The Promise of Self Serve

How many times in your life have you had an excellent experience with self service? The vending machine saves a lot of time in standing in line and making a transaction with a cashier. This is something that we all understand and take for granted. We are all familiar now with self service gasoline stations, but can you remember what it was like the first time you pumped your own gas? Do you remember the days before online banking? The first iterations of online banking were very cumbersome, with 3rd party software required to connect and non-intuitive interfaces. There is definitely a learning curve for the consumer when adopting self service, and we ignore this fact at our peril.

In the Business Intelligence field, self service is one of the most over used terms to describe the value of a BI implementation. Every vendor has a story about how their product will promote self service with your user community. Based on my experience, it is very useful to clearly categorize the types of self serve access your users can expect. This is a helpful way to segment your user base and to align your BI strategy.

Below I describe my thinking on these user segments.

Self Serve Level 1 (Report Consumer)

• Ability to access / refresh reports on demand
• Ability to chose pre-defined parameters to filter data
• Ability to drill down along pre-defined paths

Basically no ability to customize reporting or explore data outside of pre-defined reports.

Technologies required: Enterprise BI Platform (i.e. MicroStrategy, Cognos, Oracle, Business Objects, etc.)

Self Serve Level 2 (Data Explorer)

• Ability to explore exposed data sets
• Ability to import data
• Ability to created limited reports / dashboards

Technologies required: Data Visualization Tool (i.e. QlikView, Tableau, MicroStrategy Visual Insight)

Self Serve Level 3 (Report Author)

• Ability to create custom dashboards and reports against any available data source

Technologies required: Enterprise BI Platform (i.e. MicroStrategy, Cognos, Oracle, Business Objects, etc.)

Self Serve Level 4 (Data Scientist)

• Ability to load and access large volumes of unstructured, semi-structured, and structured data
• Ability to create and test complicated models

Technologies required: Big Data (i.e. Hadoop, Other Big Data vendors), Statistical Modelling Tool (i.e. SAS, R, etc)

Note that the levels are not mutually exclusive and higher levels will have the capabilities and will leverage the technologies from lower levels.

My recommendation is to segment your user base into the above or similar segments and ensure you have the right training and tools to support each segment. Arguably you should start at Level 1 and ensure you have the basics right, but it is possible to push forward on each level in parallel, but this will drive more complexity into your BI program.

While the promise of self serve BI is real, it takes a lot of work, planning and user engagement to realize the benefits for your organization.

Image licensed under Creative Commons

Engaging Your Users From Day 1

There is something very appealing about a report or dashboard that just speaks to you. There are lots of factors that contribute to this type of experience. For me the first thing that grabs me is the overall beauty of the layout. Is it using consistent formatting? Does the colour scheme appeal to the eye and make the visualizations easy to read? (white font headers on dark backgrounds are my favourite) Do I immediately zero in on an insight that I will take action on?

The funny thing is another person may look at the same dashboard and it doesn't say a word to them. Everyone's needs and insights are different, and we have to design our BI user experience to tailor to the end user. This is why it is so critical to engage our users at the earliest stages of the process.

My recommendation would be start with what business questions we are trying to answer, and what actions this will generate. Before the first mock-up or list of data elements is created, make sure we know what business value we are trying to deliver. Once you have determined the problem we are trying to solve, then understand what data you need to source. This is a critical step, at this point we should be working with our users on what data we need and how that data is generated. We can use this time to establish the quality of our data, and engage our business in ensuring that the data will meet our needs.

An interesting experiment at this point is to make the raw data available and allow the user to play. Let them at a cube or even an Excel pivot table, and see what insights can be pulled from the data. This will help us understand what key metrics and dashboards will speak back to the questions we are trying to answer and the actions we are going to take as a result.

If your users are engaged from day 1, you will save time and your users will be raving fans!

 

BI Stands for Business Integrity

One of the key aspects of working within the field of Business Intelligence is the need for an absolute commitment to Integrity. In my career I have run across a number of situations that have put this to the test.

Data Integrity Issues

No matter the maturity of your data governance program, we have all run across situations where a data quality issue is impacting the quality of our Business Intelligence information or highlights an operational issue. Quite often these issues are highly sensitive within the organization. We have a responsibility as Business Intelligence professionals to communicate these issues to our stakeholders in a sensitive way that is focused on resolving the issue.

Truth Hurts

The facts don't lie. The beauty of Business Intelligence is that once you take a look at the data, there really is no defending the various theories that people use to operate their business in the absence of facts. Again, we need to engage our stakeholders in the process of moving to fact based decision making by having them part of the solution so they can adjust their business practices to the true reality.

Oops

Nobody is perfect. The reality is that we will make mistakes in our BI implementations by inaccurate meta data, ETL coding errors, report logic and the hundreds of other places we can make mistakes in our BI implementations. We need to raise these issues and the impact this has had to business operations and decision making. This isn't about quickly fixing and hoping no one notices, we need to be up front and accept responsibility for our mistakes.

Communicating these issues is not about sending a blanket email coldly highlighting the issues, the best way is to get out of your chair or pick up the phone and talk to people. This is not the fun part of the job. We always need to do the right thing, especially when it is hard.

 

Sometimes You Can’t Make It On Your Own

One of my favourite songs is “Sometimes You Can’t Make It On Your Own”, a signature U2-styled track.  To change the song title up a bit when thinking of BI, the title becomes “You Can Never Make It On Your Own”.  Business Intelligence is a discipline that spans many areas of IT and the business, and you need all stakeholders fully engaged in your BI program to be successful.

Due to the nature of BI architecture and applications, we are always downstream of other IT systems as they are our data source.  You must be tightly integrated with the other groups within IT managing these systems, as any changes can impact the availability and quality of your data.  For example, it is a well know fact within Data Governance programs the more you push the resolution of data quality issues back to the source systems, the more effective the solution.  BI is also a data hog, requiring large amounts of storage and processing power to analyze your organization’s data.  You need to have all the IT stakeholders invested in your BI program to ensure you have the relationships to keep your BI architecture stable.

Most importantly, BI is not an IT discipline, it is a corporate discipline.  Many BI organizations sit within IT under the accountability of the CIO, but in many other organizations this function may be centralized under Finance, Marketing or your corporate governance function.  That assumes that you are lucky enough to have a centralized BI function at all, in most organizations you will find BI scattered throughout the organization.  If we go back to the concept that this is a corporate discipline, we should be doing everything we can do encourage widespread use of common BI methodologies to maximize value.

Once you realize that you have many areas in your organization doing and supporting BI, it will become apparent that a decentralized structure is what is necessary.  To drive the maximum effectiveness you need to have some common thread pulling these groups together.  Having a competency centre type approach, where you provide common standards, access to data, and a means to collaborate are essential for success.

I will be writing a lot more about developing a competency centre to support BI over the coming months, stay tuned!

Mark

What Target Are You Aiming At?

I have been spending quite a bit of time on the KPI Library website lately.  This site originally was setup as a library of common metrics categorized by industry.  Over the past number of years this site has grown and changed to include interesting forums and expert blogs.  I highly recommend signing up for this service (it’s free) and taking a look around.

http://www.kpilibrary.com

Too often we consider metrics an end in themselves, we put all our focus on discovering what data exists, acquiring that data, building reports and dashboards, and once it is up on your company’s BI portal we think our jobs are done.  This is a criminal waste of time if this is all you do.

We must think back to the original question that prompted us to do all this work to get this metric.  A common question in my industry is “Which customers are churning out of our services the most?”  We need to at a minimum define the following before we embark on our analytical journey:

1. What results or objectives are we trying to achieve?  i.e. Improve profitability by lowering acquisition costs
2. What measures can we track to see how we stand against the objective? i.e. do we only measure disconnecting customers or do we delve deeper to understand if the churn is desirable?  Some customers you are better off not having!
3. What targets do we want to set to compare to our measures?  A measure is just flapping in the wind with nothing to relate it to.  i.e. Controllable churn target of 1.5% per month
4. What actions are we going to take based on comparing our measures to targets? i.e. Contact our customers after their initial promotion expires with a value reminder communication

This sets up your framework to measure the right performance, analyze the results, determine actions to improve performance and then to act.  Rinse, repeat, rinse, repeat until you meet your objective.  The measure of success for any BI organization is that action is taken on the information you provide.

Don’t Follow the White Rabbit

There is something special ingrained in the soul of a BI practitioner.  We love to think of all the twists and turns that we can take to analyze the data we are working with.  This comes in the form of dashboards with countless dials, traffic lights and indicators, graphs with multiple series, axis and dimensions, and grids and grids of good ol’ tabular data.  All this stuff makes us very, very happy with ourselves.

Ok checkpoint, a couple issues with this approach.  First of all, all this data can detract from the very business question you are trying to answer with your analysis.  Think about it, do you know of any user that actually has the time to sit down and analyze 200 points of information on a single report?  No, we need to enable our users to make the best decision in the least amount of time.  There are likely 2 to 3 key data points that will get them there.  Yes, it is ok to show basic trends and change indicators, but for Pete’s sake keep it simple.

Secondly, the more we go with this smorgasbord approach to BI we are encouraging requirements bloat in the next BI project or iteration.  Users will simply say to myself: “I should ask for everything I think I might need just so I know I can get it if I need it.”  Why wouldn’t they think that, when that is how we have delivered BI to date?

Fight the urge to go down the rabbit hole; strive to deliver 80% of the value with 20% of the work.  How much more value could you drive for your organization if you can quickly move on to the next opportunity rather than lingering in analysis paralysis?

BI Leadership Article

It isn’t often you come across a good article that brings together developing a Business Intelligence vision and the behaviours BI leaders need to demonstrate to make this vision a reality.  Well worth a read if you are a BI leader or are just trying to understand them…8)

Creating Shared Responsibility for Success in Your Business Intelligence Team

I’m glad to see we are already doing some of this in my current organization, but there are a number of areas discussed in this article that can be improved.  The fun never ends!

P.S. Apologies for the lack of updates to this blog, good weather calls!

The BI Experience: One Size Doesn’t Fit All

The enemy of BI adoption is treating all your users the same.  Each role and person in your organization will use your BI platform differently, so you must deliver a solution that fits each individuals needs.  You should consider aspects such as tool complexity, usage of visualizations, navigation and accessibility.

Executive Users will be best served with high level dashboards, mobile access, and real time alerts.  The typical executive (an effective one anyways) doesn’t spend a great deal of time behind their computer.  When they do, they are likely short on time and are looking for the bottom line on their organization’s performance.  Dashboards are great for displaying all the relevant information on a single screen, allowing your user to have a one stop shop for when they are at their computer.  With most of an executive’s time comprising of meetings and talking with folks, having mobile access from their handheld device (BlackBerry, iPhone, etc.) will provide them with metrics where and when they need them.  Couple this with real-time alerts via email, and your Executive users will be able to leverage your BI platform according to their needs.

In contrast, front line users (Customer Service Representatives, Sales Representatives, etc.) need intelligence that is bundled in with their business processes.  To drive maximum effectiveness you want to inject Business Intelligence into the operational systems these people use every day.  This can take forms such as surfacing a Customer Lifetime Value score to assist with save strategies when a customer is calling to disconnect, or providing real time alerts into your Customer Relationship Management system when a critical order deadline has passed.  You certainly don’t want these roles that are the face to the customer wasting time flipping between their operational screens and the BI portal!

The way you provide a tailored user experience to your BI clients is critical to drive adoption towards the goal of pervasive BI.  Spend extra time and attention to this aspect when you are planning any major BI deployment and you will reap the benefits.

Related Posts:

• Icing the BI Cake with User Experience
• It Just Has to Work
• Will the BI portal last?

Icing the BI Cake with User Experience

The true power of BI is the ability to present actionable information to decision makers.  All the data quality governance, ETL excellence, and dimensional analysis will mean nothing if the business can't intuitively interact with the information through their BI tools.  You must ice the cake with an exceptional User Experience to drive the most value from your BI investments.

 

A focus on User Experience is key to your BI program's success for the following reasons:

• Enables a faster "mean time to answer".  Reducing the number of clicks and searches makes your applications more satisfying and efficient to use.
• Consistency.  Once a user understands how to read one report or dashboard, they can apply the same methodology across all BI applications.
• Expose correlations.  Simply combining data on a visualization like a chart or a scatter plot can immediately show a correlation between 2 measures that are not as readily apparent as pure numbers.
• Guide your user through tried and tested analysis processes to expose the knowledge of power users to the larger user base.
• Simplify the experience of finding information by improving navigation structures.  Ensure your search capabilities are optimized to return the most relevant results.

In order to drive the most value from your user's experience you need to have a consistent approach across all the user touch points in your environment.  My recommendation is to designate someone to be accountable for User Experience across the board.  Depending on the size of your BI program this may be part of someone's role, a full person, or a small team.  Only with a centre of excellence can you drive consistent standards across all groups creating end user content.  This team would also serve as a consultant to projects to ensure that the reporting and visualizations are designed to be intuitive and useful.

You can have all the wonderful architecture and data quality you want, but without an effective User Experience your clients will not see the value of all your hard work and will be less likely to adopt your BI environment.  First impressions only happen once.

Related Posts:

• It Just Has to Work
• Cut to the Truth: Answer a Business Question

The Tactical Data Mart Advantage

The capabilities of current BI tools allow us to drive down the costs to develop BI applications / data marts.  This means that we can quickly develop tactical applications leveraging standard tools and methodologies to reduce development cycles.  This enables the deployment of applications to support specific decision making problems that would normally not be supported by a quality BI environment due to the short turnaround required.

Some people get upset about building these so called silos that aren't integrated.  I take a different perspective that these tactical data marts expose the power of tactical Business Intelligence, and if you manage it well they are a great complement to your core infrastructure.

"A good solution applied with vigour now is better than a perfect solution applied ten minutes later.” - General George S. Patton

It is critical that you have a checkpoint or a regular environment review that you can rely on to take one of the following actions at the end of the decision making life cycle:

1. Decision has been made and there is no need for continued access to the data mart.  Simple response, retire the data mart and free up system resources for the next initiative.  Celebrate the success.
2. The information within the data mart has continuing value to the organization.  In this case you need to have room in your requirements funnel to incorporate the data mart into your permanent BI infrastructure.

Don't hold your business back by sticking to a traditional release cycle.  Build a team within your BI program that can quickly spin up the appropriate BI solution to support tactical decision making.  You can't win the war if you don't win a battle now and then!

Data Integration and Your Organization

Today we had John Schmidt, VP of Global Integration Services at Informatica come speak to our teams about best practices for setting up an Integration Competency Centre (ICC).  John made an interesting point that your various Integration functions may be included in Competency Centres in different ways depending on the organization.  

If you have researched Business Intelligence Competency Centres (BICC) you will often find a Data Integration function as one of the functions staffed within the BICC.  However, Business Intelligence is just one part of the pie when it comes to integrating information across the enterprise.  You will also need data integration expertise for data quality, Master Data Management, and other non-BI specific projects.  A number of these projects may not impact BI significantly if at all.  If you want a holistic view of data across your organization you must either create a standalone ICC with strong ties to your BICC, or incorporate all areas of integration practice within a higher level Information Management Competency Centre.

Good BI is enabled by quality data.  Since data quality lives along the information management timeline from your source systems to your analytic databases, it is critical you have technology, process and people engaged all the way along to ensure the best data is delivered for BI.

Keeping your BI environment clean

Over the past year or so the "green" revolution has been sweeping IT as a way for organizations to reduce their environmental impact and to cut costs at the same time.  This has driven the need for robust measurement systems to manage the transition to being a green organization.  A number of the big BI vendors have come out with pre-packaged solutions to drive more "green"  for organizations by encapsulating best practices.

It is great for BI teams to provide these systems to the organization, but we would be remiss to not apply these techniques within our BI environments.

What can you do?  One thing I recommend you do is to perform a complete review of your Business Intelligence data stores and reporting at least once a year.  The overall goals of this review are to remove obsolete information and to address any deficiencies.  There are a number of activities you could incorporate into this review:

• Validate with your business stakeholders that the information has continuing value.  Solicit feedback on the information that is valuable, but is somehow incomplete.
• Review usage statistics for data marts and reports to determine what content is accessed regularly and by whom
• Validate that the report or data store reflects current business logic
• Gather metrics on batch load processes, response time, and storage usage to ensure your environment is stable and is scaling as volumes grow.  I've never heard of a Data Warehouse that shrinks in size over time.

The majority of these activities do not cost you a dime other than people's time.  This is especially attractive in a financial environment like today, when capital spending is being closely scrutinized.  Putting in place a regular program such as this will enable your organization to realize the following benefits:

• Reduce maintenance costs for your BI environment.  Less data marts and reports to support as your environment grows and changes.  You can quickly find yourself drowning in too many items to support.
• Lower the related data centre costs to support your infrastructure.  Believe it or not, all your ETL processes, database servers and storage solutions consume significant amounts of power and cooling.
• Identify duplicated information that is redundant.  Root out conflicting business definitions.
• Reduce the opportunity for information overload.  Having focused information will allow your users to find what they are looking for quicker.

Spare the environment and lower your costs by supporting less while extracting maximum value from your existing investments.  You can't beat this type of ROI!

BI for BI

One of the core disciplines within a Business Intelligence team or program is to develop internal metrics and performance management practices. The idea of "BI for BI" is to exploit your internal strengths and live the fact-based decision mantra.

Business Intelligence is a core component of corporate performance management systems. Conversely, Business Intelligence can only deliver maximum value when it is coupled by a strong performance management culture. In order to communicate the value of spend on BI / performance management initiatives, we need to talk about both the BI technology and the culture required to support it.

If we have the management practices in place to constantly act on and refine our internal program / departmental metrics, we can then use this as a working example to the rest of the organization. Not only will you be able to demo the metrics produced for your team, but you will be able to relate success stories that you had acting on this information. Not only show a dashboard trending average bugs per release, but also tell the story of how your team saw a negative trend 6 months ago, implemented an improvement to the code review process, and reduced overall bugs by 50%. How powerful is that!

Here are just a couple examples of metrics you could put in place to measure your BI program:

• ETL Job Completion Times
• Uptime by Infrastructure Component (Presentation, ETL, Database Layers)
• Volume of Requests for Information
• Average Frequency of Visits to BI Portal (i.e. 3.5 times per week)
• Volume of Data Quality Issues
• Resolution Time on Service Tickets
• Production Incidents by Severity
• Changes to Report / Asset Catalogue

A couple hints on how to put this in place:

• Make people accountable for their metrics. They own them and report on them.
• Start small and grow. Start with measuring only a handful of metrics, and add additional metrics over time. Look for patterns of key performance indicators, and their relation to other metrics.
• Review and publish your metrics. Have a regular time set for metric review. Make metrics available to anyone that interacts with you, including your customers. Open the Kimono!

It is a basic professional requirement for an organization to produce regular metrics on their performance. Leverage your team's expertise in the discipline of Business Intelligence to improve your team's performance, and show the rest of the organization how it is done!

Cut to the Truth: Answer a Business Question

As the wealth of data available to consumers of Business Intelligence in your organization exponentially increases, so does the opportunity to expose all this data... Even when we shouldn't...

What?? Don't leverage any and all information to analyze every aspect of the situation? Bite your tongue!

When we have decision makers that need to make a call and move on to the next problem, they don't have the time (or the patience) to sort through endless reams of data to find the nugget or two that will help them make the decision. It is critical that we always ask the golden question, "What business question are we trying to answer?"

Once we have identified the end purpose of our analysis, we can then focus our energies on leveraging all this data to show the proper trend, scoring or visualization as our end result. Don't try to overcomplicate your analysis, just focus on the 80% value you can deliver with 20% of the effort. Your customers will love you for making their jobs easier, and your solutions will be less costly and easier to maintain.

Don't you love the 80/20 rule? Simplicity at it's best.

Keep Your KPIs Fresh

One thing that has caught my attention over the past couple of months is the emphasis on new financial metrics in this depressed market. Employment statistics used to be mentioned in passing, and now we breathlessly await the monthly numbers. Clearly the global recession is an entirely new environment, and investors would be wise to not rely on old metrics and approaches to guide them.

This made me think of how a core component of a good performance management system is keeping your metrics relevant to the current business environment they are attempting to measure. There is no such thing as permanent performance indicators. If your metrics don't change, then there is something wrong with your process. It is critical to keep your performance metrics focused on the current objectives and strategies of your business.

If your company is spending money differently, as all companies are doing in some way, then your performance indicators must change to reflect this. In these types of rapid change it is entirely possible that your core business strategy can shift significantly. It is essential that the keepers of corporate performance make it a continuous cycle to review and evaluate the suite of KPIs to ensure they are aligned to today's business reality.

This also speaks to the need for the IT organization to be able to rapidly respond to these changes in direction to provide the necessary data for the new KPIs. The best BI infrastructures are highly adaptable and responsive to change.

Step back and take a look at your metrics. Are they still relevant?