fast. Today, simply describing an idea is enough for an AI to generate the first version of an application almost from scratch.

Even people who have never developed software have experienced this feeling when creating an image, generating a video, or producing a text with AI.

A simple prompt is enough to produce something. Whether it will meet your expectations or not is another story. The fact is that the speed of creation is no longer the main challenge.

Now the real challenge is something else.

When Fail Fast Becomes Fail Fest

The problem begins when a prototype or proof of concept stops being an experiment and starts being treated as a product.

When something that should exist solely to validate a hypothesis is released directly to the market, and the customer receives a poorly finished product, full of defects, without even realizing they are being treated as a guinea pig.

It was during a conversation with my friend Ricardo Stucchi, President of IIBA Brasil, that I first heard him use the expression fail fest to describe this kind of situation. I found the expression so accurate that it became the inspiration for this article.

Fail fest is a rushed, rather than agile, approach to product development in which:

There is no clearly defined hypothesis.
There is no carefully designed experiment.
There is no structured learning.

Instead, an insufficiently thought-out product is simply released into production with the expectation that customers will perform the testing and analysis that should have been carried out by the organization itself.

When complaints begin to appear, the team rushes to fix the problems.

Then new problems emerge. Then more fixes. And the cycle repeats itself.

This is no longer about learning quickly.

It is about improvisation.

Fail fast vs Fail Fest – Created by me using ChatGPT

The Cost of Confidence

This approach can be extremely expensive, and I am not referring only to its financial cost. A festival of failures erodes something that is far more difficult to rebuild: confidence.

Every time a customer uses a product that is clearly not ready, they lose a little confidence in the company that produced it.

The failure is no longer perceived as part of an innovation process. Instead, it is interpreted by the customer as a lack of care, competence, and respect.

While the organization believes it is being agile, the customer simply sees that the product does not work and, in the worst-case scenario, may not even complain. They may simply move to a competitor. The organization will never have the opportunity to learn from the failure and will be left in the dark trying to understand what happened.

Confidence is built slowly, but it can be destroyed very quickly.

Failing Fast Does Not Mean Putting Poor Products into Customers’ Hands

There is a huge difference between validating hypotheses quickly and outsourcing quality to your customers.

The former reduces risk.

The latter creates new risks.

True fail fast is a scientific process that seeks to identify uncertainties, define clear hypotheses, and design smart experiments to learn as quickly as possible and make better decisions before investing further.

Fail fest simply accelerates the delivery of problems into the customer’s hands and, in most cases, does not even state a hypothesis to be tested.

Artificial Intelligence has made it much easier to create products, but it has also made it much easier to produce poor software at record speed.

Do not fall for the siren song that AI’s competitive advantage lies merely in building faster and putting anything online. Success depends on learning quickly, but without turning your customers into guinea pigs.

Speed generates efficiency.

Learning drives evolution.

But confidence remains the most valuable asset an organization must build, and also one of the easiest to lose.