Nick McKenna's Blog

Agile Executive Overview

Nick McKenna — Sun, 17 Aug 2008 21:23:28 GMT

In preparation for some forthcoming Agile presentations, I have published an Executive Overview of Agile Software Development that will be of interest to managerial staff interested in Agile development.

That is the word "Agile" 3 times in one sentence...

Click here for the article.

McKenna's Hierarchy Of Needs

Nick McKenna — Wed, 09 Jul 2008 14:19:41 GMT

I was thinking about Maslow's hierarchy of needs recently, and I began to wonder how it might apply to software. I thought that you could use the Hierarchy to guide how you build software, specifically how you prioritise work. Have a look at the diagram:

You can see here that the first thing we should do is make sure that the software is reliable (i.e. it does not crash). Next we should make sure that it is secure (i.e. hackers cannot steal data from it). User Stories that are Physiological or Safety related should be the highest priorities in any product backlog. With out these, the product may be saleable, but your customers will become very unhappy very quickly.

Next the software should deliver basic business value. Without this you cannot sell the product. Any piece of software will have a minimum feature set that is required to complete in a given marketplace. For example, any web browser should be able to display HTML 4.0 compliant web pages. These User Stories should immediately follow the Physiological and Safety User Stories in the Product Backlog.

The next item is to provide one or more unique features that give the software under development a competitive niche. For example, the web browser discussed above may have features to enable users to quickly search their bookmarks.

Finally, the software should become more than a few unique selling points and should become a product that leads the marketplace in strategic terms. This means that rather than just adding the facility to search bookmarks, the web browser should lead the market by implementing new features that change the way people use the Worl Wide Web.

It is always tempting to violate the hierarchy outlined above. The marketeers are always keen to see new features rather than a stable product. I believe that the key to delivering long-term happy customers is to get the basics right first.

Estimating Business Value

Nick McKenna — Thu, 03 Jul 2008 13:57:50 GMT

I have written before about my concern about the meaning of "Business Value". As a follow up I would like to suggest a new strategy for Product Owners that should lead to a more consistent and transparent approach to the prioritisation of User Stories based on Cost and Benefit.

First of all, an organisation should define "Business Value". There is no getting away from this. It should not vague or unclear. Agile development aside, an organisation that defines Business Value is likely to be a focussed organisation that gets the right job done at the right time! The team as a whole is more likely to have a positive feeling about proposed work if it demonstrably has Business Value and is not just the latest hot potato! My suggested starting point for a definition would be:

"A User Story has Business Value if it:

Meets a minimum requirement to meet basic user expectation; or
Makes the product more strategically competitive; or
Makes the organisation more strategically competitive; or
Is a popular feature that will generate revenue; or
Is a feature that will reduce costs"

The list above is in order of importance. A product must first meet the basic expectations of the user and it must jump the minimum barriers to entry for a particular industry. Then it should seek to become strategically competitive or make the developing organisation more competitive. Finally the product should have the features required to generate revenue or improve the bottom line through reduced costs.

Every organisation will have its own definition of Business Value customised for its own market, people, political situation etc.

From here on in, the Product Owner should have enough information to relatively estimate the Business Value of a User Story. For example, adding a social networking feature to a Scuba Diving club web site could be strategically important. Interestingly, the Business Value of a User Story may change over time. For example, if all Scuba Diving club web sites have social networking features then it becomes a minimum requirement of the user rather than a strategic User Story that will make the product more competitive.

I would suggest that the Product Owner uses the same scale as the development team uses for Story Point (Cost) estimation to estimate the Business Value. This then allows the Product Owner to perform the trade-off between Business Value (Benefit) and Cost to determine the priority of the User Story.

I am not suggesting any maths (e.g. weighted averages etc) to do any of this automatically. I think that the gut feel of the Product owner based on the Business Value and Story Points should provide enough for the prioritisation to take place.

It is also worth noting that this approach does not add any additional effort to the Agile process. All it does is formalise the process.

Zero Point Stories

Nick McKenna — Sun, 22 Jun 2008 19:13:09 GMT

A team I work with estimates User Stories on a relative Story Point scale based on the Fibonacci series. We actually used the Mountain Goat voting cards to implement this. There has been a little controversy lately about “technical” stories. As far as I can tell, these are stories that do not directly affect the functionality of the software. An example might be a technical prototype to evaluate the suitability of a database for a particular piece of functionality.

The non-techies believe that these stories should be estimated as 0 points as they add no “business value”. I have a couple of problems with this:

I have yet to hear a definition of “business value”
The purpose of Story Points is to allow the team to estimate how much work is required to do a job, not to indicate the “business value”

We know that we estimate Story Points so that we can look at the team’s Velocity and then work out which stories to put into a sprint. We also know that we use estimates to allow the rest of the business to take a longer term view of when a feature or product might be ready for release. Story Points are a measure of work, NOT business value (whatever that is).

As an aside, I would also dispute that technical stories have no business value. In the example above, selecting the correct database for a job is crucial. Imagine that we are looking at implementing a transaction system that needs to process 1,000 transactions per second. The team only has experience of Microsoft Access. Since there is no “business value” in creating a prototype for the technical story, the team decides not to do it. Consequently the team creates software that will never meet the non-functional requirements. Clearly here the technical User Story has large, critical business value.

Back to Zero Point Stories! The problem with these stories is that they deliberately mislead the business and cause unreasonable expectations to be made of the team. A team with a zero point story in a sprint will not complete all of the other stories in the Sprint because the zero point story will take work. So why mislead the business? Why not just estimate the technical story in the same way as any other story?

The argument of giving technical stories zero points because they have no business value is also flawed because Story Points are not a measure of business value – the priority of a story is! If the business wants a different measure of business value then the Product Owner should relatively estimate the business value of a story separately from the Story Points.

So, my recommendation is, if you want to be as accurate as possible in your estimates for Sprint planning and reporting to the business, estimate EVERY story, not just the stories with (the mythical) business value.

Architect-As-You-Go

Nick McKenna — Sat, 21 Jun 2008 10:27:57 GMT

I have been worrying a bit recently about the architecture for a web-based portal I am building. I learned to stop worrying and love Agile development. I have written a short article giving an overview of the origin of and ideas behind Architect-As-You-Go. Click here to read the article.

Open Training

Nick McKenna — Thu, 19 Jun 2008 16:14:24 GMT

I had an idea a few days ago that I have been mulling over. I was talking to some colleagues today about it and I got some positive feedback. The idea is to provide free, open, technical training to anyone who wants it.

The inspirations for this idea are the Open Source community and BSAC scuba diving. It struck me that as techies we contribute time to open source projects that we believe in. BSAC provide free scuba diving training to anyone who wants it. I think it would be a good idea to put those two ideas together.

The plan would be to create an organisation that ran free technical training sessions with volunteer trainers. The mission statement would be something like "To ensure that the UK IT workforce becomes and remains the world's most skilled IT workforce".

The organisation would organise and manage all kinds of training events from short half day sessions to week-long marathons which were staffed by volunteers. Attendees on the courses would be expected to contribute by teaching at some point in the future. We could consider some sort of training credit scheme, but I would prefer to keep things less constrained to begin with. People completing courses successfully would get certified by the organisation. There are other ways in which people could contribute too. We would need to co-ordinate events and run a web site with a calendar on it, for example.

Ideally I would like trainers to make their training materials open for public use, but I don't think we would insist on it.

Any thoughts?

You Ask And You Receive

Nick McKenna — Wed, 11 Jun 2008 09:50:44 GMT

That's what I like about this place - you ask and you receive (thanks to the Lost Boys).

After my post yesterday I found: http://www.extremeplanner.com/

It still has more features than I would like, but it is the closest I have found yet to what I am looking for. Of course, it is not Open Source, but it is cheap at $149 + $40 / year maintenance. There is an online trial which is hosted by their partner company, so I think I will have a look...

My Agile Assistant

Nick McKenna — Tue, 10 Jun 2008 21:03:22 GMT

Agile is all about understanding Occam's Razor. That is to say "All other things being equal, the simplest solution is the best". So where is the simple Agile Assistant tool? I am deliberately avoiding saying "Agile Project Management Tool" because I think that would be missing the point completely. I have been looking at Version One, JIRA and others and they are all way too complicated! They try to satisfy too many users and have too many features. So here is my wish list of features for a tool:

Visual representation of story cards on a whiteboard
Change status of story cards by moving them between lanes on the whiteboard
Add / edit / delete stories
Prioritise stories
Enter estimates for stories
Model how much can be done by when by entering a theoretical velocity (i.e. release modelling)
Group stories into a Sprint
Sprint burndown graph
A story should have:
- Name
- Notes
- Estimate
- Priority
- Tasks (with hour estimates)

I think that covers about 99% of what I want out of an Agile Assistance Tool. Maybe one day I will build it for myself... It seems like the sort of Open Source project that the community would appreciate.

Fun With NHibernate

Nick McKenna — Sun, 01 Jun 2008 13:24:28 GMT

I have had a maddening time this morning with NHibernate. I have a set of tables like this:

aspnet_users
Profile
aspnet_membership
Club

The scenario is that a user has a membership record and a profile. A profile specifies the club that the user belongs to. The club has an instructor (essentially another user).

I am presenting this information on screen in an ASP.Net application in a grid with these columns:

Username, Surname, First Name, Initials, Club Name, Instructor Name, Email Address, Is Active

The user can sort ascending or descending by each of these fields.

A user does not have to have a profile with a valid club, so to get all the users listed, outer joins are required. I am using NHibernate and an HQL query to retrieve the data. All goes well until the user sorts by Club name. At this point, the records with no Club vanish from the grid...

A bit of digging (and SQL profiling) reveals that when an HQL query sorts by a column it has a bizarre WHERE clause that effectively eliminates any records that have null values. It took a while to work these out as the SQL generated by NHibernate is somewhat arcane.

I tried a couple of solutions relating to inserting a dummy record for "No Club", but this had its own share of issues. Eventually I hit upon the idea of creating a SQL Server View for the basic query with the outer joins. NHibernate seems quite happy to accept a View as a Table. I then let NHibernate do the relevant searching, sorting and row retrieval.

Hey Presto! The whole thing works within 10 minutes! All rows are displayed regardless of sort order and whether or not they are missing related records from a left outer join.

NHibernate is hard work. I hope that the full SP1 is out for .Net 3.5 soon so I can try Entities instead. I am resisting downloading the current Beta, but it starts to look more tempting the more I struggle with NHibernate...

Test Driven Development At HR SoftCo

Nick McKenna — Wed, 28 May 2008 20:43:43 GMT

As promised, I have completed the second article in my series on the development practices we have implemented for my client HR SoftCo (not their real name!). Have a look here:

http://mckennatribe.com/archive/2008/05/28/test-driven-development-at-hr-softco.aspx