Business Intelligence Best Practices: What are your 3 tips for implementing an agile BI strategy?

Here are what I think are three key elements to successful Agile BI:

1. A close working cooperation between business and IT to ensure clear understanding of needs, wants, and desires and the entire team held accountable for the results.

2. A dedicated group of developers and business representatives to achieve rapid development, ability to maintain an uninterrupted pace of development, and deliver timely solutions.

3. Simplicity of design of the solution that enables quick, accurate, and workable delivery of desired solution but can handle ever changing requirements. Along with this is standardization of data elements

You won't like these, but:

1. Start with truly agile new product development. Accept no substitutes.

2. Focus first on business processes, and only second on decision-makers.

3. Make the data and analysis cross existing organizational and firm lines.

1. Start with the business pain points - what are the top 10 pain points the business has today that BI / Analytics can solve? Find out what the time-line is for delivery (which in Agile should be "yesterday" according to the business.

2. Apply a standards based, repeatable methodology which has been fine tuned for delivery and implementation. One which has nearly all implementation and architecture processes optimized. Make sure that both the architecture and the methodology are "scalable" to scope control - in other words, incremental build-out.

3. Find the right team of seasoned professionals - BUT this team better understand the business side of the house, and not just IT. They have to have a working knowledge of the business processes AND the data sets, in order to fill the knolwedge gaps between the requirements and the realities of the situation. Furthermore, they better be team players - not single star players who are too independent to work together. Finally, everyone on the team needs to be accountable for their work, and measured on their progress.

Agility (in my mind) doesn't mean just producing quickly, but it also means the ability to produce quality work, repeatable work, measurable work, and of course - after delivery - the ability to go back and optimize the processes that need it. The ability to be able to build it better/faster/cheaper the second, third and fourth times around are all contributing factors to agility.

Hope this helps,
Dan Linstedt
PS: These are some of the reasons why I built the Data Vault model and methodology....

First of all it is important to agree on the definition of "agile BI"
It certainly has something to do with fast delivery and flexible for changes.
So an incremental approach and prototyping looks important to me.
And if we look at the average projects, these objectives are not met in most of the situations. And that is because a lot of manual activities are involved, not only with the initial projects, but also with any maintenance/changes.
And of course it is nice if you have all the users aligned and if all the future needs are already known. But that is simply not the case, and therefore flexibiltiy is key, and you need high productivity and efficiency in delivery.
Therefor, my golden bullit is to use appropriate methodolgies and tools.

Agile BI is business intelligence that is quickly deployed, easily updated or extended, and easy to use by many people in an organization. In that way, it easily adapts to rapidly changing business needs and stays relevant.

There's a need for Agile BI because traditional business intelligence is often powerful, but definitely not agile. Traditional BI is expensive and complex, is usually only deployed to a small number of people, and typically requires an elite, dedicated BI IT team to make changes and updates. Those changes and updates are achievable, but are resource intensive in terms of time and people (and sometimes require technology updates, as well). So you can't necessarily change as quickly as your business needs are changing.

Imagine traditional BI as a very well armed tank run by an elite cadre of experts. Very powerful, very effective within set goals, but a little slow and expensive to run. Agile BI is an SUV (I guess that's not really a military vehicle, so imagine an armed SUV) - sportier, faster, effective, and able to be driven by people with minimal training.

There are a few things that you need to do to make *any* BI implementation successful, agile or not, and some people have already spoken about them here, such as:

1. Defined needs - have it be clear what you're trying to achieve
2. Consistent dimensions and measures - ensure that it's easy to integrate data
3. Executive sponsorship - have someone leading the charge on the business side, who has clout
4. Clear, sustained project accountability - everyone should know who is responsible for what part of the project on an ongoing basis

For Agile BI, there are unique things to also keep in mind in order to be successful:

1. Pick the right technology. Not all BI solutions have the ability to be agile. Not all fast solutions have the ability to be powerful. Not all solutions are easy to use by business people. So know how much power you need, and pick a solution that achieves the right balance for your needs. Agile BI is about a solution that's right for now and right for the future, so pick a future proof option.

2. Pick an achievable starting point. Agile BI is about iteration and expansion - little bangs, not one Big Bang. Start with a key, but small project. Once it's successful, it will provide the momentum for achieving more projects.

3. Project accountability, to a higher degree. With Agile BI, there's not one mega project, there are multiple projects. There needs to be a clear line of accountability, project priority, and communication between all parties (business users, report developers, IT) to ensure that the project iteration, updates, and improvements are happening in the right priority and that everyone is getting the feedback that they need to be successful.