Uncharted Waters

Apr 8 2014   2:46PM GMT

Heusser On NoEstimates

Matt Heusser Matt Heusser Profile: Matt Heusser

I’ve covered the #NoEstimates movement as a journalist, twice now. Still, some of you may be wondering “What do you think about it Matt?”

This blog post is for you.

How I define an estimate

The estimates I am speaking about involve:

(A) A technical contributor
(B) Taking time away from production work
(C) To provide hours (‘work effort’) and dates
(D) In order to create the schedule for management to know the cost or duration, etc.

“No estimates”, means, then, just that: Running technical projects without any sort of estimation. At all. Period. Full Stop. You will notice this does not preclude estimating the number of sales next week, or revenue, or inventory in your warehouse. We are talking about schedule for knowledge work here)

Here’s a quote from Woody Zuill on the subject:

Woody Defines NoEstimates

That makes the #NoEstimates hashtag a label for a conversation thread exploring alternatives to estimates. It is not a demand that Estimates should never be done, nor a specific method to live without estimates.

In “No Estimates Land”, we look for other ways to manage the project and make decisions. In practice, the team might be making estimates. Or, as J.B. Rainsberger put so well, the implementation on an actual team might be “how close can we get” to no estimates. That means that compromise measures, the point where you “add just a little” estimation become the end of the conversation, not the focus. Compromise measures, in my mind, kill the #NoEstimates thread; they turn it into something else.

Matt, are you saying estimates are bad?

We do know that estimates are not essential to delivering projects, and that time spent estimating could be spent on making forward progress on the projects. When I ask people why they are needed, I generally get the same four categories of answers: Schedule Control, Cost Control, Comparison Shopping, and Governance. Honestly, in most cases, I find that estimates are not capable of keeping those promises: We find unexpected complexity, and the schedule and cost run out of control. The estimate is either so far off to make comparison shopping unrealistic or the cone of uncertainty is too wide to really compare

To be fair, my colleague, Justin Rohrman, argues that estimates can have benefit: They can force the team to define what the thing actually is, suss out misunderstandings and differences of expectation. Those are good things … but they are a bit like buying a car to the nice AMFM radio; there are better alternatives available.

Environments Where #NoEstimates already thrive

Say you have a task so short that estimating it isn’t worth the time — adding a column to a report or re-running a batch job. A manager or customer asks you to run the interface, or you just realize it needs to be done, and you just do it. Any of us who took a college course, of even developed our own software for fun, are probably familiar with working full-tilt to get the scope done and not worrying about number of hours, and a few of us have experienced time-bound projects where the scope was open. My second article for cio.com covered that in detail.  One of my interests right now is in finding where #NoEstimates are already happening in organizations and admitting that formally.

“You don’t get it Matt. Estimates are just a symptom, we need to fix the root cause!”

You could argue that the cause of the angst over estimates is something about trust, or social status, or some other more deep, perhaps emotional understanding of how we do our work. If you want to talk about that, that’s fine, but please allow me to respectfully suggest a different hashtag. If the hashtag is #NoEstimates, well, I am going to talk about ways of making decisions on projects without estimates.

“My boss would never go for it.”

This has the form of an invitation to dialogue, but it is not; it is, instead, a fiat. If the customer requires Estimates, you need to either to do them or turn down the contract. If you are an employee, good luck turning down the work.

When I want to predict project work, my preference is to use Yesterday’s Weather. Can I work in environments that use estimates? Of course. This is not a religion; it is a conversation. All I am saying here is the the world could be different, and let’s talk about it.

“This Would Never work at the portfolio level”

Some people ask tough questions. They want to know if a project is thirty stories or sixty – or sometimes, between $10Million and $100Million in cost. These sound to me, at least ostensibly, like reasonable questions. Is it possible to have a #NoEstimates approach with even these questions? I think so, yes.

Let’s talk about that next time.


3  Comments on this Post

There was an error processing your information. Please try again later.
Thanks. We'll let you know when a new response is added.
Send me notifications when other members comment.
  • galleman
    Nice piece. 

    Now the harder part. 

    Can the business side operate within it's "norms" (e.g. Governance processes) in the absence of knowing a probabilistic cost, schedule, and delivered capabilities.
    20 pointsBadges:
  • Robin "Roblimo" Miller
    I've owned many cabs and limousines, and I've run a considerable fleet for a large cab company. When we bought repair work, which we did nearly every day, we requested estimates *unless* it was work the garage (sometimes ours; sometimes an outside one) did repeatedly in accordance with a known price list -- which was an estimate of sorts.

    In other words, while we couldn't always know what everything was going to cost us, we liked to be able to predict next week's (and next month's) expenses as well as we could. Hand out work without estimates? Hah! 

    My son in law just called. He installs and refurbishes office cubicles and, increasingly, office furniture in general. Since all of his jobs are essentially custom one-offs, everybody wants an estimate before they approve the work. And Chucky walks around with a tablet computer full of his material cost sheets and other info he needs to make estimates on the spot. 
    930 pointsBadges:
  • Matt Heusser
    Hello Robin -

    Comparing software development to auto repair is, I think, a bit of a dangerous game. There are just too many unknowns unknowns in software.

    Still, even in Autos, there is a little known game for mechanics of getting paid in  estimated hours, not worked hours, because the ratio between the two is often 2:1. That is, a good mechanic can do a typical job in 1/2 the estimated time. You show me a work that pads its estimates that much, and I will show you a shop where estimation can work just fine!

    Still, because, as you say, the work is custom every time, there is a lot more unbounded risk than if it were repeatable and well-defined. Software projects are more like your custom upholstery example, or, perhaps, remodeling a complex old house in ways that involve running pipes and electrical where it currently does not go.

    I own a house built in 1871.

    The game in those projects, at least among the unscrupulous, is to make up the difference in up-charges and change orders. I have played the game, and got a not-great bathroom upstairs, late and over budget.

    Then you have the mr. fixit in town who charges a fixed hourly rate with no estimates. Lots of people won't hire him, but he gets enough business to stay working - he only needs 40 hours a week - and the ones who don't end up with my bathroom.

    My answer ? #NoEstimates sure can work ... if you are willing to try them.
    5,110 pointsBadges:

Forgot Password

No problem! Submit your e-mail address below. We'll send you an e-mail containing your password.

Your password has been sent to:

Share this item with your network: