The Null Terminator
Ethan Ram’s geeky blog on the seam of technology and product management.
Going Agile in a B2B Company
2011 Sep 10Posted by on
On why Agile is the right development methodology on non-internet software companies too
Of the 14 years I’ve been developing software 10 years were with companies doing B2B software (intended to be sold to another business, as opposed to B2C – software that is directed at customers online etc.) In recent years the Agile development methodology is growing strong and a recent Forrester study shows that now over 40% of development teams in the US are using some sort of Agile development methodology. I’ve heard of Agile project in some of the larger companies and had a chance of “upgrading” my own development department to work in an Agile environment (we took Kanban as our preferred Agile approach). Now this blog post is not going to be about my experience with Agile. Instead I’m going to tell you about a talk I’ve had with a friend who told me Agile was not for his (awesome) B2B software company and my response to that. So these are the reasons why he thought Agile was not for him:
- The company’s product release cycle is 3-6 months – They have a larger release and a couple of smaller releases, or Feature Packs, yearly. This is what their sales and channel guys are used to and can handle. They will not accept a daily/weekly/monthly release anyway – this is not how the work.
- Every other year they have at least one of those larger development projects where they tear apart large parts of the system and re-architect them. This takes several months to develop and runs in parallel to other projects. Those projects will not fit into an Agile development environment.
- They cannot afford the time it would take to go Agile. They are too tight with schedule to allow their devs to go back and add unit testing to existing code and development infrastructure for automation – and test automation is key to any kind of Agile-ness. Right?
- Their QA manager is happy with the code he gets and the released product quality is fair.
- The company is doing well in general – why change?!
At first this all seemed logical to me as I knew that the real power of Agile development lies in the quick release cycles (“give something small to your customers often”) and in cases that the software quality socks. Anyway, with another thought, these are the questions I’ve asked–
- Are you, the product manager (or company manager in a small operation), happy with the time it takes from when you identify a clear customer need to the time you have a new feature that you can sell that customer? If I told you could potentially be in a place to define customer-specific releases that would have the new feature out in 3 weeks – how much does that worth to you? ((AND – your R&D manager is not going to freak-out about this last-minute change, because his yearly timeline is getting fucked and his people don’t like working Saturdays.))
- How much time do you spend before every major release reviewing product requirements and development estimations, trying to fit 30 new features into a 10 features development time. Prioritizing again and again, only to find out 4 weeks before release that 20% of features you were guaranteed to have will not make it on time. Or the deadline must be push.
- How many times did you want something added to the upcoming release, as a large customer deal is pending on it, and your R&D managers said there’s no time and you’ll have to take something out of the release – – or the deadline must be pushed…
- Why is it taking the QA teams 2 full weeks to test a new version release? Does it have to be like this: they first get a version that just doesn’t works 4 weeks before deadline. Then this Ping-Pong between R&D and QA till the version stabilizes: The open bug count starts dropping to the point of reaching the ultimate go-no-go meeting, where the QA manager signs-off the version at the end of an extremely hectic and night-less month. No better way to meet the same goal?
- It happens that the more lines of code we have (and we have more every day) the more QA personal we need. We Often get to the point that the QA manager tells me “…we haven’t had the time to complete the testing of this version…”, “…we need more QA engineers just to complete the STP in time…” – We often release without completing the QA cycle and indeed we often need to patch the version a week or two after release because the version has some major bugs. WTF???
- The product manager often complains that the developers missed some of his instructions in the PRD and that a feature is f**ked as a result. Then the QA manager complains that the version has some features missing and some new features he did not know of and one of the team-leaders told him that some things were changed and showed him a couple this email thread from 2 months ago were those new features were detailed. Well sort of detailed.
- Your development manager is asking for 8 weeks of developer time to conduct a code merge of the new feature that is already tested into the main code branch. The Mac OS-X product was written on a branch that was never merged back to the main branch and devs keep complaining that it’s time we move to a better/modern/faster source control server because those XML files are never merged right and there’s so much manual work to do with every code commit. There must be a better way.
- And the competition is there. They are fast. How come they are faster to react then we are?
Well – you’re expecting this – go Agile. 🙂
I’m not saying its magic. You’ll have to invest time to make it happen. You’ll have to give it some chance and believe it can greatly improve your performance. Why “Believe???” – – We are engineers (or sales guys) and we have targets and methodologies to work. Why do we need to believe? Because you’ll have to change the way you work. People don’t like to change. People like to stick with processes they know. They mostly don’t see the flaws. They find it hard to believe that it can be so much better.
This is why I think that the goals for an Agile project must be set by a high ranking manager. The R&D manager is going to be involved for sure, but also marketing/product manager, and sales exec for sure as one of the main goals for an Agility project would ultimately be to improve the sales cycle and time to market. So, it seems the division head or CEO is probably the person that should set the goals and allocate the resources for an Agility project.
OK. So you are saying “the above is exactly the problems I see in my company but I’m not the CEO. I’m merely a team leader in the R&D…” – Now what? Well, send this post to your CEO – ask for a meeting to discuss the topic. Come prepared. Bring along an Agile couch/consultant. They are used to talking high mgmt. into investing in Agile.
Next up – some guidelines on how to go Agile without missing your yearly quota / deadlines.
Qs and comments are welcome as always.
Pingback: Service Quality Central – Getting Started with QA: Getting Your Feet Wet