Lessons From The Frontline

Steve Bennett · 13th May 2009, 10:48 PM

Recently I have just finished a long stint building an analytics capability in a large organisation. This has given me time to reflect on the experience and I'd like to share some of my conclusions.

First off, the analytic capability is very successful and is driving better decisions in many areas of the organisation. That's the easy bit to say - and I would say that anyway :).

What's more interesting are the lessons that I have learned - because they will help me do better next time. What are the lessons and how do I intend putting them into practice?

Lesson #1 - Get the Data Right

You can't achieve advanced analytics if the data isn't right.

Investing in high performing Business Analysis and Data Integration teams is the first step towards delivering a high performing organisation.

I say teams but even one of each is enough to kick-start things. I also mention two teams because together they give you an unbeatable combination of analytic power within your team.

Lesson #2 - Deliver Fast and Avoid 'Big Bangs'

If you are delivering a complex thing like an enterprise self-service analytic capability then you have to deliver value much faster than is possible.

How do you overcome this?

The method I use is to identify existing information problems that are long standing issues. They are never hard to find and the application of a little logic and creativity invariably delivers a tactical solution that you can implement in a matter of months.

We found a number of these problems (over 40 if I remember) and we delivered solutions to 20 or more in the months before we were able to deliver the enterprise data warehouse and self-service elements.

Just as importantly, the deliveries were spread consistently over the 2.5 years so that each quarter we could point to concrete benefits.

Lesson #3 - Deliver to the Direct Beneficiaries of Quality Data

There will be one or two groups within the organisation that can benefit directly from the high quality data that the Data Integration team make possible. The key point here is that these groups do not need the presentation layer (reports, dashboards, etc.). This makes it much quicker to deliver. Find out who they are and make them the second customers you deliver to.

In this case, Risk Management and Securitisation were departments that were identified and fit this bill. Delivering to these areas also gave us time to build the self-service analytic tools to other areas.

Lesson #4 - You've Got To Build An Exclusive Club

In the corporate world (the 'big end of town' we say in Australia) you are faced with people with a range of skills and talents. Some very, very good people with great skills and attitudes that I admire. The majority quite capable and with leadership can also deliver exceptional performance. The remaining 10% however are just not good enough. Nothing surprising in that I hear you say?

Well I disagree. I spent 10 years starting, growing and selling start-up companies and the 10% 'not good enough' don't last long in that environment. This makes a fundamental difference in the speed at which a team can deliver. Just as importantly - it's less fun for everybody. I think that you should have fun, be proud of the work and learn from your team. All of these things are diminished when you don't have a high performing team.

The organisation I worked for had an admirably cooperative and positive culture. To illustrate: you could approach just about anyone in the company and ask them a question and their first thought would be "How can I help this person?". Fantastic!

So what is the lesson?

I let the 'cooperative and positive culture' influence me to keep the 10% on in their jobs. On reflection, it would have been much better for all involved if I had done what I usually do: quickly and fairly get rid of them. Delaying or avoiding doing this hurt the performance of the analytic team I was building. Ultimately this also hurt our ability to get analytics adopted by the wider business.

Lesson #5 - New Tools Are Required

My final lesson has to do with the underlying technologies employed. I inherited a set of 'big tool' platforms (you know - the ones like IBM, Oracle and SAP). Early on I decided to keep them and go the upgrade path. I think this was wrong in a couple of areas - mainly around reporting and end user analytics.

I should have looked at more agile alternatives. Ultimately I believe they would have been easier and cheaper to deploy.

What are these alternatives? Here's is a list of a few that have caught my attention:

Open Source - tools are ready for the big time. Don't hesitate, start experimenting with them today.

QlikView, Zap and others - support more rapid deployment of analytics than the mega-platforms like IBM, Oracle and SAP.

Cloud9 Analytics - a role-based, on-demand business intelligence applications using a software-as-a-service (SaaS) model.

eThority - a web and desktop based, user-focused, accessible and scalable approach to business-driven analysis of enterprise data.

illuminate - a correlation database that brings agility to the back-end data integration tasks that are barriers to agile analytics.

Lyzasoft - depolarizing and “filling the middle” with desktop-based data gathering, data analysis, reporting and analytic publishing.

Netmap - a visual approach to discovery-based analysis.

PolyVista - extending OLAP with prepackaged, easy-to-use data mining and discovery automation capabilities.

Lesson #6 - Ignore Service Providers and Recruitment Consultants

There are many good BI specialist Service Providers and Recruitment Consultants out there and they all want to develop a relationship with you. This is great and they can add value but when you are doing what we were doing there is not enough money or time to allow relationships to develop.

Our (my) strategy was to use a team of existing staff, a number of new specialist hires and specialist contractors. If I couldn't find the new hires and specialist contractors in my own network, then Service Providers and Recruitment Consultants were a vital component of the strategy. My existing network ensured that key appointment could be made with confidence. It certainly wasn't all 'plain sailing' but it worked well enough. We did go to market via recruitment agencies for a number of roles and the results were good. The success rate through agencies over the 2.5 years was about 50% (1 out of 2 hires were successful).

In hindsight, a lot of time was spent with lots of Service Providers and Recruitment Consultants. It would have been very useful to hear from my own network which ones they already knew were good. I still wonder why I didn't do more reference checking. The lesson is to be more disciplined in what you spend time on. Building an analytic capability is a full time job. So be firm and resist the chance to have 10 coffees a day!

More...

Thread Tools	Search this Thread
Show Printable Version Email this Page	Search this Thread: Advanced Search
Display Modes
Linear Mode Switch to Hybrid Mode Switch to Threaded Mode

13th May 2009, 10:48 PM	#1
Steve Bennett Member Join Date: Oct 2007 Posts: 386 Blog Entries: 26	Lessons From The Frontline Recently I have just finished a long stint building an analytics capability in a large organisation. This has given me time to reflect on the experience and I'd like to share some of my conclusions. First off, the analytic capability is very successful and is driving better decisions in many areas of the organisation. That's the easy bit to say - and I would say that anyway :). What's more interesting are the lessons that I have learned - because they will help me do better next time. What are the lessons and how do I intend putting them into practice? Lesson #1 - Get the Data Right You can't achieve advanced analytics if the data isn't right. Investing in high performing Business Analysis and Data Integration teams is the first step towards delivering a high performing organisation. I say teams but even one of each is enough to kick-start things. I also mention two teams because together they give you an unbeatable combination of analytic power within your team. Lesson #2 - Deliver Fast and Avoid 'Big Bangs' If you are delivering a complex thing like an enterprise self-service analytic capability then you have to deliver value much faster than is possible. How do you overcome this? The method I use is to identify existing information problems that are long standing issues. They are never hard to find and the application of a little logic and creativity invariably delivers a tactical solution that you can implement in a matter of months. We found a number of these problems (over 40 if I remember) and we delivered solutions to 20 or more in the months before we were able to deliver the enterprise data warehouse and self-service elements. Just as importantly, the deliveries were spread consistently over the 2.5 years so that each quarter we could point to concrete benefits. Lesson #3 - Deliver to the Direct Beneficiaries of Quality Data There will be one or two groups within the organisation that can benefit directly from the high quality data that the Data Integration team make possible. The key point here is that these groups do not need the presentation layer (reports, dashboards, etc.). This makes it much quicker to deliver. Find out who they are and make them the second customers you deliver to. In this case, Risk Management and Securitisation were departments that were identified and fit this bill. Delivering to these areas also gave us time to build the self-service analytic tools to other areas. Lesson #4 - You've Got To Build An Exclusive Club In the corporate world (the 'big end of town' we say in Australia) you are faced with people with a range of skills and talents. Some very, very good people with great skills and attitudes that I admire. The majority quite capable and with leadership can also deliver exceptional performance. The remaining 10% however are just not good enough. Nothing surprising in that I hear you say? Well I disagree. I spent 10 years starting, growing and selling start-up companies and the 10% 'not good enough' don't last long in that environment. This makes a fundamental difference in the speed at which a team can deliver. Just as importantly - it's less fun for everybody. I think that you should have fun, be proud of the work and learn from your team. All of these things are diminished when you don't have a high performing team. The organisation I worked for had an admirably cooperative and positive culture. To illustrate: you could approach just about anyone in the company and ask them a question and their first thought would be "How can I help this person?". Fantastic! So what is the lesson? I let the 'cooperative and positive culture' influence me to keep the 10% on in their jobs. On reflection, it would have been much better for all involved if I had done what I usually do: quickly and fairly get rid of them. Delaying or avoiding doing this hurt the performance of the analytic team I was building. Ultimately this also hurt our ability to get analytics adopted by the wider business. Lesson #5 - New Tools Are Required My final lesson has to do with the underlying technologies employed. I inherited a set of 'big tool' platforms (you know - the ones like IBM, Oracle and SAP). Early on I decided to keep them and go the upgrade path. I think this was wrong in a couple of areas - mainly around reporting and end user analytics. I should have looked at more agile alternatives. Ultimately I believe they would have been easier and cheaper to deploy. What are these alternatives? Here's is a list of a few that have caught my attention: Open Source - tools are ready for the big time. Don't hesitate, start experimenting with them today. QlikView, Zap and others - support more rapid deployment of analytics than the mega-platforms like IBM, Oracle and SAP. Cloud9 Analytics - a role-based, on-demand business intelligence applications using a software-as-a-service (SaaS) model. eThority - a web and desktop based, user-focused, accessible and scalable approach to business-driven analysis of enterprise data. illuminate - a correlation database that brings agility to the back-end data integration tasks that are barriers to agile analytics. Lyzasoft - depolarizing and “filling the middle” with desktop-based data gathering, data analysis, reporting and analytic publishing. Netmap - a visual approach to discovery-based analysis. PolyVista - extending OLAP with prepackaged, easy-to-use data mining and discovery automation capabilities. Lesson #6 - Ignore Service Providers and Recruitment Consultants There are many good BI specialist Service Providers and Recruitment Consultants out there and they all want to develop a relationship with you. This is great and they can add value but when you are doing what we were doing there is not enough money or time to allow relationships to develop. Our (my) strategy was to use a team of existing staff, a number of new specialist hires and specialist contractors. If I couldn't find the new hires and specialist contractors in my own network, then Service Providers and Recruitment Consultants were a vital component of the strategy. My existing network ensured that key appointment could be made with confidence. It certainly wasn't all 'plain sailing' but it worked well enough. We did go to market via recruitment agencies for a number of roles and the results were good. The success rate through agencies over the 2.5 years was about 50% (1 out of 2 hires were successful). In hindsight, a lot of time was spent with lots of Service Providers and Recruitment Consultants. It would have been very useful to hear from my own network which ones they already knew were good. I still wonder why I didn't do more reference checking. The lesson is to be more disciplined in what you spend time on. Building an analytic capability is a full time job. So be firm and resist the chance to have 10 coffees a day! More...