Posts tagged "Facebook"


The Butler Did It

February 21st, 2022 Posted by Resources 0 thoughts on “The Butler Did It”

The Case for the Minimum Viable Product

Conventional wisdom says the startups should first build a Minimum Viable Product (MVP) to test customers’ reactions.

The MVP is a barebones version of the actual product the startup intends to sell. Unlike the genuine product on the shelves (or in the app store), the MVP would provide only the key functionalities – those that make it worthwhile for the clients.

The startup would analyze the feedback from the clients and then refine the MVP into the final product. The process usually takes several phases. Each phase addresses the issues identified by the client as well as new ideas the startup team comes up with. Eventually, the product is polished and ready to sell.

Starting with an MVP allows the startup to save money and time by avoiding developing features the customers won’t need. It also helps the startup find out early on if customers care about such a product.

People often say they would like something, yet when they get it, it turns out they don’t need it badly enough to generate a profitable business around it. Therefore, using an MVP and collecting their feedback goes a long way towards avoiding a costly failure.

The Trouble with the Minimum Viable Product

Sometimes, it takes many months just to develop the MVP. Even worse, it might require some expensive parts. For example, building the MVP for a flying car would hardly come cheap.

In these cases, entrepreneurs sometimes build mock-ups or models instead and use them to extract as much helpful feedback as possible before moving on.

Such models are OK for getting feedback on the product’s looks, but since the models cannot perform the job the client is interested in, there is no way to gather usability data. If the feedback on the looks is positive, the team enthusiastically starts working on the MVP, pouring resources into it.

Juicero, a startup that burned through USD $120 million venture capital, did precisely that: worked for 3 years on a device that was squeezing juice bags. The product ended up having an initial retail price of USD $700, soon slashed down to USD $400, only to go belly up when the clients realized the obvious: that they could squeeze the juice bags manually. So why pay for the expensive bag squeezer?

In practice, Juicero’s business plan relied on selling juice bags at USD $8 per piece.

In theory, the thought process was that people could have continued buying the bags and squeezing them manually. That would have allowed the company to break even and then become profitable. The problem was that once you removed the high-tech bag squeezer from the mix, all that was left was an expensive bag of juice with no justification for the price. That killed the business very quickly.

Was there any way to test the viability of the business without building a Minimum Viable Product?

The answer is yes. There is a way to test if a business can be built around it for any product or app, even without investing in the MVP, and we call it the Butler Criterion.


Image via Freepik


The Butler Criterion

Clients don’t buy cars. They buy a means of getting from here to there. A taxi, a bus, a train, or a bicycle can and do replace the need to own a car. Before cars existed, people used horses or porters to carry them in rickshaws.

The fundamental idea is people have needs, and just like in the case of the rickshaw, they are willing to hire other people to fulfill their needs. A butler was such a hire. The butler directly served his master, and in addition to that, he was managing the other servants, each of them providing a particular service like cleaning, cooking, gardening, etc. Therefore, the butler was the most important servant of the household and, as such, received the best pay and enjoyed a lot of other perks (better lodgings, nicer clothes, etc.)

All the modern appliances around the house perform duties that used to be assigned to hired help, which laboured manually, under the supervision of the butler. If the person who owns the appliances is well off, they usually hire someone to operate them, even nowadays. The reasoning is simple: the time spent vacuuming a large mansion has a higher hourly cost if performed by the estate owner than if the maid performs it. The owner could afford the building precisely because they earned more by the hour than the maid.

That brings us to the Butler Criterion: is there a human who gets paid to do the work our product or app would do for the client?

If the answer is yes, then it makes sense to build the MVP, which would replace or improve that specific human activity.

Applying the Butler Criterion to Juicero would have quickly revealed that people willing to pay for somebody to squeeze juice bags for them would rather pay a cook or a maid to prepare fresh juice directly from fruits and vegetables. That meant there was no real market for an expensive bag of juice and even less for a high-tech bag squeezer.

There is always a market of premium juices, but they come in packages that suggest their premium status, not in plain-looking bags. Therefore, since the business was about selling premium juices, there were other ways to do it properly, and none included building an expensive high-tech squeezer.

Let’s apply the Butler Criterion to Facebook. The paper variant of Facebook, the yearbook, existed for 200 years before its Internet version. Students hired photographers and printing shop workers to make them, and local businesses hired copywriters to create ads for these yearbooks, which appeared on certain pages.

So, when Mark Zuckerberg went to build an online version of the yearbook, he was confident the yearbook was something humans were willing to pay other humans to make. Even though he chose to monetize only the advertising part of the yearbook, he knew for sure Facebook passed the Butler Criterion.

The beauty of the Butler Criterion is it helps more than just evaluating the viability of the business built around the app or product. It also helps detect the right market segments.


Image via Unsplash


Imagine we know how to make a robot which can put clothes on us. At first glance, the robot would replace a 19th-century valet. The thing is, nowadays, people who can afford to hire a valet would find it strange to have somebody dress them.

Yet, even today, people are paid to dress other people, such as those who care for the elderly or differently-abled people. This point relates to our supposed startup because it can identify interested parties by seeing who is willing to pay a person for dressing another person. As a result, they might discover that were a robot to work with the elderly or those suffering from disabilities, some additional features might be needed, which could not have been imagined otherwise. All of that can be achieved before even buying the first screw for the MVP.

The Butler Criterion can be performed through observation or practice as part of an experiment.

For instance, before attempting to build a high-tech-based business platform, it is worth checking if a low-tech version of it can work. Don’t build the high-tech one first. Use manual work to test if there is any need for the platform’s service. If customers keep coming, then build the high-tech platform to replace the “butler” (the manual work). Amazon started that way. So did and, both later acquired by Amazon.

These examples prove that the method works time & time again and, therefore, every startup should use it before starting to work on the MVP.

If the butler did it, we can (and should) make it using technology.

right features

Getting the Right Features Right Early On

March 23rd, 2021 Posted by Resources 0 thoughts on “Getting the Right Features Right Early On”

In the previous episode, “Going From Good To Great – Ideas That Your Investors Will Love“, we’ve talked about the risk of focusing on the wrong features after the prototype is successfully tested. It is now time to learn to get the right features right early on.

One of the most instructive examples of how easy it is to get them wrong and be blinded by the initial success is the story of how MySpace lost to Facebook. What’s scary is MySpace lost almost “overnight”, despite being the dominant social network for four years in a row before that.

A bit of history helps seeing how things unfolded:

  • MySpace had been launched in August 2003 and Facebook 6 months later, in February 2004;
  • In October 2007, Facebook was still behind by a large margin, with 20 million accounts, 27.9 million unique visitors, and 14.8 billion page views, vs. My Space’s 200 million accounts, 71.9 million unique visitors, and 46.4 billion page views;
  • In May 2008, Facebook overtook MySpace internationally, with 123.9 million unique visitors and 50.6 billion page views versus 114.6 million unique visitors and 45.4 billion page views. From then on, things went downhill for MySpace pretty fast.

While it is good to know what happened, the reasons why that happened are the ones relevant for any startup.

It all started with a bug. Back in 2003, MySpace site had been developed in a matter of weeks, in a very agile way, as a clone of another successful social network at that time, called Friendster.

That rushed development resulted in a bug. Due to it, the users could change how their pages looked like by inserting HTML code in places where they were supposed only to write plain text. It goes without saying that was a very dangerous cybersecurity flaw. Fortunately, instead of doing bad things, the overwhelming majority of the users simply exploited the bug to make their pages look unique.

Seeing that, MySpace “listened to the customers” and turned the bug into a feature.

Soon almost everybody took advantage of the possibilities to personalize their pages. That led the MySpace management team to conclude the ability to customize one’s page was indeed one of the essential features for their users.

In the short term, the decision seemed to pay off: musicians and other creative professionals took advantage of the possibility to embed sound and video into the pages and started to showcase their work directly to the public. 2.2 million bands, 8,000 comedians, thousands of filmmakers had their MySpace pages. As a result, MySpace became even more attractive, and membership went through the roof.

Then Facebook happened.

While anybody could register on MySpace, until September 2006, Facebook membership had been restricted only to college students. In addition to that, all the Facebook pages had the same look because the only two things users could do were post text and pictures. Video would only be added in 2007. Audio files have yet to be allowed, 17 years after the launch.

Yet, in spite of all the restrictions on members’ creative ways of expression, once Facebook opened membership to everybody, MySpace lost most of its members to Facebook within the next two years.

Three questions come to mind:

  1. Why did the MySpace users leave in such large numbers?
  2. Could this have been predicted?
  3. Could we learn anything useful for any startup?
right features

Image via Unsplash

Why did they leave?

The most frequent complaint of those who left was MySpace had a “ghetto vibe”. All the “bling” and photos occasionally revealing a bit too much skin was quoted as the main reason for labeling it “ghettoish”.

However, independent studies done at the time showed the majority of the pages, while very colorful, actually had very decent content. But, as the saying goes, “perception is reality”.

Once the users felt “the ghetto vibe”, they concluded the “aseptic” or “boring” Facebook look and feel was preferable. After they left in large enough numbers, even their “ghetto-tolerant” friends had to follow them. A social network’s main appeal is our friends are also on that network. When our friends leave, there’s little for us left to do but follow them.

What made matters worse and enhanced the unflattering perception was MySpace had a cavalier attitude to spammers and advertisers of dubious products. As long as those spammers and snake-oil peddlers paid the fees, they were allowed to bombard the users with their unwelcomed messages.

This tolerance might be explained by the pressure exercised by News Corp, MySpace’s managing parent company. For instance, in early 2008, News Corp boss and media mogul Rupert Murdoch announced that MySpace would make $ 1 billion from advertising by the end of the fiscal year.

At the time of the announcement, the platform was generating less than 10% of that target. One can easily understand the pressure put on the MySpace management by such statements. They had to monetize the social network by any possible means by the end of the year. Even then, they failed in coming close to the target.

All in all, the combination of “ghetto vibes” and aggressively spamming users with ads for shady products resulted in antagonizing enough members and pushing them to Facebook. And that opened the floodgates.

Could this have been predicted?

Let’s start by looking at the fundamentals:

  • MySpace was a social network.
  • Its founders knew it was a social network.
  • Its founders wanted it to be a social network.

In such a case, one would have expected all the 70+ years of scientific knowledge about social networks would have been used to design the features of MySpace. That didn’t happen!

Sociology and psychology started to study in the 1930s how people network with each other and form groups. By the time MySpace was launched, it was well known social class, ethnic culture, race, and religion are four compelling factors in shaping the connections people form.

People live in specific neighborhoods, go to certain pubs and clubs, take part in certain public events, go to certain places of worship according to their social class, ethnicity, race, and religion. Those are places where people from the other groups don’t go. We may like or dislike such self-segregation, but that is how things are in any society.

MySpace allowed people unlimited freedom of expression and, at the same time, failed to provide an equivalent of the separation existing in the real world. That unfettered freedom of expression made apparent the class, cultural, racial and religious divisions while “forcing” everybody to “live” in the same “place”. This meant most of the MySpace users would leave for a platform providing a better “separation” as soon as such a platform became available.

Facebook, with its limited freedom of expression, offered that equivalent to physical separation. Inside Facebook, people would still group themselves mainly according to class, ethnicity, race, and religion because they would be connected to the same people they know in their real lives. As it’s often quipped, people on Facebook live in their own bubbles.

Nevertheless, their social class, race, ethnicity, and religion would be less “in your face” thanks to the uniform design of the Facebook pages.

The phenomenon of physical separation and social class and racial lines even had a consecrated name in sociology: “the flight to the suburbs”. It started in the 1950s when the upper- and middle-class left the downtowns of the American cities.

The advertisers covet most the upper and middle class, and social networks make their money from advertising. So, when MySpace decided to allow users to do anything they wanted to their pages, they were setting themselves up for a “flight to the suburbs” and the loss of advertising revenue resulting from it.

So, what are some valuable lessons to remember?

  1. Users might love our prototype simply because that’s the best available solution at that moment in time. Iteration, by itself, can only confirm a prototype is acceptable. Meanwhile, the investors and we need to know if the prototype is actually good;
  2. A good product could hold its own against the competition, buying us time to redress the situation. An acceptable one would be replaced quickly before we have time to react;
  3. The iterative process of improving prototypes needs to be combined with other methods, like checking our prototype against the relevant body of scientific knowledge. If we discover discrepancies, we make the corrections and then iterate with the users to see how they like the new prototype. This is the privilege of starting anything: corrections are cheaper to make earlier than later.

If we join the right accelerator, we will benefit from a whole set of methods to sort out the good from the acceptable. However, checking the scientific state of the art should be done first, as scientific knowledge is public and generally freely accessible.


46, General H. M. Berthelot Street
Bucharest, 010163
T: +40 770 167 618