# How Much For This Software?

Source: https://www.yegor256.com/2015/06/02/how-to-estimate-software-cost.html

"Here is the specification; how much will it cost to create this software?"
I hear this almost every day from
[clients]({% pst 2015/jan/2015-01-26-happy-boss-false-objective %})
of [Zerocracy](https://www.zerocracy.com)
and prospects that are planning to become our clients and
[outsource]({% pst 2015/oct/2015-10-27-outsourcing-doesnt-work %}) the software
development to us. My best answer is "_I don't know; it depends._"
Sounds like a strange response for someone who claims
he knows what he is doing, doesn't it? "Here is the 20-page
[specification]({% pst 2015/nov/2015-11-10-ten-mistakes-in-specs %})
that explains all the features of the product; how come you can't estimate
the cost?" I can, but I won't. Here is why.


{% jb_picture_body %}

Let me ask you something else: _Why do you need an estimate?_
Yes, I mean it---why do you ask me how much it will cost
to develop the software for you? I can tell you why.

Because you don't [trust]({% pst 2017/nov/2017-11-21-trust-pay-lose %}) me.

{% quote Any software, no matter how big or good it is, always needs more and more improvements and bug fixes %}

And obviously you have good reasons for that, simply because we both
know that a software product is something that can stay in development forever
and never be finished. Look at YouTube, for example. How much do you think
it would take to create a website like this, where users are able to
upload their videos and then stream them? A few days for a good web developer.
Will it stream video? Yes, it will. Will it be ready to compete with YouTube?
No, it won't. Add a few hundred developers to the team, a few years, and a
few million dollars, and even then you will be behind YouTube. Simply because
it's a never-ending process. Any software, no matter how big or good it is,
always needs more and more improvements and bug fixes.

{% youtube lgScAwsYWCc %}

Thus, when you ask me how much it will cost to create a system similar
to YouTube, according to your specifications, my honest and accurate answer
should be: "_All of your money, and it won't be enough._" Will you
sign a contract and
[outsource]({% pst 2015/oct/2015-10-27-outsourcing-doesnt-work %})
the project to me after that answer? No, you won't.
That's why I have to lie and promise something like "three months and $40,000."
Why would you trust me? If you're smart enough, you won't.

My point is that no matter what I promise you, I will be wrong. Terribly wrong.

What is the solution? What do you do? I totally understand that you
do need a number to make your plans and secure the money.
You need to choose the right
[software outsourcing]({% pst 2015/oct/2015-10-27-outsourcing-doesnt-work %})
partner, and you need to know what to expect and when, but ...

You're asking the _wrong question_!

This question has only one valid answer, and it's the answer nobody
will ever give you---the development will take forever and will
consume all your money. All other answers are simply a lie.

I'm sorry if I've delivered bad news to you.

But let's get back to the original problem: Why are you asking
me how much it will take to develop the software if you know it's a
never-ending process and there is basically no limit? Because you
want to make sure your $40,000 will be used the right way and will
produce the maximum it can produce. To get this assurance from me, you're
asking for an estimate. I'm telling you that your software will be ready
for $40K, and you sleep well. Until the moment you realize you've been
fooled. By your own self.

{% quote The development will take forever and will consume all your money %}

Your concern is perfectly valid. You want to spend no more than $40K, and you want
to get a product that will help you achieve your business goals. For example,
you want to get into the market and acquire your first few thousand users. In other words,
your biggest worry is that your dollars will be turned into the right amount
of the right software.

Any software team can consume your $40K, but each team will produce a different
_amount_ of software with different _quality_.

My point is that instead of asking how much a software project will cost, you should
ask [how much software]({% pst 2016/jan/2016-01-19-how-expensive-is-outsourcing %})
we can create for each dollar you give us and what
quality will it be.

Basically, don't ask us to estimate how much gas it will take to get
to the finish line, because there is no finish line. Instead, ask us
how much we charge per gallon and how many gallons we consume per mile.

Different teams use different metrics to measure their results
(to be honest, most of them don't use any). We, at
[Zerocracy](https://www.zerocracy.com), use
[hits of code]({% pst 2014/nov/2014-11-14-hits-of-code %}),
bugs, pull requests, test coverage, and a few other metrics
as measurable indicators of _quantity and quality_. We know exactly how much
software we can produce for each $100 you pay us.

Collect those numbers from other teams and compare them. Also, make sure
you can control these numbers during the course of the project. That's the
guarantee you're looking for. Now you know what you're buying and how
much you're paying for it. In other words, like I said above, having these
numbers in front of you will assure you that your money is producing
the maximum _amount_ of software it can produce, at the highest _quality_.

The only question left is how you can know you're buying
the _right_ software. In other words, you know how much
we charge per gallon and how many gallons we use per mile, but
how do you know we're driving in the right direction and not making
too many circles or detours?

There is only one mechanism to guarantee that: regular checkpoints.
You should ask us whether we deliver the software in small and regular
increments, and whether we allow you to conduct regular
[independent reviews]({% pst 2014/dec/2014-12-18-independent-technical-reviews %})
of our progress. Also, make sure we prioritize our technical goals,
use milestones, versionalize releases, publish release notes, etc.
Make sure that in the course of our journey, you are able to control
the progress and make corrections.

To summarize, this is wrong (because there is no "there"):

> Hey driver, how much will it cost to get there?

And this is right:

> Hey driver, how much do you charge per mile, and do you have a map?

Hope I've made my point clear.