Why is Scrum so hard to implement?

by Graffen 27. October 2009 21:14

A little over a year ago, I got my Scrum Master certification. The two day certification was extremely encouraging and inspiring, and I really felt I was ready to take on the world with my newly-acquired skills. And Certified Scrum Master sure does sound sexy, doesn’t it?

One of the things I like the most about the Scrum Master role is the coaching aspect. I really enjoy coaching, and I love the satisfaction I feel inside as the process begins to grow on the development team. When they start gathering at one-minute-to-daily-scrum-time so they can report the day’s progress to each other.

But why is Scrum so hard to get to work properly?

At the company I work, I have been trying hard for a year now to implement, if not all, then some of the Scrum process into our daily workflow. And my boss has been backing me as best as he can. My main problem is that he (and basically everyone else in the organization) don’t really understand Scrum. Some know of the basics (daily standup “scrums”, sprints, timeboxing, burn down charts), but I’m learning the hard way that if Scrum is to be successful in our department, everyone around us needs to somehow be aligned to the process and work model. Basically, I think that sending everyone to get Scrum Master certified would definitely help. But of course this isn’t really an option.

One of our biggest impediments is that we don’t have real product owners. This role is basically missing, and we “get by” at every sprint start by asking someone in the organization with domain knowledge to play the role of product owner. We also face the challenge that many of our projects are either complete re-writes of older products, or they’re something new altogether. They aren’t simply existing products that we’re just adding features to. This means that the scrum teams are facing a huge amount of architectural decisions during sprints. And seeing as we don’t really have an “architecture team”, many decisions are just made at meetings they hold together. These meetings are expensive – they often come up completely unplanned and can easily throw the burn down chart way off (I’ve seen it happen many times).

Another impediment is that we’re in the process of integrating systems from a competitor that we acquired earlier this year. This is of course a top-priority from the management group and they will spare no means to get the job done. In practice this has the effect that team members can get pulled at moments notice to do something else. And there is little I can do to stop that.

Another challenge I’m facing is the whole aspect of reporting. Finance would like us to begin reporting hours spent on each and every development project. The problem is that “time spent” is not measured in Scrum. At least not if you want to follow the process completely (which is a goal for me – Ken Schwaber, the father of Scrum, says “If you are not an expert in something, you are not at liberty to change it” – and I totally agree) I cannot possibly adapt Scrum to “work better for us” if I haven’t fully tried it in its basic form first. And we’re still getting there. But nonetheless I still have to figure out an effective way of reporting hours from the team, without asking them to actually start logging work done.

The next thing I’m going to try to push into the team is the concept of using story points when planning the product backlog. I want to start pushing the concept of relative size of backlog items, as opposed to using hours or man-days. The reason I like the idea of story points is that managers have a tendency to accept hour or day estimates as a commitment from the team. By using story points we abstract this away and convey the fact that backlog estimates are extremely rough and are solely based on hunches (because of the nature of backlog items, in that they are not at all very specific).

Next week I’ll be attending the ÖreDev conference in Malmö, Sweden. I’ll be following quite a few Agile sessions and hope to get some answers to some of my questions. Like “How can we get better at following the Scrum process as a company and not just as a small isolated team”

Tags: , , , ,

Programming | Scrum

Comments

10/28/2009 7:28:48 AM #

pingback

Pingback from topsy.com

Twitter Trackbacks for
        
        Graffens Blog | Why is Scrum so hard to implement?
        [graffen.dk]
        on Topsy.com

topsy.com

1/5/2010 11:46:56 PM #

Hamish Graham

Hi, interesting post. Scrum to me is definately more than a process I feel, it's almost more like a state of mind/culture/belief, or general "feeling" that takes more than a description of a product backlog and daily standups to explain. Maybe you need to gather everyone into a room and try and impart the agile "feeling" Smile

Having people pulled off is a pain and unpredictable for velocity, one project I was on we decided to just make one person full time on the "distractions", but I understand that won't always be possible.

RE the hours reporting that can be horrible if you need to do it per work item (trying to remember when you started, distractions, multitasking etc...). Would a flat number of hours worked in the day be enough for finance? I've always had to do that at least anyway.

My last project we did the initial estimates with ideal days (and pretty roughly). They actually ended up turning into story points effectively, the actual time per task ended up being about 170% of our original estimates in the end. Which, as you say does have the effect on making the project look as though it's struggling, as opposed to acknowledging that the estimates were simply very rough in the first place and probably more effectively just a relative size to each other.

Hamish Graham New Zealand

1/12/2010 11:41:11 AM #

Graffen

Hamish,

Thanks a lot for your comment. I definitely agree that getting everyone together in a room and really explaining all the "why is agile so cool?" would be a great thing.

Regarding the financing part, we have reached a state where we report per month the total number of hours spent on each of five overall categories (things like maintenance, support, development (and a reference to which project/product), Out of Office etc.). This, I feel, works really well and keeps us away from the distractions of spending too much time on the nitty-gritty of work item logging.

I presented the concept of "Ideal Days" estimation to the team at a meeting recently but it wasn't all that well received. The biggest issue is that they don't overlap much, so they find it difficult to estimate on each others behalf. Which of course is totally understandable.
I find it quite interesting that you landed at about 170% of initial ideal-days estimates - not all that far from what I would have expected.

Thanks again for your comment. I will probably write up another post soon with some more of my experiences.

Graffen Denmark

Add comment


(Will show your Gravatar icon)

  Country flag

biuquote
  • Comment
  • Preview
Loading



Copyright © 2009 Jesper Hess Nielsen. This work is licensed under the Creative Commons License.
Powered by BlogEngine.NET 1.5.0.7
Theme by Extensive SEO

About the author

A blog about me, R/C planes, .NET, Twitter and whatever else I feel like writing about.